CT based Quantification of Intratumoral and Peritumoral Heterogeneity for diagnosing Lymphovascular Invasion for Early Stage Non-small cell lung cancer

CT based Quantification of Intratumoral and Peritumoral Heterogeneity for diagnosing Lymphovascular Invasion for Early Stage Non-small cell lung cancer | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article CT based Quantification of Intratumoral and Peritumoral Heterogeneity for diagnosing Lymphovascular Invasion for Early Stage Non-small cell lung cancer Yun Long, Hanfei Zhang, Yang Guo, Tian Gan, Jingting Wang, Ting Li, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6764701/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 19 Dec, 2025 Read the published version in BMC Medical Imaging → Version 1 posted 9 You are reading this latest preprint version Abstract Objective: To establish a model integrates clinical, traditional radiologic, intratumoral and peritumoral radiomics (ITR and PTR), and intratumoral and peritumoral heterogeneity (ITH and PTH) features to diagnose lymphovascular invasion (LVI) status for early stage non small cell lung cancer (NSCLC). Materials and Methods: Clinical data and chest CT imaging data of NSCLC patients who underwent surgical resection of the lungs from January 2019 to May 2021 were collected. Surgical pathology were the diagnostic gold standard to clarify the LVI status. ITR and PTR features and ITH and PTH features from the total tumor volume and peritumoral tumor volume were extracted. Then clinical, traditional radiologic, ITR and PTR, ITH and PTH models were established to diagnose LVI status. Finally, a column chart diagnostic model was constructed and the diagnostic efficacy was evaluated. Results: 308 NSCLC cases met the inclusion criteria, including 117 cases in the LVI positive group and 191 cases in the LVI negative group. They were randomly divided into training group and validation group in a 1:1 ratio. In the three cohorts of PTR_(0 - 3, -3 - 3 and 0 - 6), the PTR_0 - 6 model has better predictive performance, with area under the curve (AUC) of 0.882 and 0.824 for the training and validation groups, respectively. Gender, Vascular Convergence Sign, and N stagewere significantly related to LVI status, Finally, the combined model integrated ITH, PTR_0-6, and PTH_0-6 models, N stage and Vascular Convergence Sign has the highest diagnostic accuracy. The AUC in the training group is 0.962 and in the validation group is 0.882. Conclusions: A comprehensive diagnostic model based on clinical features, traditional radiological features, radiomic features, and heterogeneity features of NSCLC were established to diagnose LVI for early stage NSCLC, which has the highest diagnostic efficiency and can help to guide treatment decisions. Non small cell lung cancer CT imaging Lymphovascular invasion radiomics Tumor heterogeneity Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Cancer is a major societal, public health, and economic problem in the 21st century; and lung cancer has the highest morbidity and mortality among all cancer types globally [ 1 ] . Regrettably, about half of all new lung cancer cases were in Asia [ 2 ] . There are 2 main forms of lung cancer: non small cell lung cancer (NSCLC) and small cell lung cancer, and NSCLC account for 85% [ 3 ] . Treatment of NSCLC is stage specific. Patients with stage I or II should be treated with complete surgical resection when not contraindicated. Nonsurgical patients should be considered for radiotherapy or chemotherapy [ 3 ] . Whereas 25% -70% of surgical patients eventually relapse despite complete resection, with 5-year survival of only 35% -65% [4] . Distant metastasis is the most common form of postoperative recurrence [5] . Studies have shown that cancer must first enter and spread to the entire vasculature before it can spread and metastasize through blood vessels or lymphatic vessels [6,7] . The detection of neoplastic cells in arterial, venous, or lymphatic lumen with H&E stain is called lymphovascular invasion (LVI) [8-10] . The presence of LVI in patients with lung cancer is associated with poor prognosis [11] , and LVI is an independent prognostic factor for recurrence-free survival in patients with stage I lung adenocarcinoma undergoing lobectomy [12] . Studies have shown that LVI can be used as an indication for preoperative neoadjuvant chemotherapy in NSCLC patients. Preoperative neoadjuvant chemotherapy can reduce tumor volume, shorten tumor stage, and provide long-term survival benefit [13,14] . Lobectomy plus lymphadenectomy is more effective in patients with LVI-positive NSCLC compared with subresection alone [15] .Therefore, it is very important to identify LVI invasion before surgery or other therapy. At present, the diagnosis of LVI of lung cancer is mainly by HE staining, immunohistochemistry and special staining after operation. LVI is a histological condition that can only be recognized postoperatively by surgical specimens [16,17] . NSCLC exhibits strong intratumor heterogeneity (ITH), and more heterogeneous tumors have higher invasive and metastatic potential [18] . Studies have shown that invasive tumors with LVI have significantly higher ITH scores than non-invasive tumors [19] . Because of different ITH, patients with other similar clinical features, such as clinical stage, molecular subtype, have different LVI status. Therefore, ITH could help to predict LVI status. Previous studies have focused on the relationship between clinical outcomes and radiological features within the primary tumor volume, yet the lung parenchyma surrounding the primary tumor may also be involved in tumor invasion and metastasis [20,21] . Pathological studies have shown that lung tumors can cause worse clinical manifestations through lymphatic metastasis, hematogenous metastasis, or direct invasion of surrounding lung tissues [22-29] . Furthermore, the association of tumor peripheral cancer cell infiltration with local recurrence or distant metastasis was significantly stronger compared with intratumoral cancer tissure [22,25,27] Among preoperative non-invasive diagnostic methods, conventional computer tomography (CT) is the most commonly used one. But studies have shown that CT-based diagnosis of vascular and mediastinal tumor invasion has limited utility [30-32] , subtle differences in some of the underlying features can not be identified with the naked eye. Radiomics, as an image analysis technique, can extract complex information that is difficult for the human eye to recognize or quantify from a diagnostic image and convert it into quantitative data [33,34] . Thus, based on the importance of diagnosing LVI before surgery and the lack of corresponding method, we establish a model integrated clinical, traditional radiologic, intratumoral and peritumoral radiomics (ITR and PTR), and intratumoral and peritumoral heterogeneity (ITH and PTH) features to diagnose LVI status for early stage NSCLC. Methods Patients This retrospective study was approved by the Ethics Committee of Zhongnan Hospital of Wuhan University, with the ethical approval number 2021048 (IRB number). According to the CIMOS guidelines, informed consent was waived for this study. It involved collecting pathological data and chest CT images from patients (Figure S1) who underwent surgical resection for non-small cell lung cancer (NSCLC) between January 2019 and May 2021. All patients had their surgeries within two weeks of their CT scans, which included routine blood tests measuring tumor markers such as Carbohydrate Antigen 125 (CA125, reference value ≤35 U/mL), Carcinoembryonic Antigen (CEA, reference value 0.0-5.0 ng/mL), and Neuron Specific Enolase (NSE, reference value ≤16.0 ng/mL). The inclusion criteria were: 1) pathologically confirmed NSCLC; 2) availability of a pathological report with clear LVI status; 3) a diagnostic-compliant chest CT performed within two weeks prior to surgery; and 4) complete baseline clinical data. The exclusion criteria included: 1) lack of complete clinical-pathological data or preoperative CT images; 2) presence of other malignant tumors; and 3) prior anti-tumor treatment before surgery. Data Flowchart As shown in Figure 1, the data analysis process in this study consists of two main parts: clinical risk factors analysis/modeling, and image data analysis/modeling. The analysis of clinical risk factors is conducted through univariate and multivariate analyses step by step, while image data processing includes three steps: preprocessing, segmentation, and model construction. Notably, regions of interest (ROI) within the tumor (intratumoral) and around the tumor (peritumoral) were analyzed respectively, and both the traditional radiomics models and the heterogeneity models that well reflecting tumor heterogeneity were developed separately. Ultimately, integrating selected clinical risk factors with the intratumoral and peritumoral models, we established a combined model and draw the corresponding nomogram to enhance its clinical utility. During data processing, all patients were randomly divided into training and testing sets in a 1:1 ratio using the random stratified sampling method. The training set was used for model development and construction, while the testing set served as an independent sample for model validation. All model performance metrics were calculated and compared across both the training and testing sets. Image Processing Before the formal analysis of the images, all patients' CT images underwent a two-step preprocessing: resampling and normalization of CT values. This ensures consistency and comparability of images acquired from different devices and batches, thereby enhancing the accuracy and generalizability of the diagnostic models. Specifically, we resampled all patients' chest CT images to a resolution of 1mm×1mm×1mm using linear interpolation, and normalized the CT values to a standard normal distribution N(0,1) using the Z-scoring method. Subsequently, we manually segmented the tumor regions using ITK-SNAP (version 3.8.0; http://www.itksnap.org) to obtain the tumor regions of interest (ROIs). Based on this, the SimpleITK package was used to automatically contract the tumor ROI (gross tumor volume, GTV) by 3mm or expand it by 3mm and 6mm, defining three different peritumoral tumor volume (PTV) areas (Figure 2): from 3mm inside to 3mm outside the tumor boundary (PTV_-3-3), from the tumor boundary to 3mm outside (PTV_0-3), and from the tumor boundary to 6mm outside (PTV_0-6). We manually corrected the automatic expansion results to exclude adjacent pleura and bones included during the process. The delineation of the tumor ROI and the adjustment of the peritumoral areas were completed jointly by two radiologists under CT lung window settings (width 1500 HU; level -500 HU), and the ICC method was used to ensure stability and repeatability of the results. This process ultimately generated one intratumoral ROI and three peritumoral ROIs for subsequent modeling and analysis. Model Construction Firstly, univariate analysis was used to evaluate clinical variables and radiological semantic features, selecting variables with p<0.05 as clinical risk factors. Subsequently, a multivariate linear regression method was employed to construct the Clinical model. Next, we conducted traditional radiomics modeling. For the four tumor ROIs (GTV, PTV_-3-3, PTV_0-3, and PTV_0-6), 105 radiomics features were extracted (details in the supplementary document). Feature reduction was performed as follows step by step: (1) features with variances below a threshold of 1 were removed using the variance threshold method; (2) highly correlated features (R > 0.9) were eliminated using the Pearson correlation analysis method; (3) features showing significant differences between classification groups were identified using the t-test (p < 0.05); (4) the least absolute shrinkage and selection operator (LASSO) method was used to select the optimal feature subset and construct the corresponding traditional radiomics models(ITR, PTR_-3-3, PTR_0-3, and PTR_0-6). Following this, we constructed heterogeneity models. The predictive performance of traditional radiomics models based on three peritumoral ROIs was compared to determine the best peritumoral ROI. Then, intratumoral heterogeneity models (ITH) and peritumoral heterogeneity models (PTH) were developed separately for the intratumoral ROI and the optimal peritumoral ROI. The construction of heterogeneity models involved the following steps: (1) subregion segmentation of the tumor, objectively identifying and segmenting subregions inside the ROI to better understand the internal structure and heterogeneity of the tumor; (2) unsupervised clustering analysis, classifying tumor subregions with similar features to reveal different ecological "populations" within the tumor ROI; (3) calculation of ecological diversity indices for different ecological "populations" within the tumor ROI; (4) constructing the "Tumor Ecological Diversity Feature Vector" (TED) based on the ecological diversity indices, i.e., the heterogeneity model. For specific details on each step's design and implementation, refer to the supplementary materials (Figure S1). Finally, the selected clinical risk factors, intratumoral and peritumoral radiomics models (ITR and PTR), as well as intratumoral and peritumoral heterogeneity models (ITH and PTH) were used as independent risk factors for LVI. A multivariate linear regression method was utilized to construct a combined model, and a nomogram was created to visualize the final model, enhancing its clinical interpretability and usability. Model Validation In this study, eight models were established: four traditional radiomics models (ITR, PTR_-3-3, PTR_0-3, and PTR_0-6) for the tumor ROI and three peritumoral ROIs, two heterogeneity models (ITH and PTH_0-6) for the intratumoral and the optimal peritumoral ROIs, a clinical model, and the final combined model. The predictive accuracy of these models was evaluated using ROC curves and their derived metrics, Area under the curve (AUC), accuracy (Acc), sensitivity (Sens), and specificity (Spec). The Delong test was used to determine whether differences in ROC curves between the models were significant. Additionally, the consistency between the model predictions and actual observations was assessed using calibration curves, and the models' goodness of fit was examined with the Hosmer-Lemeshow (HL) test. Finally, decision curves were used to evaluate the clinical benefits of the models at different risk thresholds. Statistics Statistical analyses were conducted using R software (version 4.1.2, http://www.R-project.org). For quantitative data following a normal distribution, the mean ± standard deviation (mean±SD) was used for presentation, and comparisons between groups were performed using independent t-tests. For data not following a normal distribution, the median and interquartile range (IQR) were reported, and comparisons between groups were made using the Mann-Whitney U test. Categorical data were analyzed using the Chi-square test or Fisher’s exact test. A p-value of less than 0.05 was considered statistically significant. Results Patients This study included 308 patients with non-small cell lung cancer (NSCLC), of whom 117 were LVI-positive and 191 were LVI-negative. Clinical imaging characteristics of the patients are detailed in Table 1 . After a 1:1 random assignment, there were 154 patients in the training group and 154 in the validation group. The training group consisted of 59 LVI-positive patients and 95 LVI-negative patients, while the validation group included 58 LVI-positive patients and 96 LVI-negative patients. The patients enrollment details were shown in Figure S2. Table 1 Patients demographics for training and test set. Characteristic Training set (154) p -value Test set (154) p -value LVI(+) (n = 59) LVI(-). (n = 95) LVI(+) (n = 58) LVI(-) (n = 96) Age 63.00 (58.20, 67.00) 63.00 (56.20, 67.00) 0.470 64.09 ± 7.72 62.80 ± 7.95 0.328 Gender Female 13 (22.03%) 36 (37.89%) 0.04* 14(24.14%) 33(34.38%) 0.181 Male 46 (77.97%) 59 (62.11%) 44(75.86%) 63(65.62%) Pathology adenocarcinoma 42(71.19%) 77 (81.05%) 0.156 35(60.34%) 78(81.25%) 0.004 Squamous Cell Carcinoma 17(28.81%) 18 (18.95%) 23(39.66%) 18(18.75%) TNM Stages I 26(44.07%) 68(71.58%) 0.003* 20(34.48%) 72(75.00%) < 0.001 II 25(42.37%) 22(23.16%) 27(46.55%) 16(16.67%) III 8(13.56%) 5(5.26%) 10(17.24%) 8(8.33%) N Stage 0 33(55.93%) 86(91.49%) < 0.001* 35(60.34%) 84(91.30%) 0.001 1 25(42.37%) 8(8.51%) 22(37.93%) 8(8.70%) 2 1(1.69%) 0(0.00%) 1(1.72%) 0(0.00%) MaxSize (mm) 3.24 (2.20, 4.40) 2.74 (2.15, 4.01) 0.456 3.05(2.50, 4.78) 2.60(1.77, 3.50) 0.011 Location centra 26(44.07%) 31(32.63%) 0.153 22(37.93%) 31(32.29%) 0.475 periphral 33(55.93%) 64(67.37%) 36(62.07%) 65(67.71%) Density Solid 51(86.44%) 75(78.95%) 0.241 52(89.66%) 70(72.92%) 0.013 Subsolid 8(13.56%) 20(21.05%) 6(10.34%) 26(27.08%) Spiculation Sign Yes 44(74.58%) 68(71.58%) 0.685 43(74.14%) 66(68.75%) 0.476 No 15(25.42%) 27(28.42%) 15(25.86%) 30(31.25%) Lobulation Sign. Yes 54(91.53%) 88(92.63%) 1.000 56(96.55%) 90(93.75%) 0.701 No 5(8.47%) 7(7.37%) 2(3.45%) 6(6.25%) Pleural Indentation Sign. Yes 28 (47.46%) 54(56.84%) 0.256 33(56.90%) 58(60.42%) 0.667 No 31(52.54%) 41(43.16%) 125(43.10%) 38(39.58%) Bronchus Sign Yes 21(35.59%) 47(49.47%) 0.092 33(56.90%) 41(42.71%) 0.962 No 38(64.41%) 48(50.53%) 41(70.69%) 55(57.29%) Vascular Convergence Sign Yes 37(62.71%) 84(88.42%) < 0.001* 41(70.69%) 73(76.04%) 0.463 No 22(37.29%) 11(11.58%) 17(29.31%) 23(23.96%) Vacuole Sign Yes 15(25.42%) 25(26.32%) 0.022 14(24.14%) 21(21.88%) 0.745 No 44(74.58%) 70(73.68%) 44(75.86%) 75(78.12%) CEA 3.45 (2.52, 7.21) 2.96 (2.02, 4.69) 0.035* 3.75 (2.25, 5.77) 2.95 (1.77, 5.15) 0.127 CA-125 13.59 (9.78, 29.03) 13.79 (9.86, 20.38) 0.539 16.72(10.27, 25.96) 13.44(8.79, 19.71) 0.079 NSE 13.21 (10.65, 15.56) 13.04 (10.81, 14.37) 0.513 12.92(11.16, 14.85) 12.10(10.71, 14.34) 0.201 Clinical Risk Factors and Modeling In the univariate analysis of the training group, significant statistical differences were observed between the LVI-positive and LVI-negative groups in terms of gender and CEA levels (p < 0.05). In semantic features, significant differences were noted between the two groups in TNM stage, N stage, Vascular Convergence Sign (VCS), and Vacuole Sign (p < 0.05). All clinical and semantical significant variables were included into a Generalized Linear Model (GLM) for analysis to construct the Clinical model. According to the results of the GLM analysis (Table S1 ), gender, VCS, and N stage were selected for the Clinical model construction, in which N stage showing the most significant correlation with LVI status (Coef 0.480, 95% CI 0.266–0.693, p < 0.001). Radiomics and Heterogeneity Modeling This study extracted 105 radiomic features from four regional ROIs (GTV, PTV_0–3, PTV_-3-3, and PTV_0–6). Initially, features with ICC values below 0.75 were excluded. Subsequently, a four-step feature selection process reduced the number of features for GTV (Table S2), PTV_0–3, PTV_-3-3, and PTV_0–6 (Table S3) to 3, 9, 11, and 10, respectively. Among the three PTV models, the PTV_0–6 model exhibited AUC values of 0.882 in the training group and 0.824 in the validation group, significantly outperforming (DeLong test: p < 0.05) the PTV_-3-3 model (AUCs of 0.771 and 0.696) and the PTV_0–3 model (AUCs of 0.805 and 0.798). Additionally, the accuracy, sensitivity, and specificity of the three models were comprehensively evaluated. Table 2 (Figure S3) indicates that the performance of the PTV_0–6 model was superior to the other two models. Table 2 Performance of the constructed models evaluated by the ROC-related metrics. Models AUC (95%CI) Accuracy(95%CI) Sensitivity(95%CI) Specificity(95%CI) PTR_-3-3 train 0.771 (0.695–0.847) 0.701 (0.695–0.707) 0.762 (0.749–0.777) 0.663 (0.653–0.673) test 0.696 (0.610–0.783) 0.682 (0.676–0.688) 0.621 (0.604–0.637) 0.719 (0.71–0.728) PTR_0–3 train 0.805 (0.735–0.874) 0.727 (0.622–0.733) 0.831 (0.818–0.843) 0.663 (0.653–0.673) test 0.798 (0.730–0.866) 0.721 (0.715–0.726) 0.879 (0.868–0.890) 0.625 (0.615–0.635) PTR_0–6 train 0.882 (0.830–0.934) 0.812 (0.807–0.817) 0.881 (0.871–0.892) 0.768 (0.76–0.777) test 0.824 (0.760–0.888) 0.701 (0.695–0.707) 0.845 (0.833–0.857) 0.615 (0.605–0.625) Consequently, intratumoral ITH models and peritumoral PTH models were constructed from the GTV and PTV_0–6 ROIs, respectively. Initially, tumor subregions were identified within the GTV and PTV_0–6 ROIs using a simple linear interactive clustering method. Radiomic features were then extracted, and unsupervised clustering was performed on similar feature-bearing subregions using a Gaussian mixture model. Dimensionality reduction was subsequently carried out using the minimum redundancy and maximum relevance method, and ultimately, five imaging features were selected to represent the spatial heterogeneity of the GTV (Table S4) and PTV_0–6 (Table S5), and finally constructed the ITH and PTH_0–6 models, respectively. Finally, the traditional radiomic models of PTR_0–6, along with the ITH and PTH_0–6 models, were combined as independent risk factors with selected clinical semantic imaging risk factors (gender, vascular convergence sign, N staging) to construct a comprehensive nomogram diagnostic model (Table S6, Fig. 3 ). In this model, each "independent risk factor" in a patient’s lesion is converted into a score based on the model's "Points" column. The sum of the scores from the five risk factors for the lesion is in the "Total Points” row, and the corresponding probability in the "Risk of metastasis" column represents the probability of LVI for that lesion. Models Validation and Comparison Figure 4 displays the ROC curves for the constructed intratumoral and peritumoral traditional radiomics models, heterogeneity models, clinical models, and combined models, while Table 3 (Figure S4) lists more comprehensive ROC-related metrics (AUC, accuracy, sensitivity, and specificity) for these models. Figure S5-S10 showed the calibration curves with H-L test results for all six constructed models. The analysis of these figures and tables reveals: Table 3 Performance of the constructed models evaluated by the ROC-related metrics. Models AUC (95%CI) Accuracy(95%CI) Sensitivity(95%CI) Specificity(95%CI) PTR_0–6 train 0.882 (0.830–0.934) 0.812 (0.807–0.817) 0.881 (0.871–0.892) 0.768 (0.76–0.777) test 0.824 (0.760–0.888) 0.701 (0.695–0.707) 0.845 (0.833–0.857) 0.615 (0.605–0.625) PTH_0–6 train 0.725 (0.643–0.805) 0.675 (0.669–0.681) 0.610 (0.594–0.626) 0.716 (0.706–0.725) test 0.692 (0.606–0.778) 0.656 (0.65–0.662) 0.534 (0.518–0.551) 0.729 (0.72–0.738) ITR train 0.673 (0.589–0.757) 0.604 (0.598–0.61) 0.593 (0.577–0.61) 0.611 (0.6-0.621) test 0.665 (0.577–0.753) 0.604 (0.598–0.61) 0.5 (0.483–0.517) 0.667 (0.657–0.676) ITH train 0.722 (0.642–0.802) 0.662 (0.656–0.668) 0.695 (0.68–0.71) 0.642 (0.632–0.652) test 0.607 (0.517–0.697) 0.578 (0.572–0.584) 0.448 (0.431–0.465) 0.656 (0.647–0.666) Clinical train 0.767 (0.688–0.846) 0.766 (0.761–0.772) 0.610 (0.594–0.626) 0.863 (0.856–0.87) test 0.680 (0.595–0.765) 0.669 (0.663–0.675) 0.483 (0.466-0.5) 0.781 (0.773–0.79) Combined train 0.962 (0.937–0.988) 0.903 (0.899–0.906) 0.898 (0.888–0.908) 0.905 (0.899–0.911) test 0.882 (0.831–0.933) 0.779 (0.774–0.784) 0.862 (0.85–0.874) 0.729 (0.72–0.738) (1) Across both the training and testing sets, the combined model exhibits the highest predictive performance among all models, significantly outperforming the other models (p < 0.05). (2) Whether traditional radiomics or heterogeneity models, the models based on peritumoral constructions consistently outperform those based on intratumoral constructions. (3) For both intratumoral and peritumoral ROIs, the predictive performance of the heterogeneity models alone is inferior to that of the traditional radiomics models. Clinical use, Benefit and Explanation To illustrate the clinical use of the constructed nomogram, we listed two examples in Fig. 5 . As shown in the figure, we found that we could easily obtain the risk probability of LVI for the NSCLC patient by combining the clinical risk factors and radiomics and heterogeneity models constructed based on their CT images. Case 1 achieved a very high-risk score and proven to be LVI-positive pathologically, whereas case 2 was categorized to be LVI-negative (low risk score). Figure 6 displays the Decision Curve Analysis (DCA) curves for the constructed intratumoral and peritumoral traditional radiomics models, heterogeneity models, clinical models, and combined models. The observations from the figure are as follows: (1) Across most risk threshold ranges, using these six models yields higher net benefits compared to the "treat all" and "treat none" strategies. (2) Except in the risk threshold range of approximately 0.83–0.9, the combined model achieves the highest net benefit in other ranges. (3) Across most risk threshold ranges, the clinical benefits of the models align with the conclusions drawn from model comparisons in the previous subsection: the combined model shows the highest net benefit, the peritumoral model achieves higher net benefits than the intratumoral model, and the traditional radiomics model provides higher net benefits than the heterogeneity model. Discussion In this study, a model combined clinical data, radiologic features, radiolomic features and tumor heterogeneity features was established to diagnose LVI status for early stage NSCLC. ITR, ITH and PTR, PTH models were established by analyzing the subregions of tumor and peritumor on CT images. The final model included VCS, N stage, ITH model, PTR_0 + 6 model and PTH_0 + 6 model, which was used to quantify the probability of LVI in patients. This model has high diagnostic accuracy, with an AUC of 0.962(95% CI: 0.937-0.988) in the training group and 0.882(95% CI: 0.831-0.933) in the validation group. Several previous studies have constructed models to diagnose LVI status of NSCLC by either imaging features or combining imaging features with clinical features. Chen et al [35] developed an LVI predictive model combined independent predictors(smoking and clinical stage) and the GPTV9 radiomic score, with an AUC of 0.89,0.83, and 0.66 in the training, internal validation, and external validation groups, respectively. Zhang et al [36] developed a LVI predictive model based on 2D and 3D tumor and peritumoral features, with an average AUC of 0.759. The AUC of the synthetic model established in this study is respectively 0.962 and 0.882 in the training and validation corhort, which is better than the previous studies. The reason may be that we not only integrated clinical data, imaging features, ITR and PTR features, but also added ITH and PTH features. NSCLC presents a large intratumor heterogeneity [37] . A good ITH model requires both imaging features and their spatial distribution. However, current research typically only captures a portion of the information. The definition of computational features in traditional radiomics involves the assumption of uniform distribution heterogeneity, without quantifying the local features of tumors [38] . While the ITH model uses the intensity of different tumors to group pixels and identify similar subregions. And associate the statistical features of each sub region (such as its LVI state), while integrating local radiometric features and global pixel distribution patterns. The model established by Shi et al [39] integrates the clinicopathological information, ITH characteristics and c-radiation group characteristics, and has a good prediction effect on the complete remission of neoadjuvant chemotherapy pathology in breast cancer patients (AUC values were 0.83-0.87 in the test set) . However, The AUC of the combined model of clinical imaging and traditional histology was only 0.78-0.81 in the test set, and 0.74-0.76 in ITH model alone. Thus, the addition of ITH features to the pathologic complete response (pCR) prediction model of breast cancer may increase clinical utility The ITH model in this study showed good discrimination ability for the state of LVI. The AUC of validation groups in GTV-ITH and PTH_0-6model were 0.607 and 0.692, respectively. The AUC in GTV and PTR_0-6 models were 0.665 and 0.824, respectively. When we add the heterogeneity model to the traditional radiomic model, the performance of the combinatorial model is greatly improved, and the AUC value reaches 0.882. It may be that ITH models use multi-region image features to characterize intratumoral spatial heterogeneity, this could further refine the shortcomings of traditional radiomic models, adding abundant information on tumor heterogeneity within the ROI. Tumor cells are often highly aggressive, disrupting the normal structure of surrounding parenchymal tissue, leading to carcinogenic infiltration of small blood vessels and lymphatic vessels around the lesion, which is often overlooked in studies that focus on intratumoral areas [24,40] . Therefore, detecting the boundary of lung cancer may help to quantify the invasiveness of the tumor. The study of NSCLC prediction by peritumoral imaging features has become an active and important field. However, the use of peritumoral imaging features in predicting the invasiveness of NSCLC has only been studied in a few studies. Previous studies defined the peritumoral area as 1.5 to 20 mm [24,41-43] . One study quantified the distance to micrometastases in histopathology lung cancer, resulting in mean distances of 2.94 mm and 2.69 mm for adenocarcinoma and squamous cell carcinoma of the lung to surrounding micrometastases, respectively [44] . Based on this, we defined the peritumoral extent on a 3 mm gradient to explore the relationship between vascular invasion status in NSCLC and peritumoral imaging. The results showed that Compared with PTR_0 - 3 and PTR_-3 - 3 , the AUC of PTR_0 - 6 has the highest diagnostic value (0.696,0.798 and 0.824). The reason may be that the farther away from the tumor, the higher the reproducibility of imaging features. This finding may be related to the presence of homogeneous lung parenchyma in the distal peri-tumour [42] . Therefore, PTR_0 - 6 model performs better than other models in our study. In addition, the performance of the peritumoral model is generally better than that of the intratumoral model in this study, whether it is the traditional radiomic model or the heterogeneity model. Consistent with the findings of Dou TH, et al. [45] , their model for predicting distant metastasis in NSCLC patients has a higher prognostic value for tumor marginal radiological features than for tumor radiological features alone; The comparison between the two was statistically significant (p = 0.048) .This may be due to the presence of increased cancer invasion and metastatic activity around the tumour, such as epithelial-mesenchymal Metastasis [46] , Tumor-associated macrophage [47,48] , tumour budding [49] and lymphatic vascular invasion [50,51] . GLM model analysis found that vessel convergence and N stage were independent predictors of LVI, and gender was a marginal correlation factor of LVI. This study established a clinical-imaging model based on this. Of these, N stage was most strongly associated with LVI status in NSCLC patients, as LVI is an initial manifestation of nodal metastasis [52] . The vessel convergence sign is the manifestation of vascular structure being pulled by the focus to concentrate in the direction of the focus or truncated through the focus or at the edge of the focus. This is mainly due to the fact that the Vascular endothelial growth factor secreted by malignant nodules through cancer cells promote the growth of microvessels in tumor tissues and the destruction of some blood vessel walls, which is consistent with the results of this study. A study [53] has found that the density of memory B cells, which play an important role in human anti-tumor immunity, is higher in lung adenocarcinoma tissues of female patients. Additional studies have suggested that lung cancer in women may have a different genetic profile from that in men because of different natural histories as well as female characteristics (younger age at diagnosis, non-smokers, more likely to have adenocarcinoma than men) [54,55] . This may explain some of the gender differences observed in this study regarding tumor invasion of vessels. However, this study has some limitations. Firstly, in this study, semi-automatic method was used to segment ROI, which may lead to artificial differences. An accurate automatic segmentation method should be considered in future research. Secondly, only perform radiomics feature extraction on CT plain scan images. CT enhancement or PET images may contain additional valuable information. Thirdly, this study only used an internal validation queue, and due to the limited sample size of the public database, external validation cannot be conducted. More large-scale studies are needed in the future to further validate. In summary, we have constructed a total of 6 models, and within most risk threshold ranges, using these 6 models can achieve higher net benefits than the "all treatment" and "no treatment at all" strategies. Among them, the combination model that integrates clinical data, traditional radiological features, peritumor radiomic features, peritumor heterogeneity features has the highest diagnostic accuracy for the LVI status in early stage NSCLC patients. And DCA analysis shows that within most risk thresholds, combining multiple models improves the clinical value of the models. The comprehensive column chart model not only enables non-invasive preoperative risk assessment of lung lesions, but also helps to provide objective guidance for rational clinical decision-making. Declarations Date availability The datasets generated during and/or analysed during the current study are not openly available for reasons of patient confidentiality however anonymised data may be available from the corresponding author on reasonable request. Authors' contributions Conceptualization: Hanfei Zhang, Meiyan Liao Data curation: Yun Long, Hanfei Zhang, Yang Guo, Tian Gan, Jingting Wang, Ting Li, Kemeng Zhang, Jun Chen, Yang Li, Meiyan Liao, Feng Xiao Formal analysis: Yun Long, Feng Xiao Investigation: Yun Long, Hanfei Zhang, Yang Guo, Tian Gan, Jingting Wang, Ting Li, Kemeng Zhang, Jun Chen, Yang Li, Meiyan Liao, Feng Xiao Methodology: Yun Long, Hanfei Zhang, Meiyan Liao, Feng Xiao Project administration: Meiyan Liao, Feng Xiao Resources: Meiyan Liao, Feng Xiao Software: Jun Chen, Yang Li, Feng Xiao Supervision: Meiyan Liao, Feng Xiao Validation: Yun Long, Hanfei Zhang, Jun Chen, Yang Li, Meiyan Liao, Feng Xiao Visualization: Yun Long, Hanfei Zhang, Jun Chen, Yang Li, Meiyan Liao, Feng Xiao Writing-original draft: Yun Long, Hanfei Zhang Writing-review & editing: Yun Long, Hanfei Zhang, Meiyan Liao, Feng Xiao acknowledgement The authors have no acknowledgements to declare. Funding No funding was received for this study. Ethics declarations Ethics approval and consent to participate This study was approved by the Ethics Committee of Zhongnan Hospital of Wuhan University, with the ethical approval number 2021048 (IRB number). Individual informed consent was waived because this was a retrospective study. All methods were performed in accordance with the relevant guidelines and regulations. Consent for publication Manuscript is approved by all authors for publication Competing interests Conflict of interest The authors of this manuscript declare no relationships with any companies whose products or services may be related to the subject matter of the article. References Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024 May-Jun;74(3):229-263. Chen P, Liu Y, Wen Y, et al. Non-small cell lung cancer in China. Cancer Commun (Lond). 2022 Oct;42(10):937-970. Travis WD, Brambilla E, Nicholson AG, et al. The 2015 World Health Organization Classification of Lung Tumors: Impact of Genetic, Clinical and Radiologic Advances Since the 2004 Classification. J Thorac Oncol. 2015 Sep;10(9):1243-1260. Günther K., Leier J., Henning G., et al. Prediction of lymph node metastasis in colorectal carcinoma by expressionof chemokine receptor CCR7[J]. Int. J. Cancer, 2005, 116:726–733. Kodama J., Kusumoto T., Seki N., et al. Association of CXCR4 and CCR7 chemokine receptor expression and lymph node metastasis in human cervical cancer[J]. Ann. Oncol, 2007, 18:70–76. Almofti A., Uchida D., Begum N.M., et al. The clinicopathological significance of the expression of CXCR4 protein in oral squamous cell carcinoma[J]. Int. J. Oncol, 2004, 25:65–71. Schimanski C.C., Bahre R., Gockel I., et al. Dissemination of hepatocellular carcinoma is mediated via chemokine receptor CXCR4[J]. Br. J. Cancer, 2006, 95:210–217. Dieterich L.C., Kapaklikaya K., Cetintas T., et al. Transcriptional profiling of breast cancer-associated lymphatic vessels reveals VCAM-1 as regulator of lymphatic invasion and permeability[J]. Int. J. Cancer, 2019, 145:2804–2815. Garmy-Susini B., Avraamides C.J., Schmid M.C., et al. Integrin α4β1 Signaling Is Required for Lymphangiogenesis and Tumor Metastasis[J]. Cancer Res, 2010, 70:3042–3051. Vestweber D. How leukocytes cross the vascular endothelium[J]. Nat. Rev. Immunol, 2015, 15:692–704. Tacconi C., Correale C., Gandelli A., et al. Vascular Endothelial Growth Factor C Disrupts the Endothelial Lymphatic Barrier to Promote Colorectal Cancer Invasion[J]. Gastroenterology, 2015, 148:1438–1451. Nakayama H., Ohuchida K., Yonenaga A., et al. S100P regulates the collective invasion of pancreatic cancer cells into the lymphatic endothelial monolayer[J]. Int. J. Oncol, 2019, 55:211–222. Mohammed R.A., Martin S.G., Mahmmod A.M., et al. Objective assessment of lymphatic and blood vascular invasion in lymph node-negative breast carcinoma: Findings from a large case series with long-term follow-up[J]. J. Pathol, 2011, 223:358–365. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation[J]. Cell, 2011, 144: 646–674. Brown M., Assen F.P., Leithner A., et al. Lymph node blood vessels provide exit routes for metastatic tumor cell dissemination in mice[J]. Science, 2018, 359:1408–1411. Gabor S, Renner H, Popper H, et al. Invasion of blood vessels as significant prognostic factor in radically resected T1-3N0M0 non-small cell lung cancer[J]. Eur J Cardiothorac Surg, 2004, 25:439–42. Hishida T, Yoshida J, Maeda R, et al. Prognostic impact of intratumoural microvascular invasion and microlymphatic permeation on node-negative non-small cell lung cancer: which indicator is the stronger prognostic factor? [J]. Eur J Cardiothorac Surg, 2013, 43:772–7. Collier FC, Blakemore WS, Kyle R H, et al. Carcinoma of the lung: factors which influence five year survival with special reference to blood vessel invasion[J]. Ann Surg, 1957, 146(3): 417-423. Neri S, Yoshida J, Ishii G, et al. Prognostic impact of microscopic vessel invasion and visceral pleural invasion in non-small cell lung cancer: a retrospective analysis of 2,657 patients[J]. Ann Surg, 2014, 260(2): 383-388. Coroller TP, grossmann P, hou Y, et al. CT-based radiomic signature predicts distant metastasis in lung adenocarcinoma[J]. Radiother Oncol, 2015, 114:345-50. Wang X, Wan Q, Chen H, et al. Classification of pulmonary lesion based on multiparametric MRI: utility of radiomics and comparison of machine learning methods[J]. EurRadiol, 2020, 30(8):4595–4605. Dercle L, Fronheiser M, Lu L, et al. Identification of nonsmall cell lung cancer sensitive to systemic cancer therapies using radiomics[J]. Clin Cancer Res, 2020, 26(9):2151–2162. Kadota K, Nitadori J-i, Sima CS, et al. Tumor Spread through Air Spaces is an Important Pattern of Invasion and Impacts the Frequency and Location of Recurrences after Limited Resection for Small Stage I Lung Adenocarcinomas[J]. Journal of Thoracic Oncology, 2015, 10: 806–14. Matsumura Y, Ishii G, Aokage K, et al. Morphophenotypic characteristics of intralymphatic cancer and stromal cells susceptible to lymphogenic metastasis[J]. Cancer Science, 2012, 103: 1342–7. Onozato ML, Kovach AE, Yeap BY, et al. Tumor islands in resected early-stage lung adenocarcinomas are associated with unique clinicopathologic and molecular characteristics and worse prognosis[J]. Am J Surg Pathol, 2013, 37: 287–94. Shimada Y, Ishii G, Hishida T, et al. Extratumoral vascular invasion is a significant prognostic indicator and a predicting factor of distant metastasis in non-small cell lung cancer[J]. J Thorac Oncol, 2010; 5: 970–5. Uruga H, Fujii T, Fujimori S,, et al. Semiquantitative Assessment of Tumor Spread through Air Spaces (STAS) in Early-Stage Lung Adenocarcinomas[J]. J Thorac Oncol, 2017, 12: 1046–51. Saijo T, Ishii G, Ochiai A, et al. Evaluation of extratumoral lymphatic permeation in non-small cell lung cancer as a means of predicting outcome[J]. Lung Cancer, 2007, 55:61–6. Koukourakis MI, Giatromanolaki A, Thorpe PE, et al. Vascular endothelial growth factor/KDR activated microvessel density versus CD31 standard microvessel density in non-small cell lung cancer[J]. Cancer Res, 2000, 60: 3088–95. Tsutani Y,Suzuki K,Koike T, et al.High-risk factors for recurrence of stage I lung adenocarcinoma: follow-up data from JCOG0201[J].Ann Thorac Surg, 2019, 108(5):1484-1490. Nwogu I, Corso JJ. Exploratory identification of image-based biomarkers for solid mass pulmonary tumors[J]. Med Image Comput Comput Assist Interv, 2008, 11(Pt 1):612-9. Lambin P, Rios-Velazquez E, Leijenaar R, et al. Radiomics: extracting more information from medical images using advanced feature analysis[J]. Eur J Cancer, 2012, 48(4):441-6. Zwanenburg A, Vallières M, Abdalah MA, et al. The Image Biomarker Standardization Initiative: Standardized Quantitative Radiomics for High-Throughput Image-based Phenotyping,et al. Radiology, 2020, 295(2):328-338. Gillies RJ, Kinahan PE, Hricak H. Radiomics: images are more than pictures, they are data[J]. Radiology, 2016, 278:563–577. Chen Q, Shao J, Xue T, et al. Intratumoral and peritumoral radiomics nomograms for the preoperative prediction of lymphovascular invasion and overall survival in non-small cell lung cancer. Eur Radiol. 2023 Feb;33(2):947-958. Zhang X, Zhang G, Qiu X, et al. Radiomics under 2D regions, 3D regions, and peritumoral regions reveal tumor heterogeneity in non-small cell lung cancer: a multicenter study. Radiol Med. 2023 Sep;128(9):1079-1092. Zhang J, Fujimoto J, Zhang J, et al. Intratumor heterogeneity in localized lung adenocarcinomas delineated by multiregion sequencing. Science. 2014 Oct 10;346(6206):256-9. Subramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles[J]. Proc Natl Acad Sci U S A, 2005, 102(43):15545-50. Shi Z, Huang X, Cheng Z, et al. MRI-based Quantification of Intratumoral Heterogeneity for Predicting Treatment Response to Neoadjuvant Chemotherapy in Breast Cancer. Radiology. 2023 Jul;308(1):e222830. Wang XL, Fang JP, Tang RY, et al. Different significance between intratumoral and peritumoral lymphatic vessel density in gastric cancer: a retrospective study of 123 cases[J]. BMC Cancer, 2010, 17;10:299. Padera TP, Kadambi A, di Tomaso E, et al. Lymphatic metastasis in the absence of functional intratumor lymphatics[J]. Science, 2002, 296(5574):1883-6. Saijo T, Ishii G, Ochiai A, et al. Evaluation of extratumoral lymphatic permeation in non-small cell lung cancer as a means of predicting outcome[J]. Lung Cancer, 2007, 55(1):61-6. Shimada Y, Ishii G, Hishida T, et al. Extratumoral vascular invasion is a significant prognostic indicator and a predicting factor of distant metastasis in non-small cell lung cancer[J]. J Thorac Oncol, 2010, 5(7):970-5. Moon Y, Sung SW, Lee KY, et al. Clinicopathological characteristics and prognosis of non-lepidic invasive adenocarcinoma presenting as ground glass opacity nodule[J]. J Thorac Dis, 2016, 8(9):2562-2570. Dou TH, Coroller TP, van Griethuysen JJM, et al. Peritumoral radiomics features predict distant metastasis in locally advanced NSCLC. PLoS One. 2018 Nov 2;13(11):e0206108. Mahmood MQ, Ward C, Muller HK, et al. Epithelial mesenchymal transition (EMT) and non-small cell lung cancer (NSCLC): a mutual association with airway disease. Med Oncol. 2017 Mar;34(3):45. Joyce JA, Pollard JW. Microenvironmental regulation of metastasis. Nat Rev Cancer. 2009 Apr;9(4):239-52. Qian BZ, Pollard JW. Macrophage diversity enhances tumor progression and metastasis. Cell. 2010 Apr 2;141(1):39-51. Yamaguchi Y, Ishii G, Kojima M, et al. Histopathologic features of the tumor budding in adenocarcinoma of the lung: tumor budding as an index to predict the potential aggressiveness. J Thorac Oncol. 2010 Sep;5(9):1361-8.. Shimada Y, Ishii G, Hishida T, et al. Extratumoral vascular invasion is a significant prognostic indicator and a predicting factor of distant metastasis in non-small cell lung cancer. J Thorac Oncol. 2010 Jul;5(7):970-5. Tammela T, Alitalo K. Lymphangiogenesis: Molecular mechanisms and future promise. Cell. 2010 Feb 19;140(4):460-76. Lee H, Choi H. Investigating the Clinico-Molecular and Immunological Evolution of Lung Adenocarcinoma Using Pseudotime Analysis[J].. Front Oncol, 2022, 12:828505. Wu DM, Deng SH, Zhou J, et al. PLEK2 mediates metastasis and vascular invasion via the ubiquitin-dependent degradation of SHIP2 in non-small cell lung cancer[J]. Int J Cancer , 2020, 146(9):2563-2575. Alwithenani A, Bethune D, Castonguay M, et al. Profiling non-small cell lung cancer reveals that PD-L1 is associated with wild type EGFR and vascular invasion, and immunohistochemistry quantification of PD-L1 correlates weakly with RT-qPCR[J]. PLoS One, 2021, 16(5):e0251080. Ramnefjell M, Aamelfot C, Helgeland L, et al. Vascular invasion is an adverse prognostic factor in resected nonsmall-cell lung cancer[J]. APMIS, 2017, 125(3):197–206. Additional Declarations No competing interests reported. Supplementary Files BMISupplementary.docx Cite Share Download PDF Status: Published Journal Publication published 19 Dec, 2025 Read the published version in BMC Medical Imaging → Version 1 posted Editorial decision: Revision requested 18 Aug, 2025 Reviews received at journal 13 Aug, 2025 Reviews received at journal 28 Jun, 2025 Reviewers agreed at journal 21 Jun, 2025 Reviewers agreed at journal 21 Jun, 2025 Reviewers invited by journal 19 Jun, 2025 Editor assigned by journal 18 Jun, 2025 Submission checks completed at journal 18 Jun, 2025 First submitted to journal 18 Jun, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6764701","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":474390290,"identity":"7a733c1e-262b-4405-a8ec-16ec57e8fed9","order_by":0,"name":"Yun Long","email":"","orcid":"","institution":"Zhongnan Hospital of Wuhan University","correspondingAuthor":false,"prefix":"","firstName":"Yun","middleName":"","lastName":"Long","suffix":""},{"id":474390291,"identity":"ace29253-987b-4024-982b-4cdc8e832968","order_by":1,"name":"Hanfei Zhang","email":"","orcid":"","institution":"Zhongnan Hospital of Wuhan University","correspondingAuthor":false,"prefix":"","firstName":"Hanfei","middleName":"","lastName":"Zhang","suffix":""},{"id":474390292,"identity":"9aef7077-6f0d-451e-b8a4-ef6b4e565ebe","order_by":2,"name":"Yang Guo","email":"","orcid":"","institution":"Zhongnan Hospital of Wuhan University","correspondingAuthor":false,"prefix":"","firstName":"Yang","middleName":"","lastName":"Guo","suffix":""},{"id":474390293,"identity":"42a91c45-c2b6-4a8b-9ed3-fd70055e1f0f","order_by":3,"name":"Tian Gan","email":"","orcid":"","institution":"Zhongnan Hospital of Wuhan University","correspondingAuthor":false,"prefix":"","firstName":"Tian","middleName":"","lastName":"Gan","suffix":""},{"id":474390294,"identity":"38b61c75-7cc2-4c02-8d93-9aa996866305","order_by":4,"name":"Jingting Wang","email":"","orcid":"","institution":"Zhongnan Hospital of Wuhan University","correspondingAuthor":false,"prefix":"","firstName":"Jingting","middleName":"","lastName":"Wang","suffix":""},{"id":474390295,"identity":"f194efcd-4092-4a7d-9a24-b268776f2b07","order_by":5,"name":"Ting Li","email":"","orcid":"","institution":"Zhongnan Hospital of Wuhan University","correspondingAuthor":false,"prefix":"","firstName":"Ting","middleName":"","lastName":"Li","suffix":""},{"id":474390296,"identity":"7f8e9e99-2814-4506-be7b-b9f3dee641c3","order_by":6,"name":"Kemeng Zhang","email":"","orcid":"","institution":"Zhongnan Hospital of Wuhan University","correspondingAuthor":false,"prefix":"","firstName":"Kemeng","middleName":"","lastName":"Zhang","suffix":""},{"id":474390297,"identity":"c874a197-7c8b-41eb-9464-d2946bdc9480","order_by":7,"name":"Jun Chen","email":"","orcid":"","institution":"Bayer Healthcare","correspondingAuthor":false,"prefix":"","firstName":"Jun","middleName":"","lastName":"Chen","suffix":""},{"id":474390298,"identity":"538e9975-5617-434b-a8ad-7932f2635417","order_by":8,"name":"Yang Li","email":"","orcid":"","institution":"Zhongnan Hospital of Wuhan University","correspondingAuthor":false,"prefix":"","firstName":"Yang","middleName":"","lastName":"Li","suffix":""},{"id":474390299,"identity":"0c1711ad-c50c-4037-863c-51a7fe92da21","order_by":9,"name":"Meiyan Liao","email":"","orcid":"","institution":"Zhongnan Hospital of Wuhan University","correspondingAuthor":false,"prefix":"","firstName":"Meiyan","middleName":"","lastName":"Liao","suffix":""},{"id":474390300,"identity":"4fcabd51-e920-490b-877f-a486bd3fa75f","order_by":10,"name":"Feng Xiao","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA4ElEQVRIiWNgGAWjYFACxgYgIVE/n5n5wIEPP4jXYsO4sZ0t8eDMHuKtSmNsOM9jfJiDjQi18u3NbRI/GA4zMzbzfDjMwMMgzy92AL8WgzMH2yR7GA6zsTPzbjhcYMFgOHN2AgEtEoltEjwMh3kYm4FaZvAwJBjcJqBFfv7DNsk/DIclgLoeHOZhI0ILww3GNmkehjQDoBYG4rQYnElstpZhsEkwbGYzAAayBGG/yLcff3jzDYNEgjz/4ccfPvywkeeXJuQwBgYWCcZ/cI4EQeUgwPyBKGWjYBSMglEwcgEAhL1C9a7d6V4AAAAASUVORK5CYII=","orcid":"","institution":"Zhongnan Hospital of Wuhan University","correspondingAuthor":true,"prefix":"","firstName":"Feng","middleName":"","lastName":"Xiao","suffix":""}],"badges":[],"createdAt":"2025-05-28 06:23:14","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6764701/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6764701/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s12880-025-02041-0","type":"published","date":"2025-12-19T15:57:43+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":85347154,"identity":"6690873d-65f0-497d-adb4-40676dd27093","added_by":"auto","created_at":"2025-06-25 02:17:20","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":116686,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eThe data flowchart of this study. \u003c/strong\u003eThe upper part of the figure illustrates the screening process for clinical risk factors. The lower part depicts the modeling process based on CT images. Four models (including ITR, ITH, PTR, and PTH) were developed separately for intratumoral and peritumoral regions. These models, along with the selected clinical variables, were then used as independent risk factors to construct the combined model.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-6764701/v1/810b8fe28480fbf35006a6de.png"},{"id":85348513,"identity":"73243d10-6aeb-4227-a85c-d331d1b974fd","added_by":"auto","created_at":"2025-06-25 02:25:20","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":154352,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eA example of the manual segmentation for the GTV ROI and its corresponding automatically dilated PTV ROIs. \u003c/strong\u003eFirstly, the GTV ROI for the tumors were manually segmented by the radiologists, and then automatically dilated into three PTV ROIs using the SimpleITK program packages.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-6764701/v1/ccf52a2b98451242d5356ab2.png"},{"id":85347156,"identity":"23a5a1f7-dcf2-4dbc-86e3-fdea06463970","added_by":"auto","created_at":"2025-06-25 02:17:20","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":38623,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eThe nomogram draw based on the final constructed Combined model. \u003c/strong\u003eIn the nomogram, each risk factor is assigned a specific score (Points). These individual scores are summed to calculate a total risk score (Total Points), which corresponds to the final risk probability value, indicating the likelihood of LVI.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-6764701/v1/ffdb3debb6d962f28a7f6f75.png"},{"id":85347161,"identity":"eac9a5d4-0402-47c0-a7f3-34ab07b85bed","added_by":"auto","created_at":"2025-06-25 02:17:21","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":88440,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eThe ROC of the constructed models in this study. \u003c/strong\u003eROC was used to\u003cstrong\u003e \u003c/strong\u003eassess the prediction accuracy of the constructed models\u003cstrong\u003e \u003c/strong\u003ein both the train set (A)and test set (B). The diagonal dashed line represented the AUC value of 0.5, which means a completely random prediction.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-6764701/v1/9edbbb6961c258479e346a4b.png"},{"id":85347162,"identity":"c62481f8-4e5f-4936-8b7d-149f3561f626","added_by":"auto","created_at":"2025-06-25 02:17:21","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":443499,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eExamples for the use of nomogram to predict the LVI risk of the NSCLC. \u003c/strong\u003e(A)\u003cstrong\u003e \u003c/strong\u003ea patient with a clinical stage of T1cN1M0 presented with imaging findings showing a vascular convergence sign. The nomogram predicted a 0.965 probability of LVI positive. This was subsequently confirmed through surgical pathology, diagnosing the patient with adenocarcinoma, and identifying the presence of vascular invasion; (B) a patient with a clinical stage of T1bN0M0 presented with imaging findings showing no evidence of a vascular convergence sign. The nomogram predicted a 0.0024 probability of LVI positive. Surgical pathology confirmed the diagnosis of squamous cell carcinoma, with no evidence of vascular invasion.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-6764701/v1/425f840ce80ca136f3b359ca.png"},{"id":85348519,"identity":"9b7d2812-fbf3-4a87-93a3-101d085fe908","added_by":"auto","created_at":"2025-06-25 02:25:22","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":92893,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eThe DCA of the constructed models in this study.\u003c/strong\u003e All models were evaluated in both training set (A) and test set (B). Treat-all strategy: The net benefit in the condition that all patients were treated as Positive; Treat-none: All patients were treated as Negative.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-6764701/v1/0d9cb612276f2a4445913859.png"},{"id":98815229,"identity":"8f025577-9b37-477b-b775-5f6f80cc04b3","added_by":"auto","created_at":"2025-12-22 16:14:13","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2154510,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6764701/v1/dd5cbe0c-0ffe-46bc-8132-56e583d8f170.pdf"},{"id":85347157,"identity":"4aca5e30-5516-4e46-8ae7-42471a15fff1","added_by":"auto","created_at":"2025-06-25 02:17:20","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":1399374,"visible":true,"origin":"","legend":"","description":"","filename":"BMISupplementary.docx","url":"https://assets-eu.researchsquare.com/files/rs-6764701/v1/716f99c1fbc31821cdc46ad0.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"CT based Quantification of Intratumoral and Peritumoral Heterogeneity for diagnosing Lymphovascular Invasion for Early Stage Non-small cell lung cancer","fulltext":[{"header":"Introduction","content":"\u003cp\u003eCancer is a major societal, public health, and economic problem in the 21st century; and lung cancer has the highest morbidity and mortality among all cancer types globally\u003csup\u003e[\u003c/sup\u003e\u003csup\u003e1\u003c/sup\u003e\u003csup\u003e]\u003c/sup\u003e. Regrettably, about half of all new lung cancer cases were in Asia\u003csup\u003e[\u003c/sup\u003e\u003csup\u003e2\u003c/sup\u003e\u003csup\u003e]\u003c/sup\u003e. There are 2 main forms of lung cancer: non small cell lung cancer (NSCLC) and small cell lung cancer, and NSCLC account for 85%\u003csup\u003e[\u003c/sup\u003e\u003csup\u003e3\u003c/sup\u003e\u003csup\u003e]\u003c/sup\u003e. Treatment of NSCLC is stage specific. Patients with stage I or II should be treated with complete surgical resection when not contraindicated. Nonsurgical patients should be considered for radiotherapy or chemotherapy\u003csup\u003e[\u003c/sup\u003e\u003csup\u003e3\u003c/sup\u003e\u003csup\u003e]\u003c/sup\u003e. Whereas 25% -70% of surgical patients eventually relapse despite complete resection, with 5-year survival of only 35% -65% \u003csup\u003e[4]\u003c/sup\u003e . Distant metastasis is the most common form of postoperative recurrence\u003csup\u003e[5]\u003c/sup\u003e. Studies have shown that cancer must first enter and spread to the entire vasculature before it can spread and metastasize through blood vessels or lymphatic vessels \u003csup\u003e[6,7]\u003c/sup\u003e .\u003c/p\u003e\n\u003cp\u003eThe detection of neoplastic cells in arterial, venous, or lymphatic lumen with H\u0026amp;E stain is called lymphovascular invasion (LVI) \u003csup\u003e[8-10]\u003c/sup\u003e . The presence of LVI in patients with lung cancer is associated with poor prognosis \u003csup\u003e[11]\u003c/sup\u003e , and LVI is an independent prognostic factor for recurrence-free survival in patients with stage I lung adenocarcinoma undergoing lobectomy \u003csup\u003e[12]\u003c/sup\u003e . Studies have shown that LVI can be used as an indication for preoperative neoadjuvant chemotherapy in NSCLC patients. Preoperative neoadjuvant chemotherapy can reduce tumor volume, shorten tumor stage, and provide long-term survival benefit\u003csup\u003e[13,14]\u003c/sup\u003e. Lobectomy plus lymphadenectomy is more effective in patients with LVI-positive NSCLC compared with subresection alone \u003csup\u003e[15]\u003c/sup\u003e .Therefore, it is very important to identify LVI invasion before surgery or other therapy. At present, the diagnosis of LVI of lung cancer is mainly by HE staining, immunohistochemistry and special staining after operation. LVI is a histological condition that can only be recognized postoperatively by surgical specimens\u003csup\u003e[16,17]\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eNSCLC exhibits strong intratumor heterogeneity (ITH), and more heterogeneous tumors have higher invasive and metastatic potential \u003csup\u003e[18]\u003c/sup\u003e . Studies have shown that invasive tumors with LVI have significantly higher ITH scores than non-invasive tumors \u003csup\u003e[19]\u003c/sup\u003e . Because of different ITH, patients with other similar clinical features, such as clinical stage, molecular subtype, have different LVI status. Therefore, ITH could help to predict LVI status.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ePrevious studies have focused on the relationship between clinical outcomes and radiological features within the primary tumor volume, yet the lung parenchyma surrounding the primary tumor may also be involved in tumor invasion and metastasis \u003csup\u003e[20,21]\u003c/sup\u003e . Pathological studies have shown that lung tumors can cause worse clinical manifestations through lymphatic metastasis, hematogenous metastasis, or direct invasion of surrounding lung tissues \u003csup\u003e[22-29]\u003c/sup\u003e . Furthermore, the association of tumor peripheral cancer cell infiltration with local recurrence or distant metastasis was significantly stronger compared with intratumoral cancer tissure \u003csup\u003e[22,25,27]\u003c/sup\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAmong preoperative non-invasive diagnostic methods, conventional computer tomography (CT) is the most commonly used one. But studies have shown that CT-based diagnosis of vascular and mediastinal tumor invasion has limited utility \u003csup\u003e[30-32]\u003c/sup\u003e , subtle differences in some of the underlying features can not be identified with the naked eye. Radiomics, as an image analysis technique, can extract complex information that is difficult for the human eye to recognize or quantify from a diagnostic image and convert it into quantitative data \u003csup\u003e[33,34]\u003c/sup\u003e .\u003c/p\u003e\n\u003cp\u003eThus, based on the importance of diagnosing LVI before surgery and the lack of corresponding method, we establish a model integrated clinical, traditional radiologic, intratumoral and peritumoral radiomics (ITR and PTR), and intratumoral and peritumoral heterogeneity (ITH and PTH) features to diagnose LVI status for early stage NSCLC.\u0026nbsp;\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cstrong\u003ePatients\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis retrospective study was approved by the Ethics Committee of Zhongnan Hospital of Wuhan University, with the ethical approval number 2021048 (IRB number). According to the CIMOS guidelines, informed consent was waived for this study. It involved collecting pathological data and chest CT images from patients (Figure S1) who underwent surgical resection for non-small cell lung cancer (NSCLC) between January 2019 and May 2021. All patients had their surgeries within two weeks of their CT scans, which included routine blood tests measuring tumor markers such as Carbohydrate Antigen 125 (CA125, reference value ≤35 U/mL), Carcinoembryonic Antigen (CEA, reference value 0.0-5.0 ng/mL), and Neuron Specific Enolase (NSE, reference value ≤16.0 ng/mL). The inclusion criteria were: 1) pathologically confirmed NSCLC; 2) availability of a pathological report with clear LVI status; 3) a diagnostic-compliant chest CT performed within two weeks prior to surgery; and 4) complete baseline clinical data. The exclusion criteria included: 1) lack of complete clinical-pathological data or preoperative CT images; 2) presence of other malignant tumors; and 3) prior anti-tumor treatment before surgery.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Flowchart\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAs shown in Figure 1, the data analysis process in this study consists of two main parts: clinical risk factors analysis/modeling, and image data analysis/modeling. The analysis of clinical risk factors is conducted through univariate and multivariate analyses step by step, while image data processing includes three steps: preprocessing, segmentation, and model construction. Notably, regions of interest (ROI) within the tumor (intratumoral) and around the tumor (peritumoral) were analyzed respectively, and both the traditional radiomics models and the heterogeneity models that well reflecting tumor heterogeneity were developed separately. Ultimately, integrating selected clinical risk factors with the intratumoral and peritumoral models, we established a combined model and draw the corresponding nomogram to enhance its clinical utility.\u003c/p\u003e\n\u003cp\u003eDuring data processing, all patients were randomly divided into training and testing sets in a 1:1 ratio using the random stratified sampling method. The training set was used for model development and construction, while the testing set served as an independent sample for model validation. All model performance metrics were calculated and compared across both the training and testing sets.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eImage Processing\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBefore the formal analysis of the images, all patients' CT images underwent a two-step preprocessing: resampling and normalization of CT values. This ensures consistency and comparability of images acquired from different devices and batches, thereby enhancing the accuracy and generalizability of the diagnostic models. Specifically, we resampled all patients' chest CT images to a resolution of 1mm×1mm×1mm using linear interpolation, and normalized the CT values to a standard normal distribution N(0,1) using the Z-scoring method.\u003c/p\u003e\n\u003cp\u003eSubsequently, we manually segmented the tumor regions using ITK-SNAP (version 3.8.0; http://www.itksnap.org) to obtain the tumor regions of interest (ROIs). Based on this, the SimpleITK package was used to automatically contract the tumor ROI (gross tumor volume, GTV) by 3mm or expand it by 3mm and 6mm, defining three different peritumoral tumor volume (PTV) areas (Figure 2): from 3mm inside to 3mm outside the tumor boundary (PTV_-3-3), from the tumor boundary to 3mm outside (PTV_0-3), and from the tumor boundary to 6mm outside (PTV_0-6). We manually corrected the automatic expansion results to exclude adjacent pleura and bones included during the process. The delineation of the tumor ROI and the adjustment of the peritumoral areas were completed jointly by two radiologists under CT lung window settings (width 1500 HU; level -500 HU), and the ICC method was used to ensure stability and repeatability of the results. This process ultimately generated one intratumoral ROI and three peritumoral ROIs for subsequent modeling and analysis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eModel Construction\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFirstly, univariate analysis was used to evaluate clinical variables and radiological semantic features, selecting variables with p\u0026lt;0.05 as clinical risk factors. Subsequently, a multivariate linear regression method was employed to construct the Clinical model.\u003c/p\u003e\n\u003cp\u003eNext, we conducted traditional radiomics modeling. For the four tumor ROIs (GTV, PTV_-3-3, PTV_0-3, and PTV_0-6), 105 radiomics features were extracted (details in the supplementary document). Feature reduction was performed as follows step by step: (1) features with variances below a threshold of 1 were removed using the variance threshold method; (2) highly correlated features (R \u0026gt; 0.9) were eliminated using the Pearson correlation analysis method; (3) features showing significant differences between classification groups were identified using the t-test (p \u0026lt; 0.05); (4) the least absolute shrinkage and selection operator (LASSO) method was used to select the optimal feature subset and construct the corresponding traditional radiomics models(ITR, PTR_-3-3, PTR_0-3, and PTR_0-6).\u003c/p\u003e\n\u003cp\u003eFollowing this, we constructed heterogeneity models. The predictive performance of traditional radiomics models based on three peritumoral ROIs was compared to determine the best peritumoral ROI. Then, intratumoral heterogeneity models (ITH) and peritumoral heterogeneity models (PTH) were developed separately for the intratumoral ROI and the optimal peritumoral ROI. The construction of heterogeneity models involved the following steps: (1) subregion segmentation of the tumor, objectively identifying and segmenting subregions inside the ROI to better understand the internal structure and heterogeneity of the tumor; (2) unsupervised clustering analysis, classifying tumor subregions with similar features to reveal different ecological \"populations\" within the tumor ROI; (3) calculation of ecological diversity indices for different ecological \"populations\" within the tumor ROI; (4) constructing the \"Tumor Ecological Diversity Feature Vector\" (TED) based on the ecological diversity indices, i.e., the heterogeneity model. For specific details on each step's design and implementation, refer to the supplementary materials (Figure S1).\u003c/p\u003e\n\u003cp\u003eFinally, the selected clinical risk factors, intratumoral and peritumoral radiomics models (ITR and PTR), as well as intratumoral and peritumoral heterogeneity models (ITH and PTH) were used as independent risk factors for LVI. A multivariate linear regression method was utilized to construct a combined model, and a nomogram was created to visualize the final model, enhancing its clinical interpretability and usability.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eModel Validation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn this study, eight models were established: four traditional radiomics models (ITR, PTR_-3-3, PTR_0-3, and PTR_0-6) for the tumor ROI and three peritumoral ROIs, two heterogeneity models (ITH and PTH_0-6) for the intratumoral and the optimal peritumoral ROIs, a clinical model, and the final combined model. The predictive accuracy of these models was evaluated using ROC curves and their derived metrics, Area under the curve (AUC), accuracy (Acc), sensitivity (Sens), and specificity (Spec). The Delong test was used to determine whether differences in ROC curves between the models were significant. Additionally, the consistency between the model predictions and actual observations was assessed using calibration curves, and the models' goodness of fit was examined with the Hosmer-Lemeshow (HL) test. Finally, decision curves were used to evaluate the clinical benefits of the models at different risk thresholds.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatistics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eStatistical analyses were conducted using R software (version 4.1.2, http://www.R-project.org). For quantitative data following a normal distribution, the mean ± standard deviation (mean±SD) was used for presentation, and comparisons between groups were performed using independent t-tests. For data not following a normal distribution, the median and interquartile range (IQR) were reported, and comparisons between groups were made using the Mann-Whitney U test. Categorical data were analyzed using the Chi-square test or Fisher’s exact test. A p-value of less than 0.05 was considered statistically significant.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e \u003cb\u003ePatients\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThis study included 308 patients with non-small cell lung cancer (NSCLC), of whom 117 were LVI-positive and 191 were LVI-negative. Clinical imaging characteristics of the patients are detailed in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. After a 1:1 random assignment, there were 154 patients in the training group and 154 in the validation group. The training group consisted of 59 LVI-positive patients and 95 LVI-negative patients, while the validation group included 58 LVI-positive patients and 96 LVI-negative patients. The patients enrollment details were shown in Figure S2.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePatients demographics for training and test set.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eCharacteristic\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003eTraining set (154)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cem\u003ep\u003c/em\u003e-value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003eTest set (154)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cem\u003ep\u003c/em\u003e-value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLVI(+)\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;59)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLVI(-).\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;95)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eLVI(+)\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;58)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eLVI(-)\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;96)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAge\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e63.00 (58.20, 67.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e63.00 (56.20, 67.00)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.470\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e64.09\u0026thinsp;\u0026plusmn;\u0026thinsp;7.72\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e62.80\u0026thinsp;\u0026plusmn;\u0026thinsp;7.95\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.328\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eGender\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e13 (22.03%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e36 (37.89%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cb\u003e0.04*\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e14(24.14%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e33(34.38%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e0.181\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e46 (77.97%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e59 (62.11%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e44(75.86%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e63(65.62%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003ePathology\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eadenocarcinoma\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e42(71.19%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e77 (81.05%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e0.156\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e35(60.34%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e78(81.25%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e0.004\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSquamous Cell Carcinoma\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e17(28.81%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e18 (18.95%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e23(39.66%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e18(18.75%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eTNM Stages\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eI\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e26(44.07%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e68(71.58%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e\u003cb\u003e0.003*\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e20(34.48%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e72(75.00%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eII\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e25(42.37%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e22(23.16%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e27(46.55%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e16(16.67%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIII\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e8(13.56%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e5(5.26%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e10(17.24%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e8(8.33%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eN Stage\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e33(55.93%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e86(91.49%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;0.001*\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e35(60.34%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e84(91.30%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e25(42.37%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e8(8.51%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e22(37.93%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e8(8.70%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1(1.69%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0(0.00%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1(1.72%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0(0.00%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eMaxSize (mm)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3.24 (2.20, 4.40)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2.74 (2.15, 4.01)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.456\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.05(2.50, 4.78)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e2.60(1.77, 3.50)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.011\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eLocation\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ecentra\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e26(44.07%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e31(32.63%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e0.153\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e22(37.93%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e31(32.29%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e0.475\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eperiphral\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e33(55.93%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e64(67.37%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e36(62.07%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e65(67.71%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eDensity\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSolid\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e51(86.44%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e75(78.95%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e0.241\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e52(89.66%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e70(72.92%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e0.013\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSubsolid\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e8(13.56%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e20(21.05%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e6(10.34%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e26(27.08%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eSpiculation Sign\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e44(74.58%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e68(71.58%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e0.685\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e43(74.14%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e66(68.75%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e0.476\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e15(25.42%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e27(28.42%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e15(25.86%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e30(31.25%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eLobulation Sign.\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e54(91.53%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e88(92.63%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e56(96.55%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e90(93.75%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e0.701\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e5(8.47%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e7(7.37%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2(3.45%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e6(6.25%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003ePleural Indentation Sign.\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e28 (47.46%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e54(56.84%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e0.256\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e33(56.90%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e58(60.42%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e0.667\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e31(52.54%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e41(43.16%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e125(43.10%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e38(39.58%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eBronchus Sign\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e21(35.59%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e47(49.47%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e0.092\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e33(56.90%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e41(42.71%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e0.962\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e38(64.41%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e48(50.53%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e41(70.69%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e55(57.29%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eVascular Convergence Sign\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e37(62.71%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e84(88.42%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;0.001*\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e41(70.69%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e73(76.04%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e0.463\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e22(37.29%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e11(11.58%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e17(29.31%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e23(23.96%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eVacuole Sign\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e15(25.42%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e25(26.32%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cb\u003e0.022\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e14(24.14%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e21(21.88%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e0.745\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e44(74.58%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e70(73.68%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e44(75.86%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e75(78.12%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCEA\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3.45 (2.52, 7.21)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2.96 (2.02, 4.69)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u003cb\u003e0.035*\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.75 (2.25, 5.77)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e2.95 (1.77, 5.15)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.127\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCA-125\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e13.59 (9.78, 29.03)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e13.79 (9.86, 20.38)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.539\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e16.72(10.27, 25.96)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e13.44(8.79, 19.71)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.079\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eNSE\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e13.21 (10.65, 15.56)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e13.04 (10.81, 14.37)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.513\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e12.92(11.16, 14.85)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e12.10(10.71, 14.34)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.201\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eClinical Risk Factors and Modeling\u003c/b\u003e \u003c/p\u003e \u003cp\u003eIn the univariate analysis of the training group, significant statistical differences were observed between the LVI-positive and LVI-negative groups in terms of gender and CEA levels (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05). In semantic features, significant differences were noted between the two groups in TNM stage, N stage, Vascular Convergence Sign (VCS), and Vacuole Sign (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05). All clinical and semantical significant variables were included into a Generalized Linear Model (GLM) for analysis to construct the Clinical model. According to the results of the GLM analysis (Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e), gender, VCS, and N stage were selected for the Clinical model construction, in which N stage showing the most significant correlation with LVI status (Coef 0.480, 95% CI 0.266\u0026ndash;0.693, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e \u003cp\u003e \u003cb\u003eRadiomics and Heterogeneity Modeling\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThis study extracted 105 radiomic features from four regional ROIs (GTV, PTV_0\u0026ndash;3, PTV_-3-3, and PTV_0\u0026ndash;6). Initially, features with ICC values below 0.75 were excluded. Subsequently, a four-step feature selection process reduced the number of features for GTV (Table S2), PTV_0\u0026ndash;3, PTV_-3-3, and PTV_0\u0026ndash;6 (Table S3) to 3, 9, 11, and 10, respectively.\u003c/p\u003e \u003cp\u003eAmong the three PTV models, the PTV_0\u0026ndash;6 model exhibited AUC values of 0.882 in the training group and 0.824 in the validation group, significantly outperforming (DeLong test: p\u0026thinsp;\u0026lt;\u0026thinsp;0.05) the PTV_-3-3 model (AUCs of 0.771 and 0.696) and the PTV_0\u0026ndash;3 model (AUCs of 0.805 and 0.798). Additionally, the accuracy, sensitivity, and specificity of the three models were comprehensively evaluated. Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e (Figure S3) indicates that the performance of the PTV_0\u0026ndash;6 model was superior to the other two models.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePerformance of the constructed models evaluated by the ROC-related metrics.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModels\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAUC (95%CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAccuracy(95%CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eSensitivity(95%CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eSpecificity(95%CI)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" morerows=\"1\" nameend=\"c2\" namest=\"c1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cb\u003ePTR_-3-3\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003etrain\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.771 (0.695\u0026ndash;0.847)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.701 (0.695\u0026ndash;0.707)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.762 (0.749\u0026ndash;0.777)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.663 (0.653\u0026ndash;0.673)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003etest\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.696 (0.610\u0026ndash;0.783)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.682 (0.676\u0026ndash;0.688)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.621 (0.604\u0026ndash;0.637)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.719 (0.71\u0026ndash;0.728)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" morerows=\"1\" nameend=\"c2\" namest=\"c1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cb\u003ePTR_0\u0026ndash;3\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003etrain\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.805 (0.735\u0026ndash;0.874)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.727 (0.622\u0026ndash;0.733)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.831 (0.818\u0026ndash;0.843)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.663 (0.653\u0026ndash;0.673)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003etest\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.798 (0.730\u0026ndash;0.866)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.721 (0.715\u0026ndash;0.726)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.879 (0.868\u0026ndash;0.890)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.625 (0.615\u0026ndash;0.635)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" morerows=\"1\" nameend=\"c2\" namest=\"c1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cb\u003ePTR_0\u0026ndash;6\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003etrain\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.882 (0.830\u0026ndash;0.934)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.812 (0.807\u0026ndash;0.817)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.881 (0.871\u0026ndash;0.892)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.768 (0.76\u0026ndash;0.777)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003etest\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.824 (0.760\u0026ndash;0.888)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.701 (0.695\u0026ndash;0.707)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.845 (0.833\u0026ndash;0.857)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.615 (0.605\u0026ndash;0.625)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eConsequently, intratumoral ITH models and peritumoral PTH models were constructed from the GTV and PTV_0\u0026ndash;6 ROIs, respectively. Initially, tumor subregions were identified within the GTV and PTV_0\u0026ndash;6 ROIs using a simple linear interactive clustering method. Radiomic features were then extracted, and unsupervised clustering was performed on similar feature-bearing subregions using a Gaussian mixture model. Dimensionality reduction was subsequently carried out using the minimum redundancy and maximum relevance method, and ultimately, five imaging features were selected to represent the spatial heterogeneity of the GTV (Table S4) and PTV_0\u0026ndash;6 (Table S5), and finally constructed the ITH and PTH_0\u0026ndash;6 models, respectively.\u003c/p\u003e \u003cp\u003eFinally, the traditional radiomic models of PTR_0\u0026ndash;6, along with the ITH and PTH_0\u0026ndash;6 models, were combined as independent risk factors with selected clinical semantic imaging risk factors (gender, vascular convergence sign, N staging) to construct a comprehensive nomogram diagnostic model (Table S6, Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). In this model, each \"independent risk factor\" in a patient\u0026rsquo;s lesion is converted into a score based on the model's \"Points\" column. The sum of the scores from the five risk factors for the lesion is in the \"Total Points\u0026rdquo; row, and the corresponding probability in the \"Risk of metastasis\" column represents the probability of LVI for that lesion.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eModels Validation and Comparison\u003c/b\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e displays the ROC curves for the constructed intratumoral and peritumoral traditional radiomics models, heterogeneity models, clinical models, and combined models, while Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e (Figure S4) lists more comprehensive ROC-related metrics (AUC, accuracy, sensitivity, and specificity) for these models. Figure S5-S10 showed the calibration curves with H-L test results for all six constructed models. The analysis of these figures and tables reveals:\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePerformance of the constructed models evaluated by the ROC-related metrics.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModels\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAUC (95%CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAccuracy(95%CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eSensitivity(95%CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eSpecificity(95%CI)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" morerows=\"1\" nameend=\"c2\" namest=\"c1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cb\u003ePTR_0\u0026ndash;6\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003etrain\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.882 (0.830\u0026ndash;0.934)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.812 (0.807\u0026ndash;0.817)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.881 (0.871\u0026ndash;0.892)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.768 (0.76\u0026ndash;0.777)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003etest\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.824 (0.760\u0026ndash;0.888)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.701 (0.695\u0026ndash;0.707)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.845 (0.833\u0026ndash;0.857)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.615 (0.605\u0026ndash;0.625)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" morerows=\"1\" nameend=\"c2\" namest=\"c1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cb\u003ePTH_0\u0026ndash;6\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003etrain\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.725 (0.643\u0026ndash;0.805)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.675 (0.669\u0026ndash;0.681)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.610 (0.594\u0026ndash;0.626)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.716 (0.706\u0026ndash;0.725)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003etest\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.692 (0.606\u0026ndash;0.778)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.656 (0.65\u0026ndash;0.662)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.534 (0.518\u0026ndash;0.551)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.729 (0.72\u0026ndash;0.738)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" morerows=\"1\" nameend=\"c2\" namest=\"c1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cb\u003eITR\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003etrain\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.673 (0.589\u0026ndash;0.757)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.604 (0.598\u0026ndash;0.61)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.593 (0.577\u0026ndash;0.61)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.611 (0.6-0.621)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003etest\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.665 (0.577\u0026ndash;0.753)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.604 (0.598\u0026ndash;0.61)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.5 (0.483\u0026ndash;0.517)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.667 (0.657\u0026ndash;0.676)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" morerows=\"1\" nameend=\"c2\" namest=\"c1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cb\u003eITH\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003etrain\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.722 (0.642\u0026ndash;0.802)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.662 (0.656\u0026ndash;0.668)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.695 (0.68\u0026ndash;0.71)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.642 (0.632\u0026ndash;0.652)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003etest\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.607 (0.517\u0026ndash;0.697)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.578 (0.572\u0026ndash;0.584)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.448 (0.431\u0026ndash;0.465)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.656 (0.647\u0026ndash;0.666)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" morerows=\"1\" nameend=\"c2\" namest=\"c1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cb\u003eClinical\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003etrain\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.767 (0.688\u0026ndash;0.846)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.766 (0.761\u0026ndash;0.772)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.610 (0.594\u0026ndash;0.626)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.863 (0.856\u0026ndash;0.87)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003etest\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.680 (0.595\u0026ndash;0.765)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.669 (0.663\u0026ndash;0.675)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.483 (0.466-0.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.781 (0.773\u0026ndash;0.79)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" morerows=\"1\" nameend=\"c2\" namest=\"c1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cb\u003eCombined\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003etrain\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.962 (0.937\u0026ndash;0.988)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.903 (0.899\u0026ndash;0.906)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.898 (0.888\u0026ndash;0.908)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.905 (0.899\u0026ndash;0.911)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003etest\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.882 (0.831\u0026ndash;0.933)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.779 (0.774\u0026ndash;0.784)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.862 (0.85\u0026ndash;0.874)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.729 (0.72\u0026ndash;0.738)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e(1) Across both the training and testing sets, the combined model exhibits the highest predictive performance among all models, significantly outperforming the other models (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05).\u003c/p\u003e \u003cp\u003e(2) Whether traditional radiomics or heterogeneity models, the models based on peritumoral constructions consistently outperform those based on intratumoral constructions.\u003c/p\u003e \u003cp\u003e(3) For both intratumoral and peritumoral ROIs, the predictive performance of the heterogeneity models alone is inferior to that of the traditional radiomics models.\u003c/p\u003e \u003cp\u003e \u003cb\u003eClinical use, Benefit and Explanation\u003c/b\u003e \u003c/p\u003e \u003cp\u003eTo illustrate the clinical use of the constructed nomogram, we listed two examples in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e. As shown in the figure, we found that we could easily obtain the risk probability of LVI for the NSCLC patient by combining the clinical risk factors and radiomics and heterogeneity models constructed based on their CT images. Case 1 achieved a very high-risk score and proven to be LVI-positive pathologically, whereas case 2 was categorized to be LVI-negative (low risk score).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e displays the Decision Curve Analysis (DCA) curves for the constructed intratumoral and peritumoral traditional radiomics models, heterogeneity models, clinical models, and combined models. The observations from the figure are as follows:\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e(1) Across most risk threshold ranges, using these six models yields higher net benefits compared to the \"treat all\" and \"treat none\" strategies.\u003c/p\u003e \u003cp\u003e(2) Except in the risk threshold range of approximately 0.83\u0026ndash;0.9, the combined model achieves the highest net benefit in other ranges.\u003c/p\u003e \u003cp\u003e(3) Across most risk threshold ranges, the clinical benefits of the models align with the conclusions drawn from model comparisons in the previous subsection: the combined model shows the highest net benefit, the peritumoral model achieves higher net benefits than the intratumoral model, and the traditional radiomics model provides higher net benefits than the heterogeneity model.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study, a model combined clinical data, radiologic features, radiolomic features and tumor heterogeneity\u0026nbsp;features was established to diagnose LVI status for early stage NSCLC. ITR, ITH and PTR,\u0026nbsp;PTH models were established by analyzing the subregions of tumor and peritumor on CT images. The final model included VCS, N stage, ITH model, PTR_0 + 6 model and PTH_0 + 6 model, which was used to quantify the probability of LVI in patients. This model has high diagnostic accuracy, with an AUC of 0.962(95% CI: 0.937-0.988) in the training group and 0.882(95% CI: 0.831-0.933) in the validation group.\u003c/p\u003e\n\u003cp\u003eSeveral previous studies have constructed models to diagnose LVI status of NSCLC by either imaging features or combining imaging features with clinical features. Chen et al\u003csup\u003e[35]\u003c/sup\u003e developed an LVI predictive model combined independent predictors(smoking and clinical stage) and the GPTV9 radiomic score, with an AUC of 0.89,0.83, and 0.66 in the training, internal validation, and external validation groups, respectively. Zhang et al\u003csup\u003e[36]\u0026nbsp;\u003c/sup\u003edeveloped a LVI predictive model based on 2D and 3D tumor and peritumoral features, with an average AUC of 0.759. The AUC of the synthetic model established in this study is respectively 0.962 and 0.882 in the training and\u0026nbsp;validation\u0026nbsp;corhort, which is better than the previous studies. The reason may be that we not only integrated clinical data, imaging features, ITR and PTR features, but also added ITH and PTH features.\u003c/p\u003e\n\u003cp\u003eNSCLC presents a large intratumor heterogeneity\u003csup\u003e[37]\u003c/sup\u003e. A good ITH model requires both imaging features and their spatial distribution. However, current research typically only captures a portion of the information.\u0026nbsp;The definition of computational features in traditional\u0026nbsp;radiomics involves the assumption of uniform distribution heterogeneity, without quantifying the local features of tumors \u003csup\u003e[38]\u003c/sup\u003e. While the ITH model uses the intensity of different tumors to group pixels and identify similar subregions. And associate the statistical features of each sub region (such as its LVI state), while integrating local radiometric features and global pixel distribution patterns. The model established by Shi et al\u003csup\u003e[39]\u003c/sup\u003e integrates the clinicopathological information, ITH characteristics and c-radiation group characteristics, and has a good prediction effect on the complete remission of neoadjuvant chemotherapy pathology in breast cancer patients (AUC values were 0.83-0.87 in the test set) . However, The AUC of the combined model of clinical imaging and traditional histology was only 0.78-0.81 in the test set, and 0.74-0.76 in ITH model alone. Thus, the addition of ITH features to the pathologic complete response (pCR) prediction model of breast cancer may increase clinical utility\u003c/p\u003e\n\u003cp\u003eThe ITH model in this study showed good discrimination ability for the state of LVI. The AUC of validation groups in GTV-ITH and PTH_0-6model were 0.607 and 0.692, respectively. The AUC in GTV and\u0026nbsp;PTR_0-6 models were 0.665 and 0.824, respectively. When we add the heterogeneity model to the traditional radiomic model, the performance of the combinatorial model is greatly improved, and the AUC value reaches 0.882. It may be that ITH models use multi-region image features to characterize intratumoral spatial heterogeneity, this could further refine the shortcomings of traditional radiomic models, adding abundant information on tumor heterogeneity within the ROI.\u003c/p\u003e\n\u003cp\u003eTumor cells are often highly aggressive, disrupting the normal structure of surrounding parenchymal tissue, leading to carcinogenic infiltration of small blood vessels and lymphatic vessels around the lesion, which is often overlooked in studies that focus on intratumoral areas\u003csup\u003e[24,40]\u003c/sup\u003e. Therefore, detecting the boundary of lung cancer may help to quantify the invasiveness of the tumor. The study of NSCLC prediction by peritumoral imaging features has become an active and important field. However, the use of peritumoral imaging features in predicting the invasiveness of NSCLC has only been studied in a few studies. Previous studies defined the peritumoral area as 1.5 to 20 mm\u003csup\u003e[24,41-43]\u003c/sup\u003e. One study quantified the distance to micrometastases in histopathology lung cancer, resulting in mean distances of 2.94 mm and 2.69 mm for adenocarcinoma and squamous cell carcinoma of the lung to surrounding micrometastases, respectively\u003csup\u003e[44]\u003c/sup\u003e. Based on this, we defined the peritumoral extent on a 3 mm gradient to explore the relationship between vascular invasion status in NSCLC and peritumoral imaging. The results showed that Compared with PTR_0 - 3 and PTR_-3 - 3 , the AUC of PTR_0 - 6 has the highest diagnostic value (0.696,0.798 and 0.824). The reason may be that the farther away from the tumor, the higher the reproducibility of imaging features. This finding may be related to the presence of homogeneous lung parenchyma in the distal peri-tumour\u003csup\u003e[42]\u003c/sup\u003e. Therefore, PTR_0 - 6 model performs better than other models in our study. In addition, the performance of the peritumoral model is generally better than that of the intratumoral model in this study, whether it is the traditional radiomic model or the\u0026nbsp;heterogeneity\u0026nbsp;model. Consistent with the findings of\u0026nbsp;Dou TH, et al.\u003csup\u003e[45]\u003c/sup\u003e, their model for predicting distant metastasis in NSCLC patients has a higher prognostic value for tumor marginal radiological features than for tumor radiological features alone; The comparison between the two was statistically significant (p = 0.048) .This may be due to the presence of increased cancer invasion and metastatic activity around the tumour, such as epithelial-mesenchymal Metastasis\u003csup\u003e[46]\u003c/sup\u003e, Tumor-associated macrophage\u003csup\u003e[47,48]\u003c/sup\u003e, tumour budding\u003csup\u003e[49]\u003c/sup\u003e and lymphatic vascular invasion\u003csup\u003e[50,51]\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eGLM model analysis found that vessel convergence and N stage were independent predictors of LVI, and gender was a marginal correlation factor of LVI. This study established a clinical-imaging model based on this. Of these, N stage was most strongly associated with LVI status in NSCLC patients, as LVI is an initial manifestation of nodal metastasis\u003csup\u003e[52]\u003c/sup\u003e. The vessel convergence sign is the manifestation of vascular structure being pulled by the focus to concentrate in the direction of the focus or truncated through the focus or at the edge of the focus. This is mainly due to the fact that the Vascular endothelial growth factor secreted by malignant nodules through cancer cells promote the growth of microvessels in tumor tissues and the destruction of some blood vessel walls, which is consistent with the results of this study. A study\u003csup\u003e[53]\u003c/sup\u003e has found that the density of memory B cells, which play an important role in human anti-tumor immunity, is higher in lung adenocarcinoma tissues of female patients. Additional studies have suggested that lung cancer in women may have a different genetic profile from that in men because of different natural histories as well as female characteristics (younger age at diagnosis, non-smokers, more likely to have adenocarcinoma than men)\u003csup\u003e[54,55]\u003c/sup\u003e. This may explain some of the gender differences observed in this study regarding tumor invasion of vessels.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eHowever, this study has some limitations. Firstly, in this study, semi-automatic method was used to segment ROI, which may lead to artificial differences. An accurate automatic segmentation method should be considered in future research. Secondly, only perform radiomics feature extraction on CT plain scan images. CT enhancement or PET images may contain additional valuable information. Thirdly, this study only used an internal validation queue, and due to the limited sample size of the public database, external validation cannot be conducted. More large-scale studies are needed in the future to further validate.\u003c/p\u003e\n\u003cp\u003eIn summary, we have constructed a total of 6 models, and within most risk threshold ranges, using these 6 models can achieve higher net benefits than the \u0026quot;all treatment\u0026quot; and \u0026quot;no treatment at all\u0026quot; strategies. Among them, the combination model \u0026nbsp;that integrates clinical data, traditional radiological features, peritumor radiomic features, peritumor heterogeneity features has the highest diagnostic accuracy for \u0026nbsp;the LVI status in early stage NSCLC patients. And DCA analysis shows that within most risk thresholds, combining multiple models improves the clinical value of the models. The comprehensive column chart model not only enables non-invasive preoperative risk assessment of lung lesions, but also helps to provide objective guidance for rational clinical decision-making.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eDate availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets generated during and/or analysed during the current study are not openly available for reasons of patient confidentiality however anonymised data may be available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026apos; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eConceptualization: Hanfei Zhang, Meiyan Liao\u003c/p\u003e\n\u003cp\u003eData curation: Yun Long, Hanfei Zhang, Yang Guo, Tian Gan, Jingting Wang, Ting Li, Kemeng Zhang, Jun Chen, Yang Li, Meiyan Liao, Feng Xiao\u003c/p\u003e\n\u003cp\u003eFormal analysis: Yun Long, Feng Xiao\u003c/p\u003e\n\u003cp\u003eInvestigation: Yun Long, Hanfei Zhang, Yang Guo, Tian Gan, Jingting Wang, Ting Li, Kemeng Zhang, Jun Chen, Yang Li, Meiyan Liao, Feng Xiao\u003c/p\u003e\n\u003cp\u003eMethodology: Yun Long, Hanfei Zhang, Meiyan Liao, Feng Xiao\u003c/p\u003e\n\u003cp\u003eProject administration: Meiyan Liao, Feng Xiao\u003c/p\u003e\n\u003cp\u003eResources: Meiyan Liao, Feng Xiao\u003c/p\u003e\n\u003cp\u003eSoftware: Jun Chen, Yang Li, Feng Xiao\u003c/p\u003e\n\u003cp\u003eSupervision: Meiyan Liao, Feng Xiao\u003c/p\u003e\n\u003cp\u003eValidation: Yun Long, Hanfei Zhang, Jun Chen, Yang Li, Meiyan Liao, Feng Xiao\u003c/p\u003e\n\u003cp\u003eVisualization: Yun Long, Hanfei Zhang, Jun Chen, Yang Li, Meiyan Liao, Feng Xiao\u003c/p\u003e\n\u003cp\u003eWriting-original draft: Yun Long, Hanfei Zhang\u003c/p\u003e\n\u003cp\u003eWriting-review \u0026amp; editing: Yun Long, Hanfei Zhang, Meiyan Liao, Feng Xiao\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eacknowledgement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors have no acknowledgements to declare.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNo funding was received for this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics declarations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEthics approval and consent to participate\u003c/p\u003e\n\u003cp\u003eThis study was approved by the Ethics Committee of Zhongnan Hospital of Wuhan University, with the ethical approval number 2021048 (IRB number).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIndividual informed consent was waived because this was a retrospective study.\u003c/p\u003e\n\u003cp\u003eAll methods were performed in accordance with the relevant guidelines and regulations.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eManuscript is approved by all authors for publication\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eConflict of interest The authors of this manuscript declare no relationships with any companies whose products or services may be related to the subject matter of the article.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eBray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024 May-Jun;74(3):229-263. \u003c/li\u003e\n\u003cli\u003eChen P, Liu Y, Wen Y, et al. Non-small cell lung cancer in China. Cancer Commun (Lond). 2022 Oct;42(10):937-970.\u003c/li\u003e\n\u003cli\u003eTravis WD, Brambilla E, Nicholson AG, et al. The 2015 World Health Organization Classification of Lung Tumors: Impact of Genetic, Clinical and Radiologic Advances Since the 2004 Classification. J Thorac Oncol. 2015 Sep;10(9):1243-1260. \u003c/li\u003e\n\u003cli\u003eG\u0026uuml;nther K., Leier J., Henning G., et al. Prediction of lymph node metastasis in colorectal carcinoma by expressionof chemokine receptor CCR7[J]. Int. J. Cancer, 2005, 116:726\u0026ndash;733. \u003c/li\u003e\n\u003cli\u003eKodama J., Kusumoto T., Seki N., et al. Association of CXCR4 and CCR7 chemokine receptor expression and lymph node metastasis in human cervical cancer[J]. Ann. Oncol, 2007, 18:70\u0026ndash;76. \u003c/li\u003e\n\u003cli\u003eAlmofti A., Uchida D., Begum N.M., et al. The clinicopathological significance of the expression of CXCR4 protein in oral squamous cell carcinoma[J]. Int. J. Oncol, 2004, 25:65\u0026ndash;71. \u003c/li\u003e\n\u003cli\u003eSchimanski C.C., Bahre R., Gockel I., et al. Dissemination of hepatocellular carcinoma is mediated via chemokine receptor CXCR4[J]. Br. J. Cancer, 2006, 95:210\u0026ndash;217. \u003c/li\u003e\n\u003cli\u003eDieterich L.C., Kapaklikaya K., Cetintas T., et al. Transcriptional profiling of breast cancer-associated lymphatic vessels reveals VCAM-1 as regulator of lymphatic invasion and permeability[J]. Int. J. Cancer, 2019, 145:2804\u0026ndash;2815. \u003c/li\u003e\n\u003cli\u003eGarmy-Susini B., Avraamides C.J., Schmid M.C., et al. Integrin \u0026alpha;4\u0026beta;1 Signaling Is Required for Lymphangiogenesis and Tumor Metastasis[J]. Cancer Res, 2010, 70:3042\u0026ndash;3051. \u003c/li\u003e\n\u003cli\u003eVestweber D. How leukocytes cross the vascular endothelium[J]. Nat. Rev. Immunol, 2015, 15:692\u0026ndash;704. \u003c/li\u003e\n\u003cli\u003eTacconi C., Correale C., Gandelli A., et al. Vascular Endothelial Growth Factor C Disrupts the Endothelial Lymphatic Barrier to Promote Colorectal Cancer Invasion[J]. Gastroenterology, 2015, 148:1438\u0026ndash;1451. \u003c/li\u003e\n\u003cli\u003eNakayama H., Ohuchida K., Yonenaga A., et al. S100P regulates the collective invasion of pancreatic cancer cells into the lymphatic endothelial monolayer[J]. Int. J. Oncol, 2019, 55:211\u0026ndash;222.\u003c/li\u003e\n\u003cli\u003eMohammed R.A., Martin S.G., Mahmmod A.M., et al. Objective assessment of lymphatic and blood vascular invasion in lymph node-negative breast carcinoma: Findings from a large case series with long-term follow-up[J]. J. Pathol, 2011, 223:358\u0026ndash;365. \u003c/li\u003e\n\u003cli\u003eHanahan D, Weinberg RA. Hallmarks of cancer: the next generation[J]. Cell, 2011, 144: 646\u0026ndash;674.\u003c/li\u003e\n\u003cli\u003eBrown M., Assen F.P., Leithner A., et al. Lymph node blood vessels provide exit routes for metastatic tumor cell dissemination in mice[J]. Science, 2018, 359:1408\u0026ndash;1411. \u003c/li\u003e\n\u003cli\u003eGabor S, Renner H, Popper H, et al. Invasion of blood vessels as significant prognostic factor in radically resected T1-3N0M0 non-small cell lung cancer[J]. Eur J Cardiothorac Surg, 2004, 25:439\u0026ndash;42.\u003c/li\u003e\n\u003cli\u003eHishida T, Yoshida J, Maeda R, et al. Prognostic impact of intratumoural microvascular invasion and microlymphatic permeation on node-negative non-small cell lung cancer: which indicator is the stronger prognostic factor? [J]. Eur J Cardiothorac Surg, 2013, 43:772\u0026ndash;7.\u003c/li\u003e\n\u003cli\u003eCollier FC, Blakemore WS, Kyle R H, et al. Carcinoma of the lung: factors which influence five year survival with special reference to blood vessel invasion[J]. Ann Surg, 1957, 146(3): 417-423.\u003c/li\u003e\n\u003cli\u003eNeri S, Yoshida J, Ishii G, et al. Prognostic impact of microscopic vessel invasion and visceral pleural invasion in non-small cell lung cancer: a retrospective analysis of 2,657 patients[J]. Ann Surg, 2014, 260(2): 383-388.\u003c/li\u003e\n\u003cli\u003eCoroller TP, grossmann P, hou Y, et al. CT-based radiomic signature predicts distant metastasis in lung adenocarcinoma[J]. Radiother Oncol, 2015, 114:345-50. \u003c/li\u003e\n\u003cli\u003eWang X, Wan Q, Chen H, et al. Classification of pulmonary lesion based on multiparametric MRI: utility of radiomics and comparison of machine learning methods[J]. EurRadiol, 2020, 30(8):4595\u0026ndash;4605.\u003c/li\u003e\n\u003cli\u003eDercle L, Fronheiser M, Lu L, et al. Identification of nonsmall cell lung cancer sensitive to systemic cancer therapies using radiomics[J]. Clin Cancer Res, 2020, 26(9):2151\u0026ndash;2162.\u003c/li\u003e\n\u003cli\u003eKadota K, Nitadori J-i, Sima CS, et al. Tumor Spread through Air Spaces is an Important Pattern of Invasion and Impacts the Frequency and Location of Recurrences after Limited Resection for Small Stage I Lung Adenocarcinomas[J]. Journal of Thoracic Oncology, 2015, 10: 806\u0026ndash;14.\u003c/li\u003e\n\u003cli\u003eMatsumura Y, Ishii G, Aokage K, et al. Morphophenotypic characteristics of intralymphatic cancer and stromal cells susceptible to lymphogenic metastasis[J]. Cancer Science, 2012, 103: 1342\u0026ndash;7. \u003c/li\u003e\n\u003cli\u003eOnozato ML, Kovach AE, Yeap BY, et al. Tumor islands in resected early-stage lung adenocarcinomas are associated with unique clinicopathologic and molecular characteristics and worse prognosis[J]. Am J Surg Pathol, 2013, 37: 287\u0026ndash;94. \u003c/li\u003e\n\u003cli\u003eShimada Y, Ishii G, Hishida T, et al. Extratumoral vascular invasion is a significant prognostic indicator and a predicting factor of distant metastasis in non-small cell lung cancer[J]. J Thorac Oncol, 2010; 5: 970\u0026ndash;5. \u003c/li\u003e\n\u003cli\u003eUruga H, Fujii T, Fujimori S,, et al. Semiquantitative Assessment of Tumor Spread through Air Spaces (STAS) in Early-Stage Lung Adenocarcinomas[J]. J Thorac Oncol, 2017, 12: 1046\u0026ndash;51. \u003c/li\u003e\n\u003cli\u003eSaijo T, Ishii G, Ochiai A, et al. Evaluation of extratumoral lymphatic permeation in non-small cell lung cancer as a means of predicting outcome[J]. Lung Cancer, 2007, 55:61\u0026ndash;6. \u003c/li\u003e\n\u003cli\u003eKoukourakis MI, Giatromanolaki A, Thorpe PE, et al. Vascular endothelial growth factor/KDR activated microvessel density versus CD31 standard microvessel density in non-small cell lung cancer[J]. Cancer Res, 2000, 60: 3088\u0026ndash;95.\u003c/li\u003e\n\u003cli\u003eTsutani Y,Suzuki K,Koike T, et al.High-risk factors for recurrence of stage I lung adenocarcinoma: follow-up data from JCOG0201[J].Ann Thorac Surg, 2019, 108(5):1484-1490.\u003c/li\u003e\n\u003cli\u003eNwogu I, Corso JJ. Exploratory identification of image-based biomarkers for solid mass pulmonary tumors[J]. Med Image Comput Comput Assist Interv, 2008, 11(Pt 1):612-9. \u003c/li\u003e\n\u003cli\u003eLambin P, Rios-Velazquez E, Leijenaar R, et al. Radiomics: extracting more information from medical images using advanced feature analysis[J]. Eur J Cancer, 2012, 48(4):441-6.\u003c/li\u003e\n\u003cli\u003eZwanenburg A, Valli\u0026egrave;res M, Abdalah MA, et al. The Image Biomarker Standardization Initiative: Standardized Quantitative Radiomics for High-Throughput Image-based Phenotyping,et al. Radiology, 2020, 295(2):328-338.\u003c/li\u003e\n\u003cli\u003eGillies RJ, Kinahan PE, Hricak H. Radiomics: images are more than pictures, they are data[J]. Radiology, 2016, 278:563\u0026ndash;577. \u003c/li\u003e\n\u003cli\u003eChen Q, Shao J, Xue T, et al. Intratumoral and peritumoral radiomics nomograms for the preoperative prediction of lymphovascular invasion and overall survival in non-small cell lung cancer. Eur Radiol. 2023 Feb;33(2):947-958. \u003c/li\u003e\n\u003cli\u003eZhang X, Zhang G, Qiu X, et al. Radiomics under 2D regions, 3D regions, and peritumoral regions reveal tumor heterogeneity in non-small cell lung cancer: a multicenter study. Radiol Med. 2023 Sep;128(9):1079-1092.\u003c/li\u003e\n\u003cli\u003eZhang J, Fujimoto J, Zhang J, et al. Intratumor heterogeneity in localized lung adenocarcinomas delineated by multiregion sequencing. Science. 2014 Oct 10;346(6206):256-9.\u003c/li\u003e\n\u003cli\u003eSubramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles[J]. Proc Natl Acad Sci U S A, 2005, 102(43):15545-50.\u003c/li\u003e\n\u003cli\u003eShi Z, Huang X, Cheng Z, et al. MRI-based Quantification of Intratumoral Heterogeneity for Predicting Treatment Response to Neoadjuvant Chemotherapy in Breast Cancer. Radiology. 2023 Jul;308(1):e222830.\u003c/li\u003e\n\u003cli\u003eWang XL, Fang JP, Tang RY, et al. Different significance between intratumoral and peritumoral lymphatic vessel density in gastric cancer: a retrospective study of 123 cases[J]. BMC Cancer, 2010, 17;10:299.\u003c/li\u003e\n\u003cli\u003ePadera TP, Kadambi A, di Tomaso E, et al. Lymphatic metastasis in the absence of functional intratumor lymphatics[J]. Science, 2002, 296(5574):1883-6.\u003c/li\u003e\n\u003cli\u003eSaijo T, Ishii G, Ochiai A, et al. Evaluation of extratumoral lymphatic permeation in non-small cell lung cancer as a means of predicting outcome[J]. Lung Cancer, 2007, 55(1):61-6.\u003c/li\u003e\n\u003cli\u003eShimada Y, Ishii G, Hishida T, et al. Extratumoral vascular invasion is a significant prognostic indicator and a predicting factor of distant metastasis in non-small cell lung cancer[J]. J Thorac Oncol, 2010, 5(7):970-5.\u003c/li\u003e\n\u003cli\u003eMoon Y, Sung SW, Lee KY, et al. Clinicopathological characteristics and prognosis of non-lepidic invasive adenocarcinoma presenting as ground glass opacity nodule[J]. J Thorac Dis, 2016, 8(9):2562-2570.\u003c/li\u003e\n\u003cli\u003eDou TH, Coroller TP, van Griethuysen JJM, et al. Peritumoral radiomics features predict distant metastasis in locally advanced NSCLC. PLoS One. 2018 Nov 2;13(11):e0206108.\u003c/li\u003e\n\u003cli\u003eMahmood MQ, Ward C, Muller HK, et al. Epithelial mesenchymal transition (EMT) and non-small cell lung cancer (NSCLC): a mutual association with airway disease. Med Oncol. 2017 Mar;34(3):45.\u003c/li\u003e\n\u003cli\u003eJoyce JA, Pollard JW. Microenvironmental regulation of metastasis. Nat Rev Cancer. 2009 Apr;9(4):239-52.\u003c/li\u003e\n\u003cli\u003eQian BZ, Pollard JW. Macrophage diversity enhances tumor progression and metastasis. Cell. 2010 Apr 2;141(1):39-51.\u003c/li\u003e\n\u003cli\u003eYamaguchi Y, Ishii G, Kojima M, et al. Histopathologic features of the tumor budding in adenocarcinoma of the lung: tumor budding as an index to predict the potential aggressiveness. J Thorac Oncol. 2010 Sep;5(9):1361-8..\u003c/li\u003e\n\u003cli\u003eShimada Y, Ishii G, Hishida T, et al. Extratumoral vascular invasion is a significant prognostic indicator and a predicting factor of distant metastasis in non-small cell lung cancer. J Thorac Oncol. 2010 Jul;5(7):970-5. \u003c/li\u003e\n\u003cli\u003eTammela T, Alitalo K. Lymphangiogenesis: Molecular mechanisms and future promise. Cell. 2010 Feb 19;140(4):460-76.\u003c/li\u003e\n\u003cli\u003eLee H, Choi H. Investigating the Clinico-Molecular and Immunological Evolution of Lung Adenocarcinoma Using Pseudotime Analysis[J].. Front Oncol, 2022, 12:828505.\u003c/li\u003e\n\u003cli\u003eWu DM, Deng SH, Zhou J, et al. PLEK2 mediates metastasis and vascular invasion via the ubiquitin-dependent degradation of SHIP2 in non-small cell lung cancer[J]. Int J Cancer , 2020, 146(9):2563-2575.\u003c/li\u003e\n\u003cli\u003eAlwithenani A, Bethune D, Castonguay M, et al. Profiling non-small cell lung cancer reveals that PD-L1 is associated with wild type EGFR and vascular invasion, and immunohistochemistry quantification of PD-L1 correlates weakly with RT-qPCR[J]. PLoS One, 2021, 16(5):e0251080.\u003c/li\u003e\n\u003cli\u003eRamnefjell M, Aamelfot C, Helgeland L, et al. Vascular invasion is an adverse prognostic factor in resected nonsmall-cell lung cancer[J]. APMIS, 2017, 125(3):197\u0026ndash;206.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-imaging","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bmim","sideBox":"Learn more about [BMC Medical Imaging](http://bmcmedimaging.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bmim/default.aspx","title":"BMC Medical Imaging","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Non small cell lung cancer, CT imaging, Lymphovascular invasion, radiomics, Tumor heterogeneity","lastPublishedDoi":"10.21203/rs.3.rs-6764701/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6764701/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eObjective: \u003c/strong\u003eTo establish a model integrates clinical, traditional radiologic, intratumoral and peritumoral radiomics (ITR and PTR), and intratumoral and peritumoral heterogeneity (ITH and PTH) features to diagnose lymphovascular invasion (LVI) status for early stage non small cell lung cancer (NSCLC).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMaterials and Methods:\u003c/strong\u003e Clinical data and chest CT imaging data of NSCLC patients who underwent surgical resection of the lungs from January 2019 to May 2021 were collected. Surgical pathology were the diagnostic gold standard to clarify the LVI status. ITR and PTR features and ITH and PTH features from the total tumor volume and peritumoral tumor volume were extracted. Then clinical, traditional radiologic, ITR and PTR, ITH and PTH models were established to diagnose LVI status. Finally, a column chart diagnostic model was constructed and the diagnostic efficacy was evaluated.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults:\u003c/strong\u003e 308 NSCLC cases met the inclusion criteria, including 117 cases in the LVI positive group and 191 cases in the LVI negative group. They were randomly divided into training group and validation group in a 1:1 ratio. In the three cohorts of PTR_(0 - 3, -3 - 3 and 0 - 6), the PTR_0 - 6 model has better predictive performance, with area under the curve (AUC) of 0.882 and 0.824 for the training and validation groups, respectively. Gender, Vascular Convergence Sign, and N stagewere significantly related to LVI status, Finally, the combined model integrated ITH, PTR_0-6, and PTH_0-6 models, N stage and Vascular Convergence Sign has the highest diagnostic accuracy. The AUC in the training group is 0.962 and in the validation group is 0.882.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions:\u003c/strong\u003e A comprehensive diagnostic model based on clinical features, traditional radiological features, radiomic features, and heterogeneity features of NSCLC were established to diagnose LVI for early stage NSCLC, which has the highest diagnostic efficiency and can help to guide treatment decisions.\u003c/p\u003e","manuscriptTitle":"CT based Quantification of Intratumoral and Peritumoral Heterogeneity for diagnosing Lymphovascular Invasion for Early Stage Non-small cell lung cancer","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-06-25 02:17:11","doi":"10.21203/rs.3.rs-6764701/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-08-18T09:46:12+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-14T02:38:38+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-06-28T08:28:32+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"27276535952825115692724343680212886962","date":"2025-06-21T07:41:15+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"187186905017940038827589232152295938109","date":"2025-06-21T06:10:03+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-06-19T06:50:01+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-06-18T17:30:37+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-06-18T16:59:39+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Medical Imaging","date":"2025-06-18T16:56:06+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-medical-imaging","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bmim","sideBox":"Learn more about [BMC Medical Imaging](http://bmcmedimaging.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bmim/default.aspx","title":"BMC Medical Imaging","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"708d7676-158b-4258-82d3-9ffc1864414f","owner":[],"postedDate":"June 25th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-12-22T16:11:39+00:00","versionOfRecord":{"articleIdentity":"rs-6764701","link":"https://doi.org/10.1186/s12880-025-02041-0","journal":{"identity":"bmc-medical-imaging","isVorOnly":false,"title":"BMC Medical Imaging"},"publishedOn":"2025-12-19 15:57:43","publishedOnDateReadable":"December 19th, 2025"},"versionCreatedAt":"2025-06-25 02:17:11","video":"","vorDoi":"10.1186/s12880-025-02041-0","vorDoiUrl":"https://doi.org/10.1186/s12880-025-02041-0","workflowStages":[]},"version":"v1","identity":"rs-6764701","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6764701","identity":"rs-6764701","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}