Impact of Multi-Parameter Images Obtained from Dual-Energy CT on Radiomicis to Predict Pathological Grading of Bladder Urothelial Carcinoma

Impact of Multi-Parameter Images Obtained from Dual-Energy CT on Radiomicis to Predict Pathological Grading of Bladder Urothelial Carcinoma | 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 Impact of Multi-Parameter Images Obtained from Dual-Energy CT on Radiomicis to Predict Pathological Grading of Bladder Urothelial Carcinoma Wei Wei, Shigeng Wang, Mengting Hu, Xiaoyu Tong, Yong Fan, Jingyi Zhang, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4722594/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 12 Aug, 2024 Read the published version in Abdominal Radiology → Version 1 posted 9 You are reading this latest preprint version Abstract Objective : To investigate the effect of radiomics models obtained from dual-energy CT (DECT) material decomposition images and virtual monoenergetic images (VMIs) in predicting the pathological grading of bladder urothelial carcinoma (BUC). Materials and Methods : Preoperative Energy-Spectrum CT images were retrospectively collected from 112 pathologically confirmed cases of BUC patients, including 76 cases of high-grade urothelial carcinoma and 36 cases of low-grade urothelial carcinoma. Enhanced CT venous phase images of all patients were reconstructed at 40 to 140 keV VMIs (interval 10 keV), Iodine maps, and Water maps, and a total of 13 sets of images were obtained, and imaging features were extracted in each of the 13 sets of images. The best features related to BUC were identified by recursive feature elimination (RFE), the Minimum Redundancy Maximum Relevance (mRMR), and the Least Absolute Shrinkage and Selection Operator (LASSO) in order. A five-fold cross-validation method was used to divide the samples into training and testing sets, and models for pathological prediction of BUC grading were constructed by a random forest (RF) classifier. ROC curves were plotted to evaluate the performance of 13 models obtained from reconstructed images. Results: There were no statistically significant differences in the area under the curve (AUC) between the training set and the testing set for all 13 models, with the AUC ranging from 0.91 to 0.96 in the training set and 0.84 to 0.90 in the testing set for each group of reconstructed images. Although the features selected for the reconstructed images were very different among the groups, all the features selected from 40 to 100 keV VMIs had dependencevariance of the GLDM feature set. Conclusion: The variation of spectral CT parameters did no effect on the radiomics-based prediction of the pathological grading of BUC and did not affect the accuracy of the model even if the relevant features differed between reconstructed images. Bladder cancer Radiomics Multi -parameter images Dual-energy CT Figures Figure 1 Figure 2 Figure 3 Introduction Bladder cancer (BCa) is currently one of the most common malignant tumors of the urinary system[ 1 – 3 ]. Among these, bladder urothelial carcinoma (BUC) accounts for more than 90% of BCa and can be classified into high-grade BUC (HBUC) and low-grade BUC (LBUC) based on pathological grading. HBUC is more invasive, with a high risk of recurrence, metastasis, and poor prognosis[ 4 , 5 ]. Moreover, there are significant differences in the treatment methods for HBUC and LBUC[ 6 – 9 ]. After the initial transurethral resection of the bladder tumor (TURBT), if HBUC is found, a repeat TURBT or bladder removal surgery is required to complete the lesion removal[ 10 ]. Thus, effective and accurate preoperative classification of BUC histopathological grading is crucial for improving early diagnosis and patient survival. Currently, pathological examination remains the gold standard for BUC pathological grading[ 11 , 12 ]. However, this invasive examination may lead to inaccurate classification due to inadequate samples, sometimes requiring repetition[ 13 – 15 ]. Therefore, exploring accurate non-invasive preoperative prediction methods for pathological grading may contribute to the personalized management of BUC. Dual-energy computed tomography (DECT) can obtain multi-parameter images from two imaging datasets with different energy spectra, such as virtual monoenergetic images (VMIs) and material decomposition images[ 16 , 17 ]. While previous studies have extracted additional radiomics quantitative parameters to predict BUC pathological grades in multi-parameter images, the inherent complexity of multi-parameter images has often been overlooked[ 5 , 18 ]. VMIs encompass 101 discrete image sets, with energy levels spanning from 40 to 140 kilo-electron volts (keV); low-energy VMIs enhance BUC visualization whereas high-energy VMIs mitigate image artifacts[ 19 ]. Material decomposition images elucidate iodine maps, depicting BUC microvasculature, and water maps, highlighting peritumoral edema linked to tumor infiltration[ 20 , 21 ]. We hypothesized that the best multi-parameter images to predict BUC pathological grades may exist when using radiomics. Thus, the purpose of this study is to determine the diagnostic accuracy of radiomics derived from various VMIs and material decomposition images from DECT in predicting the pathologic grading of BUC and to identify models that show superior diagnostic performance. Materials and methods Study population This retrospective study was approved by the Institutional Ethics Committee, and the consent from patients was waived. Patients diagnosed with BUC who met the following inclusion and exclusion criteria were collected between June 2021 to June 2023. The inclusion criteria were as follows: 1) having undergone dual-energy spectral computed tomography urography (DEsCTU) within a fortnight before surgery; 2) availability of clinical and imaging datasets, such as pathological grade and DEsCTU raw data. Exclusion criteria were as follows: 1) inadequate image quality for analysis due to metal artifacts or motion artifacts; 2) presence of multiple lesions; 3) lesions with a maximum diameter less than 10 mm, or showing only limited thickening of the bladder wall without a defined tumor. This study included a total of 112 patients, consisting of 76 cases of HBUC and 36 cases of LBUC. CT acquisition The DECT imaging was performed using a 256-row CT scanner (Revolution CT, GE HealthCare, Milwaukee, WI, USA). The imaging parameters were as follows: Gemstone Spectral Imaging (GSI) mode with fast tube voltages switching between 80 kVp and 140 kVp, 400 mA tube current, pitch of 0.992:1, tube rotation speed of 0.6 s/r, detector width of 80 mm. The nonionic contrast media (Ioversol, 350 mgI/mL, Jiangsu Hengrui Pharmaceutical Co. Ltd., China) was used for contrast-enhanced imaging. The contrast media was injected through the antecubital vein at a dose of 500 mgI/kg (maximum of 100 ml), with a flow of 3.0–4.0 ml/s using a high-pressure syringe, succeeded by a saline flush. Image acquisition during the venous phase of contrast enhancement started 70 seconds after the injection of the contrast media. After the scanning is complete, 13 sets of images are generated for analysis, including 11 sets of VMIs with photon energies from 40 to 140 keV VMIs (10 keV intervals), iodine maps, and water maps. The adaptive statistical iterative reconstruction-V (ASIR-V) algorithm at 60% strength with 1.25 mm slice thickness and slice interval was used for all reconstructions. Tumor segmentation and feature extraction The tumor segmentation, radiomics feature extraction, selection, model building, and evaluation were established on the uAI Research Portal V1.1 (Shanghai United Imaging Intelligence, Co., Ltd.), as shown in Fig. 1 . Tumor segmentation The tumor segmentation was performed by a radiologist (W W, with 10 years of experience in CT abdominal imaging) in 40 keV VMI. During the segmentation of the entire tumor volume of interest (VOI), efforts were made to avoid necrosis, cystic degeneration, and calcifications. To ensure accuracy, another senior radiologist (Y L, with 20 years of experience in abdominal imaging) conducted a visual examination. The radiologists were unaware of the pathological findings. Subsequently, the segmentation VOI was replicated onto 12 other sets of images. Feature extraction : First, the preprocessing of the 13 sets of images involved three key steps: resampling the voxel size to 1×1×1 mm using the B-spline interpolation method, and discretizing gray values into 25 width bins. Then, from each set of images, a total of 90 radiomic features of 6 classes were extracted: first-order features (n = 18), gray-level co-occurrence matrix (GLCM) features (n = 21), gray-level size zone matrix (GLSZM) features (n = 16), gray-level run length matrix (GLRLM) features (n = 16), gray-level difference matrix (GLDM) features (n = 14), and neighborhood gray-tone difference matrix (NGTDM,) features (n = 5). Feature selection and model construction Before radiomic feature selection, all radiomic features were standardized using the Z-score normalization method, which rescaled the values to a range of 0 to 1. In the analysis of each feature set, a step-wise feature selection strategy was adopted to reduce the dimension of the features and prevent overfitting. Firstly, 30 features with the highest classification accuracy were selected using recursive feature elimination (RFE) which combines the advantages of wrapper and embedded methods. Subsequently, the Minimum Redundancy Maximum Relevance (mRMR) algorithm was applied to diminish mutual feature redundancy while maintaining 10 features of utmost relevance. Finally, the Least Absolute Shrinkage and Selection Operator (LASSO) was utilized to identify the optimal subset of features for each feature set. Thirteen predictive models were established to identify HBUC and LBUC using the random forest (RF) classifier. To avoid over-fitting the model in practical application, the model was trained and verified by the method of 5-fold cross verification. The predictive performance of each model was evaluated by constructing receiver operating curves (ROC) and calculating metrics such as area under the curve (AUC), sensitivity, specificity, and accuracy. Furthermore, Decision Curve Analysis (DCA) was utilized to assess the net benefits associated with each of the 13 models. Statistical analysis Statistical analysis was performed using MedCalc version 20.2 (MedCalc, Ltd, Ostend, Belgium), R package (version 4.2.1), and SPSS version 26.0 (SPSS Inc., Chicago, IL, USA). The age distribution, height, and weight were compared using one-way ANOVA, while chi-square tests were employed to evaluate gender differences. Data normality was assessed using the Kolmogorov-Smirnov test, and continuous variables were presented as mean ± standard deviation. Independent sample t-tests were used for parameters with normal distribution and homogeneity of variance, while the Mann-Whitney U-test was used for parameters without normal distribution or homogeneity of variance. The AUC of 13 models was compared using the Delong test. A significance level of p < 0.05 was used to determine statistical significance. Results Clinical characteristics The study encompassed a cohort of 112 patients, comprising 76 individuals with HBUC and 36 individuals with LBUC. Patient characteristics are summarized in Table 1 . There was no significant difference in sex, age, height, weight, and body mass index (BMI) between the HBUC and LBUC (all p > 0.05). Table 1 Clinical characteristics between low- and high- grade groups Parameters Low grade (n = 36) High grade (n = 76) Test value P-value Gender 0.895 a 0.436 F 7 11 M 29 65 Age (y) 60.25 ± 10.60 67.67 ± 7.75 -1.224 b 0.226 Height (m) 169.83 ± 8.26 171.14 ± 7.08 -0.867 b 0.388 Weight (kg) 71.10 ± 13.58 72.21 ± 11.03 -0.462 b 0.645 BMI 24.53 ± 3.61 24.66 ± 3.47 -0.182 b 0.856 Note: a represents X 2 statistics; b represents t values; BMI, body mass index Diagnostic performance of different image types Selecting for radiomics features thirteen models each utilized an optimal subset of features to construct predictive models for the grading of BUC, with the details of these features presented in Table 2 . Among them, GLCM features were incorporated into the construction of 10 models (except for 80 keV VMI, 130 keV VMI, and iodine maps). About different image types, GLDM features were employed in the construction of models for low keV (40keV-70keV VMI, excluding 60 keV VMI); for high keV VMI (80 keV-140 keV VMI, excluding 100 keV VMI) GLRLM features were selected; and for the base material images, GLSZM features were consistently used. Table 2 The selected feature sets for the radiomics models Model Number of Features Features 40keV 3 GLCM: Idn GLDM: DependenceVariance GLSZM: LowGrayLevelZoneEmphasis 50keV 2 GLCM: Idmn GLDM: LargeDependenceEmphasis 60keV 4 GLCM: Correlation GLSZM: SmallAreaEmphasis GLRLM: RunPercentage NGTDM: Coarseness 70keV 7 GLCM: Idmn, Imc2 GLDM: SmallDependenceEmphasis, DependenceVariance GLSZM: ZoneEntropy, SmallAreaEmphasis, GrayLevelNonUniformity 80keV 2 GLDM: DependencEvariance GLRLM: RunEntropy 90keV 7 GLCM: Idmn, Imc2 GLDMS: mallDependenceLowGrayLevelEmphasis, SmallDependenceEmphasis GLSZM: SizeZoneNonUniformityNormalized GLRLM: RunEntropy NGTDM: Busyness 100keV 8 GLCM: Idmn, Correlation GLDM: DependenceVariance, SmallDependenceEmphasis, SmallDependenceLowGrayLevelEmphasis GLSZM: SmallAreaEmphasis NGTDM: Busyness, Coarseness 110keV 3 GLCM: Imc2 GLSZM: ZoneEntropy GLRLM: RunPercentage 120keV 5 GLCM: Imc2 GLSZM: SmallAreaEmphasis GLRLM: RunEntropy, ShortRunEmphasis NGTDM: Busyness 130keV 1 GLRLM: RunEntropy 140keV 4 GLCM: dmn, Imc2 GLRLM: RunEntropy NGTDM: Busyness I(W) 2 GLSZM: ZoneEntropy GLRLM: RunEntropy W(I) 5 GLCM: Correlation GLDM: LargeDependenceEmphasis, SmallDependenceEmphasis GLSZM: SmallAreaEmphasis, GrayLevelNonUniformity Note: GLCM = gray level co-occurrence matrix, GLDM = gray level dependence matrix, GLSZM = gray level size zone matrix, GLRLM = gray level run length matrix, NGTDM = neighborhood gray tone difference matrix. Performance different image types The training set ROC curve area under the curve (AUC) for predicting the pathological grading of BUC in 13 models ranged from 0.91 to 0.96, while the AUC for the testing set ROC curve ranged from 0.84 to 0.90 (Table 3 ). The DeLong test results showed that there were no statistically significant differences in the AUC between the training set and the testing set for all 13 models ( p > 0.05) (Fig. 2 ). Table 3 Diagnostic performance of different models Model AUC (95% CI) Sensitivity (%) Specificity (%) Accuracy (%) Training set Testing set Training set Testing set Training set Testing set Training set Testing set 40 0.956 (0.925 ~ 0.995) 0.895 (0.776 ~ 1) 0.914 0.895 0.786 0.686 0.874 0.828 50 0.932(0.886 ~ 0.985) 0.874(0.72 ~ 0.999) 0.934 0.871 0.729 0.543 0.869 0.766 60 0.934(0.888 ~ 0.987) 0.855(0.706 ~ 0.998) 0.977 0.858 0.893 0.629 0.95 0.785 70 0.951(0.916 ~ 0.993) 0.889(0.757 ~ 1) 0.941 0.869 0.771 0.629 0.887 0.793 80 0.924(0.874 ~ 0.980) 0.841(0.67 ~ 0.993) 0.911 0.857 0.75 0.629 0.86 0.785 90 0.939(0.892 ~ 0.991) 0.868(0.727 ~ 0.989) 0.957 0.857 0.764 0.629 0.896 0.784 100 0.938(0.892 ~ 0.991) 0.852(0.696 ~ 0.991) 0.947 0.843 0.743 0.6 0.883 0.766 110 0.919(0.865 ~ 0.979) 0.847(0.695 ~ 0.985) 0.941 0.896 0.593 0.457 0.832 0.757 120 0.958(0.925 ~ 0.997) 0.889(0.766 ~ 1) 0.934 0.857 0.814 0.714 0.896 0.811 130 0.917(0.866 ~ 0.974) 0.887(0.771 ~ 0.992) 0.885 0.855 0.757 0.714 0.845 0.81 140 0.931(0.886 ~ 0.983) 0.863(0.711 ~ 0.994) 0.915 0.855 0.764 0.657 0.867 0.792 I(W) 0.906(0.846 ~ 0.970) 0.873(0.727 ~ 1) 0.901 0.813 0.708 0.632 0.839 0.758 W(I) 0.920(0.868 ~ 0.979) 0.852(0.71 ~ 0.977) 0.941 0.909 0.743 0.657 0.878 0.829 Notes: AUC = area under the receiver operating characteristic curve, CI = confidence interval. Across the majority of thresholds, the DCA of the models yielded favorable net benefits in low-energy images (40-70keV VMIs), high-energy images (80-140keV VMIs), and in iodine maps and water maps (Fig. 3 ), suggesting considerable predictive value. Discussion The DECT expands the functionality of traditional CT by providing images in the range of 40 to 140 keV VMIs and material separation technology[22, 23]. VMIs with low keV reconstruction (40-70 keV VMIs) can improve the contrast of iodine and the improve detection of BUC lesions[24]. VMIs with high keV reconstruction (80-140 keV VMI) reduce the sclerotic artifacts at the hip joint[25]. Iodine-water and water-iodine maps, based on material separation technology, can objectively reflect the iodine uptake and blood supplying changes in the lesions[26]. In this study, we investigated the impact of 13 different reconstruction parameters on the discrimination of BUC grading, including 40-140 keV VMIs, iodine-water maps, and water-iodine maps. For the models constructed using different parameters, the AUC of the training set and validation set ranged from 0.91 to 0.96 (greater than 0.9) and 0.84 to 0.90 (greater than 0.8), indicating a high diagnostic value for predicting BUC grading. Moreover, there was no significant difference in the AUC of the discrimination ability among the different parameters, indicating that changes in spectral parameters do not affect the discriminatory ability. For further validation the performance of these radiomic models, the DCA was also conducted, and the results support that within a certain threshold range, the net benefit of both high and low-energy VMIs and iodine-based image models was helpful in clinical decision-making. The imaging-genomic models constructed based on 13 different parameter images could successfully differentiate high and low-grade BUC, and spectral CT parameter variations did not affect the accuracy of the imaging-genomic discrimination models for BUC. Due to reasons such as small tumor volume and low X-ray attenuation, conventional CT scans sometimes have difficulty diagnosing some BCa[24]. Previous study has shown[27] that VMIs can improve the signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR). In addition, the reconstruction of images at 40 keV enhances the contrast of BCa and the bladder wall, representing the optimal single energy for diagnosing BCa. The venous phase provides higher temporal images for detecting bladder cancer, making it easier to segment the tumor with minimal manual delineation. Therefore, in this study, the VOI of BCa was delineated on 40 keV VMI during the venous phase. Researchers used 40-140 keV VMIs for the study of differentiating parotid tumors and adenolymphomas, the results indicate that for the classification of benign parotid tumors, the texture analysis of a multi-energy dataset is superior to the texture analysis of a single-energy dataset at 65 keV VMI[28]. In a study by Forghani et al.[29], which evaluated the lymph node metastasis of head and neck squamous cell carcinoma, texture features and iodine-based images were extracted from 40 to 140 keV VMI (at 5 keV intervals). The results showed that multi-energy texture analysis outperforms the analysis of datasets at single energies, but the texture analysis of the iodine map does not affect the model's performance. Simplifying the model (increasing the interval from 5 keV to 10 keV) has minimal impact on the overall performance. Thirteen radiomic models constructed based on different parameters can successfully distinguish between high- and low-grade bladder cancer. In this study, there was no further statistical analysis of the discriminative performance of the multi-energy model, as the AUC values of using different energy images in the ROC study were all greater than 0.91, indicating a high diagnostic efficiency. Through the analysis of the 13 selected optimal subsets, GLCM features were selected as non-zero coefficients in 10 of the 13 radiomics models, excluding 80 keV VMI, 130 keV VMI, and iodine map. This is crucial for maintaining the high efficiency of the multiple radiomics models, as GLCM features can quantify the structural changes of cells and their organelles under various conditions, serving as essential factors in distinguishing between low-grade and high-grade tumors[30]. Furthermore, non-zero coefficients for GLDM features were observed in the lower-level 40-70 keV VMIs (except for 60 keV), while non-zero coefficients for GLRLM features were present in the higher-level 80-140 keV VMIs (except for 100 keV). Additionally, non-zero coefficients for GLSZM features were identified in the base substance map. These results suggest that different types of images can capture features that correlate with high and low tumor grades. Moreover, the study findings revealed that the inverse difference normalized (IDN) features derived from GLCM were exclusively present in the 40 keV VMI for distinguishing tissue components, potentially linked to the heightened tissue contrast in the 40 keV VMI. Our study has several limitations. First, the study was retrospective and may result in inherent biases. Second, some discrepancies caused by manually outlined VOIs are unavoidable, even though we had made efforts to minimize the bias by using two trained radiologists. Third, our data were only from a single center. In the future, we will try to collect multicenter data to reinforce the conclusions of our study. In conclusion, the variation of spectral CT parameters does not affect the radiomics-based prediction of the pathological grading of BUC. Even if different relevant features are selected from images with different parameters, the accuracy of the model is not impacted. Declarations Reporting checklist: The authors have completed the STARD reporting checklist. Conflicts of Interest: The authors have no conflicts of interest to declare. Author Contribution W. W. : Data curation, Methodology, Software, Writing – original draft, Writing – review & editing. S.G. W. : Writing – original draft, Writing – review & editing.M.T. H. : Methodology, Validation. X.Y. T. : Investigation, Software. Y. F. : Methodology, Visualization. J.Y. Z. : Data curation, Visualization. Q.Y.C. : Software, Methodology D.S. D. : investigation, SupervisionL. L. : Conceptualization, Methodology, investigation, Validation, Supervision References Z.T. Dai, Y. Xiang, Y. Wang, L.Y. Bao, J. Wang, J.P. Li, H.M. Zhang, Z. Lu, S. Ponnambalam, X.H. Liao, Prognostic value of members of NFAT family for pan-cancer and a prediction model based on NFAT2 in bladder cancer, Aging (Albany NY) 13(10) (2021) 13876-13897. http://doi.org/10.18632/aging.202982. H. Sung, J. Ferlay, R.L. Siegel, M. Laversanne, I. Soerjomataram, A. Jemal, F. Bray, Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries, CA: A Cancer Journal for Clinicians 71(3) (2021) 209-249. http://doi.org/10.3322/caac.21660. Y. Wang, J. Wu, W. Luo, H. Zhang, G. Shi, Y. Shen, Y. Zhu, C. Ma, B. Dai, D. Ye, Y. Zhu, ALPK2 acts as tumor promotor in development of bladder cancer through targeting DEPDC1A, Cell Death Dis 12(7) (2021) 661. http://doi.org/10.1038/s41419-021-03947-7. H.J. Wang, Q. Cai, Y.P. Huang, M.Q. Li, Z.H. Wen, Y.Y. Lin, L.Y. Ouyang, L. Qian, Y. Guo, Amide Proton Transfer-weighted MRI in Predicting Histologic Grade of Bladder Cancer, Radiology 305(1) (2022) 127-134. http://doi.org/10.1148/radiol.211804. G. Zhang, L. Xu, L. Zhao, L. Mao, X. Li, Z. Jin, H. Sun, CT-based radiomics to predict the pathological grade of bladder cancer, European Radiology 30(12) (2020) 6749-6756. http://doi.org/10.1007/s00330-020-06893-8. P.A. Humphrey, H. Moch, A.L. Cubilla, T.M. Ulbright, V.E. Reuter, The 2016 WHO Classification of Tumours of the Urinary System and Male Genital Organs-Part B: Prostate and Bladder Tumours, Eur Urol 70(1) (2016) 106-119. http://doi.org/10.1016/j.eururo.2016.02.028. T.W. Flaig, P.E. Spiess, M. Abern, N. Agarwal, R. Bangs, S.A. Boorjian, M.K. Buyyounouski, K. Chan, S. Chang, T. Friedlander, R.E. Greenberg, K.A. Guru, H.W. Herr, J. Hoffman-Censits, A. Kishan, S. Kundu, S.M. Lele, R. Mamtani, V. Margulis, O.Y. Mian, J. Michalski, J.S. Montgomery, L. Nandagopal, L.C. Pagliaro, M. Parikh, A. Patterson, E.R. Plimack, K.S. Pohar, M.A. Preston, K. Richards, W.J. Sexton, A.O. Siefker-Radtke, M. Tollefson, J. Tward, J.L. Wright, M.A. Dwyer, C.J. Cassara, L.A. Gurski, NCCN Guidelines® Insights: Bladder Cancer, Version 2.2022, J Natl Compr Canc Netw 20(8) (2022) 866-878. http://doi.org/10.6004/jnccn.2022.0041. M. Ferro, G. Di Lorenzo, C. Buonerba, G. Lucarelli, G.I. Russo, F. Cantiello, A.R.A. Farhan, S. Di Stasi, G. Musi, R. Hurle, S. Vincenzo, G.M. Busetto, E. De Berardinis, S. Perdonà, M. Borghesi, R. Schiavina, G.L. Almeida, P. Bove, E. Lima, G. Grimaldi, D.V. Matei, F.A. Mistretta, N. Crisan, D. Terracciano, V. Paolo, M. Battaglia, G. Guazzoni, R. Autorino, G. Morgia, R. Damiano, M. Muto, R. Rocca, V. Mirone, O. de Cobelli, M.D. Vartolomei, Predictors of Residual T1 High Grade on Re-Transurethral Resection in a Large Multi-Institutional Cohort of Patients with Primary T1 High-Grade/Grade 3 Bladder Cancer, J Cancer 9(22) (2018) 4250-4254. http://doi.org/10.7150/jca.26129. Z. Ping, X. Zhan, T. Chen, Y. Zheng, M. Jiang, Y. Li, B. Fu, Survival Outcome of Partial Cystectomy versus Transurethral Bladder Tumor Resection in T1 High-Grade Bladder Cancer Patients: A Propensity Score Matching Study, J Oncol 2022 (2022) 3016725. http://doi.org/10.1155/2022/3016725. T.W. Flaig, P.E. Spiess, N. Agarwal, R. Bangs, S.A. Boorjian, M.K. Buyyounouski, S. Chang, T.M. Downs, J.A. Efstathiou, T. Friedlander, R.E. Greenberg, K.A. Guru, T. Guzzo, H.W. Herr, J. Hoffman-Censits, C. Hoimes, B.A. Inman, M. Jimbo, A.K. Kader, S.M. Lele, J. Michalski, J.S. Montgomery, L. Nandagopal, L.C. Pagliaro, S.K. Pal, A. Patterson, E.R. Plimack, K.S. Pohar, M.A. Preston, W.J. Sexton, A.O. Siefker-Radtke, J. Tward, J.L. Wright, L.A. Gurski, A. Johnson-Chilla, Bladder Cancer, Version 3.2020, NCCN Clinical Practice Guidelines in Oncology, J Natl Compr Canc Netw 18(3) (2020) 329-354. http://doi.org/10.6004/jnccn.2020.0011. Q. Cai, Z. Wen, Y. Huang, M. Li, L. Ouyang, J. Ling, L. Qian, Y. Guo, H. Wang, Investigation of Synthetic Magnetic Resonance Imaging Applied in the Evaluation of the Tumor Grade of Bladder Cancer, Journal of Magnetic Resonance Imaging 54(6) (2021) 1989-1997. http://doi.org/10.1002/jmri.27770. H. Wang, D. Hu, H. Yao, M. Chen, S. Li, H. Chen, J. Luo, Y. Feng, Y. Guo, Radiomics analysis of multiparametric MRI for the preoperative evaluation of pathological grade in bladder cancer tumors, Eur Radiol 29(11) (2019) 6182-6190. http://doi.org/10.1007/s00330-019-06222-8. Y. Song, D. Jin, N. Ou, Z. Luo, G. Chen, J. Chen, Y. Yang, X. Liu, Gene Expression Profiles Identified Novel Urine Biomarkers for Diagnosis and Prognosis of High-Grade Bladder Urothelial Carcinoma, Frontiers in Oncology 10 (2020). http://doi.org/10.3389/fonc.2020.00394. V. Panebianco, Y. Narumi, G. Barchetti, R. Montironi, J.W.F. Catto, Should We Perform Multiparametric Magnetic Resonance Imaging of the Bladder Before Transurethral Resection of Bladder? Time to Reconsider the Rules, Eur Urol 76(1) (2019) 57-58. http://doi.org/10.1016/j.eururo.2019.03.046. G.J. Netto, M.B. Amin, D.M. Berney, E.M. Compérat, A.J. Gill, A. Hartmann, S. Menon, M.R. Raspollini, M.A. Rubin, J.R. Srigley, P. Hoon Tan, S.K. Tickoo, T. Tsuzuki, S. Turajlic, I. Cree, H. Moch, The 2022 World Health Organization Classification of Tumors of the Urinary System and Male Genital Organs-Part B: Prostate and Urinary Tract Tumors, Eur Urol 82(5) (2022) 469-482. http://doi.org/10.1016/j.eururo.2022.07.002. A. Heinrich, S. Schenkl, D. Buckreus, F.V. Güttler, U.K.M. Teichgräber, CT-based thermometry with virtual monoenergetic images by dual-energy of fat, muscle and bone using FBP, iterative and deep learning–based reconstruction, European Radiology 32(1) (2021) 424-431. http://doi.org/10.1007/s00330-021-08206-z. E. Bicci, M. Mastrorosato, G. Danti, L. Lattavo, E. Bertelli, D. Cozzi, S. Pradella, S. Agostini, V. Miele, Dual-Energy CT applications in urinary tract cancers: an update, Tumori Journal 109(2) (2022) 148-156. http://doi.org/10.1177/03008916221088883. M. Kozikowski, R. Suarez-Ibarrola, R. Osiecki, K. Bilski, C. Gratzke, S.F. Shariat, A. Miernik, J. Dobruch, Role of Radiomics in the Prediction of Muscle-invasive Bladder Cancer: A Systematic Review and Meta-analysis, European Urology Focus 8(3) (2022) 728-738. http://doi.org/10.1016/j.euf.2021.05.005. D. Han, Y. Yu, N. Yu, S. Dang, H. Wu, R. Jialiang, T. He, Prediction models for clear cell renal cell carcinoma ISUP/WHO grade: comparison between CT radiomics and conventional contrast-enhanced CT, Br J Radiol 93(1114) (2020) 20200131. http://doi.org/10.1259/bjr.20200131. J. Choe, S.M. Lee, K.-H. Do, J.B. Lee, S.M. Lee, J.-G. Lee, J.B. Seo, Prognostic value of radiomic analysis of iodine overlay maps from dual-energy computed tomography in patients with resectable lung cancer, European Radiology 29(2) (2018) 915-923. http://doi.org/10.1007/s00330-018-5639-0. W. Zhang, J. Liu, W. Jin, R. Li, X. Xie, W. Zhao, S. Xia, D. Han, Radiomics from dual-energy CT-derived iodine maps predict lymph node metastasis in head and neck squamous cell carcinoma, La radiologia medica (2023). http://doi.org/10.1007/s11547-023-01750-2. Y. Wan, H. Hao, Y. Chen, Y. Zhang, Q. Yue, Z. Li, Application of spectral CT combined with perfusion scan in diagnosis of pancreatic neuroendocrine tumors, Insights into Imaging 13(1) (2022). http://doi.org/10.1186/s13244-022-01282-9. J. Li, D. Dong, M. Fang, R. Wang, J. Tian, H. Li, J. Gao, Dual-energy CT–based deep learning radiomics can improve lymph node metastasis risk prediction for gastric cancer, European Radiology 30(4) (2020) 2324-2333. http://doi.org/10.1007/s00330-019-06621-x. M. Nakagawa, T. Naiki, A. Naiki-Ito, Y. Ozawa, M. Shimohira, M. Ohnishi, Y. Shibamoto, Usefulness of advanced monoenergetic reconstruction technique in dual-energy computed tomography for detecting bladder cancer, Japanese Journal of Radiology 40(2) (2021) 177-183. http://doi.org/10.1007/s11604-021-01195-5. J. Schreck, K.R. Laukamp, J.H. Niehoff, A.E. Michael, J. Boriesosdick, M.M. Wöltjen, J.R. Kröger, R.P. Reimer, J.-P. Grunz, J. Borggrefe, S. Lennartz, Metal artifact reduction in patients with total hip replacements: evaluation of clinical photon counting CT using virtual monoenergetic images, European Radiology 33(12) (2023) 9286-9295. http://doi.org/10.1007/s00330-023-09879-4. J.S. Sung, L. Lebron, D. Keating, D. D’Alessio, C.E. Comstock, C.H. Lee, M.C. Pike, M. Ayhan, C.S. Moskowitz, E.A. Morris, M.S. Jochelson, Performance of Dual-Energy Contrast-enhanced Digital Mammography for Screening Women at Increased Risk of Breast Cancer, Radiology 293(1) (2019) 81-88. http://doi.org/10.1148/radiol.2019182660. A. Chen, A. Liu, J. Liu, S. Tian, H. Wang, Y. Liu, Application of dual-energy spectral CT imaging in differential diagnosis of bladder cancer and benign prostate hyperplasia, Medicine 95(52) (2016). http://doi.org/10.1097/md.0000000000005705. E. Al Ajmi, B. Forghani, C. Reinhold, M. Bayat, R. Forghani, Spectral multi-energy CT texture analysis with machine learning for tissue classification: an investigation using classification of benign parotid tumours as a testing paradigm, European Radiology 28(6) (2018) 2604-2611. http://doi.org/10.1007/s00330-017-5214-0. R. Forghani, A. Chatterjee, C. Reinhold, A. Pérez-Lara, G. Romero-Sanchez, Y. Ueno, M. Bayat, J.W.M. Alexander, L. Kadi, J. Chankowsky, J. Seuntjens, B. Forghani, Head and neck squamous cell carcinoma: prediction of cervical lymph node metastasis by dual-energy CT texture analysis with machine learning, European Radiology 29(11) (2019) 6172-6181. http://doi.org/10.1007/s00330-019-06159-y. X. Zhang, X. Xu, Q. Tian, B. Li, Y. Wu, Z. Yang, Z. Liang, Y. Liu, G. Cui, H. Lu, Radiomics assessment of bladder cancer grade using texture features from diffusion‐weighted imaging, Journal of Magnetic Resonance Imaging 46(5) (2017) 1281-1288. http://doi.org/10.1002/jmri.25669. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 12 Aug, 2024 Read the published version in Abdominal Radiology → Version 1 posted Editorial decision: Revision requested 13 Jul, 2024 Reviews received at journal 13 Jul, 2024 Reviews received at journal 12 Jul, 2024 Reviewers agreed at journal 12 Jul, 2024 Reviewers agreed at journal 12 Jul, 2024 Reviewers invited by journal 12 Jul, 2024 Editor assigned by journal 12 Jul, 2024 Submission checks completed at journal 12 Jul, 2024 First submitted to journal 11 Jul, 2024 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-4722594","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":326566141,"identity":"9eb8f54a-1c19-40f4-8ac3-2664383d93ae","order_by":0,"name":"Wei Wei","email":"","orcid":"","institution":"First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Wei","middleName":"","lastName":"Wei","suffix":""},{"id":326566142,"identity":"5e4a840b-b9e3-492c-92ab-b88492f87c91","order_by":1,"name":"Shigeng Wang","email":"","orcid":"","institution":"First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Shigeng","middleName":"","lastName":"Wang","suffix":""},{"id":326566143,"identity":"5e5f2cd2-df01-41a9-abdb-b3800e86b583","order_by":2,"name":"Mengting Hu","email":"","orcid":"","institution":"First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Mengting","middleName":"","lastName":"Hu","suffix":""},{"id":326566144,"identity":"139ffce1-acd0-4e56-9453-de29143cad20","order_by":3,"name":"Xiaoyu Tong","email":"","orcid":"","institution":"First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xiaoyu","middleName":"","lastName":"Tong","suffix":""},{"id":326566145,"identity":"1e6c941f-5056-4e6d-9cc1-9a475ef3c7b0","order_by":4,"name":"Yong Fan","email":"","orcid":"","institution":"First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yong","middleName":"","lastName":"Fan","suffix":""},{"id":326566146,"identity":"b3b62036-6d7c-45aa-84f8-51c7605b462c","order_by":5,"name":"Jingyi Zhang","email":"","orcid":"","institution":"First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Jingyi","middleName":"","lastName":"Zhang","suffix":""},{"id":326566147,"identity":"1c95d244-3c71-4153-bdf9-34d634cfff69","order_by":6,"name":"Qiye Cheng","email":"","orcid":"","institution":"First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Qiye","middleName":"","lastName":"Cheng","suffix":""},{"id":326566148,"identity":"54c5e569-04e1-4919-b6f2-2ae3cffa0841","order_by":7,"name":"Deshuo Dong","email":"","orcid":"","institution":"First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Deshuo","middleName":"","lastName":"Dong","suffix":""},{"id":326566149,"identity":"f12ea5d9-2138-41f9-bf0d-442e8e412cc0","order_by":8,"name":"Lei Liu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA5ElEQVRIiWNgGAWjYBACPmYGxgMMBswMDMzMBx98qJCQkyekhQ2oGKKFnS3ZcMYZC2PDBkJaGEBaGIBa+HnMpHnbKhJBXPxa2HkMDnwosI42OMxjbMw7TyKBsYH54aMbeB3GY3BwhkF67obDbIUP526TyGNnYDM2ziGg5TAQAbUwbzZ4u02imLGBh02aoJY/YC0MZhK8cyQSGw4Qo4UBrIXFTJK3gSgtbAUHe4B+mXkYFMjHJIwNmwn4hZ//8MYHP/5Y5/adPwyMypo6OXn25oeP8WnBAphJUz4KRsEoGAWjAAsAADdpSaMHstVCAAAAAElFTkSuQmCC","orcid":"","institution":"First Affiliated Hospital of Dalian Medical University","correspondingAuthor":true,"prefix":"","firstName":"Lei","middleName":"","lastName":"Liu","suffix":""}],"badges":[],"createdAt":"2024-07-11 07:47:30","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4722594/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4722594/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s00261-024-04516-0","type":"published","date":"2024-08-12T15:57:12+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":62216561,"identity":"03aeddc4-3f77-4ee9-9916-385c067a8b00","added_by":"auto","created_at":"2024-08-11 11:44:49","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":291513,"visible":true,"origin":"","legend":"\u003cp\u003eThe framework of the study\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-4722594/v1/49b0414c4af9d8565610c610.png"},{"id":62216558,"identity":"3abd8850-9b16-4d32-928d-1ebb185f5ac4","added_by":"auto","created_at":"2024-08-11 11:44:49","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":68667,"visible":true,"origin":"","legend":"\u003cp\u003eThe comparison of AUC in distinguishing low-grade and high- grade BUC\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-4722594/v1/83daa21b519160f368dd5ac0.png"},{"id":62216560,"identity":"7c36d7eb-a501-4d6c-9caa-c73e7e35ce3d","added_by":"auto","created_at":"2024-08-11 11:44:49","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":104606,"visible":true,"origin":"","legend":"\u003cp\u003eDecision curve analysis shows that the net benefits of high- and low-energy VMIs and iodine-based image models were helpful in clinical decision-making within a certain threshold range.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-4722594/v1/9f7752ecffa53d033811b16a.png"},{"id":63071129,"identity":"c2d1c09f-048d-43ff-8f04-f2a4e4d34814","added_by":"auto","created_at":"2024-08-22 20:03:41","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1012675,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4722594/v1/23531618-ef10-47e7-98d8-48c8ca527f26.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Impact of Multi-Parameter Images Obtained from Dual-Energy CT on Radiomicis to Predict Pathological Grading of Bladder Urothelial Carcinoma","fulltext":[{"header":"Introduction","content":"\u003cp\u003eBladder cancer (BCa) is currently one of the most common malignant tumors of the urinary system[\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Among these, bladder urothelial carcinoma (BUC) accounts for more than 90% of BCa and can be classified into high-grade BUC (HBUC) and low-grade BUC (LBUC) based on pathological grading. HBUC is more invasive, with a high risk of recurrence, metastasis, and poor prognosis[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Moreover, there are significant differences in the treatment methods for HBUC and LBUC[\u003cspan additionalcitationids=\"CR7 CR8\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. After the initial transurethral resection of the bladder tumor (TURBT), if HBUC is found, a repeat TURBT or bladder removal surgery is required to complete the lesion removal[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. Thus, effective and accurate preoperative classification of BUC histopathological grading is crucial for improving early diagnosis and patient survival. Currently, pathological examination remains the gold standard for BUC pathological grading[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. However, this invasive examination may lead to inaccurate classification due to inadequate samples, sometimes requiring repetition[\u003cspan additionalcitationids=\"CR14\" citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. Therefore, exploring accurate non-invasive preoperative prediction methods for pathological grading may contribute to the personalized management of BUC.\u003c/p\u003e \u003cp\u003eDual-energy computed tomography (DECT) can obtain multi-parameter images from two imaging datasets with different energy spectra, such as virtual monoenergetic images (VMIs) and material decomposition images[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. While previous studies have extracted additional radiomics quantitative parameters to predict BUC pathological grades in multi-parameter images, the inherent complexity of multi-parameter images has often been overlooked[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. VMIs encompass 101 discrete image sets, with energy levels spanning from 40 to 140 kilo-electron volts (keV); low-energy VMIs enhance BUC visualization whereas high-energy VMIs mitigate image artifacts[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. Material decomposition images elucidate iodine maps, depicting BUC microvasculature, and water maps, highlighting peritumoral edema linked to tumor infiltration[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. We hypothesized that the best multi-parameter images to predict BUC pathological grades may exist when using radiomics. Thus, the purpose of this study is to determine the diagnostic accuracy of radiomics derived from various VMIs and material decomposition images from DECT in predicting the pathologic grading of BUC and to identify models that show superior diagnostic performance.\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy population\u003c/h2\u003e \u003cp\u003e This retrospective study was approved by the Institutional Ethics Committee, and the consent from patients was waived. Patients diagnosed with BUC who met the following inclusion and exclusion criteria were collected between June 2021 to June 2023. The inclusion criteria were as follows: 1) having undergone dual-energy spectral computed tomography urography (DEsCTU) within a fortnight before surgery; 2) availability of clinical and imaging datasets, such as pathological grade and DEsCTU raw data. Exclusion criteria were as follows: 1) inadequate image quality for analysis due to metal artifacts or motion artifacts; 2) presence of multiple lesions; 3) lesions with a maximum diameter less than 10 mm, or showing only limited thickening of the bladder wall without a defined tumor. This study included a total of 112 patients, consisting of 76 cases of HBUC and 36 cases of LBUC.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003eCT acquisition\u003c/h2\u003e \u003cp\u003eThe DECT imaging was performed using a 256-row CT scanner (Revolution CT, GE HealthCare, Milwaukee, WI, USA). The imaging parameters were as follows: Gemstone Spectral Imaging (GSI) mode with fast tube voltages switching between 80 kVp and 140 kVp, 400 mA tube current, pitch of 0.992:1, tube rotation speed of 0.6 s/r, detector width of 80 mm. The nonionic contrast media (Ioversol, 350 mgI/mL, Jiangsu Hengrui Pharmaceutical Co. Ltd., China) was used for contrast-enhanced imaging. The contrast media was injected through the antecubital vein at a dose of 500 mgI/kg (maximum of 100 ml), with a flow of 3.0\u0026ndash;4.0 ml/s using a high-pressure syringe, succeeded by a saline flush. Image acquisition during the venous phase of contrast enhancement started 70 seconds after the injection of the contrast media. After the scanning is complete, 13 sets of images are generated for analysis, including 11 sets of VMIs with photon energies from 40 to 140 keV VMIs (10 keV intervals), iodine maps, and water maps. The adaptive statistical iterative reconstruction-V (ASIR-V) algorithm at 60% strength with 1.25 mm slice thickness and slice interval was used for all reconstructions.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eTumor segmentation and feature extraction\u003c/h2\u003e \u003cp\u003eThe tumor segmentation, radiomics feature extraction, selection, model building, and evaluation were established on the uAI Research Portal V1.1 (Shanghai United Imaging Intelligence, Co., Ltd.), as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eTumor segmentation\u003c/strong\u003e \u003cp\u003eThe tumor segmentation was performed by a radiologist (W W, with 10 years of experience in CT abdominal imaging) in 40 keV VMI. During the segmentation of the entire tumor volume of interest (VOI), efforts were made to avoid necrosis, cystic degeneration, and calcifications. To ensure accuracy, another senior radiologist (Y L, with 20 years of experience in abdominal imaging) conducted a visual examination. The radiologists were unaware of the pathological findings. Subsequently, the segmentation VOI was replicated onto 12 other sets of images.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eFeature extraction\u003c/b\u003e: First, the preprocessing of the 13 sets of images involved three key steps: resampling the voxel size to 1\u0026times;1\u0026times;1 mm using the B-spline interpolation method, and discretizing gray values into 25 width bins. Then, from each set of images, a total of 90 radiomic features of 6 classes were extracted: first-order features (n\u0026thinsp;=\u0026thinsp;18), gray-level co-occurrence matrix (GLCM) features (n\u0026thinsp;=\u0026thinsp;21), gray-level size zone matrix (GLSZM) features (n\u0026thinsp;=\u0026thinsp;16), gray-level run length matrix (GLRLM) features (n\u0026thinsp;=\u0026thinsp;16), gray-level difference matrix (GLDM) features (n\u0026thinsp;=\u0026thinsp;14), and neighborhood gray-tone difference matrix (NGTDM,) features (n\u0026thinsp;=\u0026thinsp;5).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003eFeature selection and model construction\u003c/h2\u003e \u003cp\u003eBefore radiomic feature selection, all radiomic features were standardized using the Z-score normalization method, which rescaled the values to a range of 0 to 1. In the analysis of each feature set, a step-wise feature selection strategy was adopted to reduce the dimension of the features and prevent overfitting. Firstly, 30 features with the highest classification accuracy were selected using recursive feature elimination (RFE) which combines the advantages of wrapper and embedded methods. Subsequently, the Minimum Redundancy Maximum Relevance (mRMR) algorithm was applied to diminish mutual feature redundancy while maintaining 10 features of utmost relevance. Finally, the Least Absolute Shrinkage and Selection Operator (LASSO) was utilized to identify the optimal subset of features for each feature set.\u003c/p\u003e \u003cp\u003eThirteen predictive models were established to identify HBUC and LBUC using the random forest (RF) classifier. To avoid over-fitting the model in practical application, the model was trained and verified by the method of 5-fold cross verification. The predictive performance of each model was evaluated by constructing receiver operating curves (ROC) and calculating metrics such as area under the curve (AUC), sensitivity, specificity, and accuracy. Furthermore, Decision Curve Analysis (DCA) was utilized to assess the net benefits associated with each of the 13 models.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003eStatistical analysis\u003c/h2\u003e \u003cp\u003eStatistical analysis was performed using MedCalc version 20.2 (MedCalc, Ltd, Ostend, Belgium), R package (version 4.2.1), and SPSS version 26.0 (SPSS Inc., Chicago, IL, USA). The age distribution, height, and weight were compared using one-way ANOVA, while chi-square tests were employed to evaluate gender differences. Data normality was assessed using the Kolmogorov-Smirnov test, and continuous variables were presented as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;standard deviation. Independent sample t-tests were used for parameters with normal distribution and homogeneity of variance, while the Mann-Whitney U-test was used for parameters without normal distribution or homogeneity of variance. The AUC of 13 models was compared using the Delong test. A significance level of \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 was used to determine statistical significance.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eClinical characteristics\u003c/h2\u003e \u003cp\u003eThe study encompassed a cohort of 112 patients, comprising 76 individuals with HBUC and 36 individuals with LBUC. Patient characteristics are summarized in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. There was no significant difference in sex, age, height, weight, and body mass index (BMI) between the HBUC and LBUC (all \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.05).\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\u003eClinical characteristics between low- and high- grade groups\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\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=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eParameters\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLow grade\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;36)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHigh grade\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;76)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eTest value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cem\u003eP-value\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGender\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\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e0.895\u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e0.436\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eF\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e11\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e29\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e65\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge (y)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e60.25\u0026thinsp;\u0026plusmn;\u0026thinsp;10.60\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e67.67\u0026thinsp;\u0026plusmn;\u0026thinsp;7.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-1.224\u003csup\u003eb\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.226\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHeight (m)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e169.83\u0026thinsp;\u0026plusmn;\u0026thinsp;8.26\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e171.14\u0026thinsp;\u0026plusmn;\u0026thinsp;7.08\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-0.867\u003csup\u003eb\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.388\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWeight (kg)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e71.10\u0026thinsp;\u0026plusmn;\u0026thinsp;13.58\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e72.21\u0026thinsp;\u0026plusmn;\u0026thinsp;11.03\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-0.462\u003csup\u003eb\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.645\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBMI\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e24.53\u0026thinsp;\u0026plusmn;\u0026thinsp;3.61\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e24.66\u0026thinsp;\u0026plusmn;\u0026thinsp;3.47\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-0.182\u003csup\u003eb\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.856\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003eNote: a represents X\u003csup\u003e2\u003c/sup\u003e statistics; b represents t values; BMI, body mass index\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003eDiagnostic performance of different image types\u003c/h2\u003e \u003cp\u003e \u003cstrong\u003eSelecting for radiomics features\u003c/strong\u003e \u003cp\u003ethirteen models each utilized an optimal subset of features to construct predictive models for the grading of BUC, with the details of these features presented in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. Among them, GLCM features were incorporated into the construction of 10 models (except for 80 keV VMI, 130 keV VMI, and iodine maps). About different image types, GLDM features were employed in the construction of models for low keV (40keV-70keV VMI, excluding 60 keV VMI); for high keV VMI (80 keV-140 keV VMI, excluding 100 keV VMI) GLRLM features were selected; and for the base material images, GLSZM features were consistently used.\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\u003eThe selected feature sets for the radiomics models\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\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=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNumber of Features\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eFeatures\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e40keV\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLCM: Idn\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLDM: DependenceVariance\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLSZM: LowGrayLevelZoneEmphasis\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e50keV\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLCM: Idmn\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLDM: LargeDependenceEmphasis\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003e60keV\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLCM: Correlation\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLSZM: SmallAreaEmphasis\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLRLM: RunPercentage\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNGTDM: Coarseness\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e70keV\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLCM: Idmn, Imc2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLDM: SmallDependenceEmphasis, DependenceVariance\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLSZM: ZoneEntropy, SmallAreaEmphasis, GrayLevelNonUniformity\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e80keV\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLDM: DependencEvariance\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLRLM: RunEntropy\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"4\" rowspan=\"5\"\u003e \u003cp\u003e90keV\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\" morerows=\"4\" rowspan=\"5\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLCM: Idmn, Imc2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLDMS: mallDependenceLowGrayLevelEmphasis, SmallDependenceEmphasis\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLSZM: SizeZoneNonUniformityNormalized\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLRLM: RunEntropy\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNGTDM: Busyness\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003e100keV\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLCM: Idmn, Correlation\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLDM: DependenceVariance, SmallDependenceEmphasis, SmallDependenceLowGrayLevelEmphasis\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLSZM: SmallAreaEmphasis\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNGTDM: Busyness, Coarseness\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e110keV\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLCM: Imc2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLSZM: ZoneEntropy\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLRLM: RunPercentage\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003e120keV\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLCM: Imc2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLSZM: SmallAreaEmphasis\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLRLM: RunEntropy, ShortRunEmphasis\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNGTDM: Busyness\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e130keV\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLRLM: RunEntropy\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e140keV\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLCM: dmn, Imc2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLRLM: RunEntropy\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNGTDM: Busyness\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eI(W)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLSZM: ZoneEntropy\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLRLM: RunEntropy\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003eW(I)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLCM: Correlation\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLDM: LargeDependenceEmphasis, SmallDependenceEmphasis\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGLSZM: SmallAreaEmphasis, GrayLevelNonUniformity\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"3\"\u003eNote: GLCM\u0026thinsp;=\u0026thinsp;gray level co-occurrence matrix, GLDM\u0026thinsp;=\u0026thinsp;gray level dependence matrix, GLSZM\u0026thinsp;=\u0026thinsp;gray level size zone matrix, GLRLM\u0026thinsp;=\u0026thinsp;gray level run length matrix, NGTDM\u0026thinsp;=\u0026thinsp;neighborhood gray tone difference matrix.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003ePerformance different image types\u003c/strong\u003e \u003cp\u003eThe training set ROC curve area under the curve (AUC) for predicting the pathological grading of BUC in 13 models ranged from 0.91 to 0.96, while the AUC for the testing set ROC curve ranged from 0.84 to 0.90 (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). The DeLong test results showed that there were no statistically significant differences in the AUC between the training set and the testing set for all 13 models (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.05) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\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\u003eDiagnostic performance of different models\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"9\"\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=\"left\" 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=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eModel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003eAUC (95% CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003eSensitivity (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e \u003cp\u003eSpecificity (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e \u003cp\u003eAccuracy (%)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTraining set\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTesting set\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eTraining set\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eTesting set\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eTraining set\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eTesting set\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eTraining set\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eTesting set\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\u003e40\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.956 (0.925\u0026thinsp;~\u0026thinsp;0.995)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.895\u003c/p\u003e \u003cp\u003e(0.776\u0026thinsp;~\u0026thinsp;1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.914\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.895\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.786\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.686\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.874\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.828\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003e50\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.932(0.886\u0026thinsp;~\u0026thinsp;0.985)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.874(0.72\u0026thinsp;~\u0026thinsp;0.999)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.934\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.871\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.729\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.543\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.869\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.766\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003e60\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.934(0.888\u0026thinsp;~\u0026thinsp;0.987)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.855(0.706\u0026thinsp;~\u0026thinsp;0.998)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.977\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.858\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.893\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.629\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.95\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.785\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003e70\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.951(0.916\u0026thinsp;~\u0026thinsp;0.993)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.889(0.757\u0026thinsp;~\u0026thinsp;1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.941\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.869\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.771\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.629\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.887\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.793\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003e80\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.924(0.874\u0026thinsp;~\u0026thinsp;0.980)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.841(0.67\u0026thinsp;~\u0026thinsp;0.993)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.911\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.857\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.629\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.86\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.785\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003e90\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.939(0.892\u0026thinsp;~\u0026thinsp;0.991)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.868(0.727\u0026thinsp;~\u0026thinsp;0.989)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.957\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.857\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.764\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.629\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.896\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.784\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003e100\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.938(0.892\u0026thinsp;~\u0026thinsp;0.991)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.852(0.696\u0026thinsp;~\u0026thinsp;0.991)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.947\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.843\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.743\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.883\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.766\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003e110\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.919(0.865\u0026thinsp;~\u0026thinsp;0.979)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.847(0.695\u0026thinsp;~\u0026thinsp;0.985)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.941\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.896\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.593\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.457\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.832\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.757\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003e120\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.958(0.925\u0026thinsp;~\u0026thinsp;0.997)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.889(0.766\u0026thinsp;~\u0026thinsp;1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.934\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.857\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.814\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.714\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.896\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.811\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003e130\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.917(0.866\u0026thinsp;~\u0026thinsp;0.974)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.887(0.771\u0026thinsp;~\u0026thinsp;0.992)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.885\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.855\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.757\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.714\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.845\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.81\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003e140\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.931(0.886\u0026thinsp;~\u0026thinsp;0.983)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.863(0.711\u0026thinsp;~\u0026thinsp;0.994)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.915\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.855\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.764\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.657\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.867\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.792\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eI(W)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.906(0.846\u0026thinsp;~\u0026thinsp;0.970)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.873(0.727\u0026thinsp;~\u0026thinsp;1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.901\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.813\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.708\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.632\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.839\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.758\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eW(I)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.920(0.868\u0026thinsp;~\u0026thinsp;0.979)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.852(0.71\u0026thinsp;~\u0026thinsp;0.977)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.941\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.909\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.743\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.657\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.878\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.829\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"9\"\u003eNotes: AUC\u0026thinsp;=\u0026thinsp;area under the receiver operating characteristic curve, CI\u0026thinsp;=\u0026thinsp;confidence interval.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/p\u003e \u003cp\u003eAcross the majority of thresholds, the DCA of the models yielded favorable net benefits in low-energy images (40-70keV VMIs), high-energy images (80-140keV VMIs), and in iodine maps and water maps (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e), suggesting considerable predictive value.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe DECT expands the functionality of traditional CT by providing images in the range of 40 to 140 keV VMIs and material separation technology[22, 23].\u0026nbsp;VMIs with low keV reconstruction\u0026nbsp;(40-70 keV VMIs) can improve the contrast of iodine and the improve detection of BUC lesions[24].\u0026nbsp;VMIs with high keV reconstruction\u0026nbsp;(80-140 keV VMI) reduce the sclerotic artifacts at the hip joint[25]. Iodine-water and water-iodine maps, based on material separation technology, can objectively reflect the iodine uptake and blood supplying changes in the lesions[26].\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp;In this study, we investigated the impact of 13 different reconstruction parameters on the discrimination of BUC grading, including 40-140 keV\u0026nbsp;VMIs, iodine-water maps, and water-iodine maps. For the models constructed using different parameters, the AUC of the training set and validation set ranged from 0.91 to 0.96 (greater than 0.9) and 0.84 to 0.90 (greater than 0.8), indicating a high diagnostic value for predicting BUC grading. Moreover, there was no significant difference in the AUC of the discrimination ability among the different parameters, indicating that changes in spectral parameters do not affect the discriminatory ability. For further validation the performance of these radiomic models, the DCA was also conducted, and the results support that within a certain threshold range, the net benefit of both high and low-energy VMIs and iodine-based image models was helpful in clinical decision-making. The imaging-genomic models constructed based on 13 different parameter images could successfully differentiate high and low-grade BUC, and spectral CT parameter variations did not affect the accuracy of the imaging-genomic discrimination models for BUC.\u003c/p\u003e\n\u003cp\u003eDue to reasons such as small tumor volume and low X-ray attenuation, conventional CT scans sometimes have difficulty diagnosing some BCa[24]. Previous study has shown[27]\u0026nbsp;that\u0026nbsp;VMIs\u0026nbsp;can improve the signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR). In addition, the reconstruction of images at 40 keV enhances the contrast of BCa and the bladder wall, representing the optimal single energy for diagnosing BCa. The venous phase provides higher temporal images for detecting bladder cancer, making it easier to segment the tumor with minimal manual delineation. Therefore, in this study, the VOI of BCa was delineated on 40 keV VMI during the venous phase.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; Researchers used 40-140 keV VMIs for the study of differentiating parotid tumors and adenolymphomas, the results indicate that for the classification of benign parotid tumors, the texture analysis of a multi-energy dataset is superior to the texture analysis of a single-energy dataset at 65 keV VMI[28]. In a study by Forghani et al.[29], which evaluated the lymph node metastasis of head and neck squamous cell carcinoma, texture features and iodine-based images were extracted from 40 to 140 keV VMI (at 5 keV intervals). The results showed that multi-energy texture analysis outperforms the analysis of datasets at single energies, but the texture analysis of the iodine map does not affect the model's performance. Simplifying the model (increasing the interval from 5 keV to 10 keV) has minimal impact on the overall performance. Thirteen radiomic models constructed based on different parameters can successfully distinguish between high- and low-grade bladder cancer. In this study, there was no further statistical analysis of the discriminative performance of the multi-energy model, as the AUC values of using different energy images in the ROC study were all greater than 0.91, indicating a high diagnostic efficiency.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; Through the analysis of the 13 selected optimal subsets, GLCM features were selected as non-zero coefficients in 10 of the 13 radiomics models, excluding 80 keV VMI, 130 keV VMI, and iodine map. This is crucial for maintaining the high efficiency of the multiple radiomics models, as GLCM features can quantify the structural changes of cells and their organelles under various conditions, serving as essential factors in distinguishing between low-grade and high-grade tumors[30]. Furthermore, non-zero coefficients for GLDM features were observed in the lower-level 40-70 keV VMIs (except for 60 keV), while non-zero coefficients for GLRLM features were present in the higher-level 80-140 keV VMIs (except for 100 keV). Additionally, non-zero coefficients for GLSZM features were identified in the base substance map. These results suggest that different types of images can capture features that correlate with high and low tumor grades. Moreover, the study findings revealed that the inverse difference normalized (IDN) features derived from GLCM were exclusively present in the 40 keV VMI for distinguishing tissue components, potentially linked to the heightened tissue contrast in the 40 keV VMI.\u003c/p\u003e\n\u003cp\u003eOur study has several limitations. First, the study was retrospective and may result in inherent biases. Second, some discrepancies caused by manually outlined VOIs are unavoidable, even though we had made efforts to minimize the bias by using two trained radiologists. Third, our data were only from a single center. In the future, we will try to collect multicenter data to reinforce the conclusions of our study.\u003c/p\u003e\n\u003cp\u003eIn conclusion, the variation of spectral CT parameters does not affect the radiomics-based prediction of the pathological grading of BUC. Even if different relevant features are selected from images with different parameters, the accuracy of the model is not impacted.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cem\u003eReporting checklist:\u0026nbsp;\u003c/em\u003eThe authors have completed the STARD reporting checklist.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eConflicts of Interest:\u0026nbsp;\u003c/em\u003eThe authors have no conflicts of interest to declare.\u003c/p\u003e\n\u003cp\u003eAuthor Contribution\u003c/p\u003e\n\u003cp\u003eW. W. : Data curation, Methodology, Software, Writing \u0026ndash; original draft, Writing \u0026ndash; review \u0026amp; editing. S.G. W. : Writing \u0026ndash; original draft, Writing \u0026ndash; review \u0026amp; editing.M.T. H. : Methodology, Validation. X.Y. T. : Investigation, Software. Y. F. : Methodology, Visualization. J.Y. Z. : Data curation, Visualization. Q.Y.C. : Software, Methodology D.S. D. : investigation, SupervisionL. L. : Conceptualization, Methodology, investigation, Validation, Supervision\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eZ.T. Dai, Y. Xiang, Y. Wang, L.Y. Bao, J. Wang, J.P. Li, H.M. Zhang, Z. Lu, S. Ponnambalam, X.H. Liao, Prognostic value of members of NFAT family for pan-cancer and a prediction model based on NFAT2 in bladder cancer, Aging (Albany NY) 13(10) (2021) 13876-13897. http://doi.org/10.18632/aging.202982.\u003c/li\u003e\n \u003cli\u003eH. Sung, J. Ferlay, R.L. Siegel, M. Laversanne, I. Soerjomataram, A. Jemal, F. Bray, Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries, CA: A Cancer Journal for Clinicians 71(3) (2021) 209-249. http://doi.org/10.3322/caac.21660.\u003c/li\u003e\n \u003cli\u003eY. Wang, J. Wu, W. Luo, H. Zhang, G. Shi, Y. Shen, Y. Zhu, C. Ma, B. Dai, D. Ye, Y. Zhu, ALPK2 acts as tumor promotor in development of bladder cancer through targeting DEPDC1A, Cell Death Dis 12(7) (2021) 661. http://doi.org/10.1038/s41419-021-03947-7.\u003c/li\u003e\n \u003cli\u003eH.J. Wang, Q. Cai, Y.P. Huang, M.Q. Li, Z.H. Wen, Y.Y. Lin, L.Y. Ouyang, L. Qian, Y. Guo, Amide Proton Transfer-weighted MRI in Predicting Histologic Grade of Bladder Cancer, Radiology 305(1) (2022) 127-134. http://doi.org/10.1148/radiol.211804.\u003c/li\u003e\n \u003cli\u003eG. Zhang, L. Xu, L. Zhao, L. Mao, X. Li, Z. Jin, H. Sun, CT-based radiomics to predict the pathological grade of bladder cancer, European Radiology 30(12) (2020) 6749-6756. http://doi.org/10.1007/s00330-020-06893-8.\u003c/li\u003e\n \u003cli\u003eP.A. Humphrey, H. Moch, A.L. Cubilla, T.M. Ulbright, V.E. Reuter, The 2016 WHO Classification of Tumours of the Urinary System and Male Genital Organs-Part B: Prostate and Bladder Tumours, Eur Urol 70(1) (2016) 106-119. http://doi.org/10.1016/j.eururo.2016.02.028.\u003c/li\u003e\n \u003cli\u003eT.W. Flaig, P.E. Spiess, M. Abern, N. Agarwal, R. Bangs, S.A. Boorjian, M.K. Buyyounouski, K. Chan, S. Chang, T. Friedlander, R.E. Greenberg, K.A. Guru, H.W. Herr, J. Hoffman-Censits, A. Kishan, S. Kundu, S.M. Lele, R. Mamtani, V. Margulis, O.Y. Mian, J. Michalski, J.S. Montgomery, L. Nandagopal, L.C. Pagliaro, M. Parikh, A. Patterson, E.R. Plimack, K.S. Pohar, M.A. Preston, K. Richards, W.J. Sexton, A.O. Siefker-Radtke, M. Tollefson, J. Tward, J.L. Wright, M.A. Dwyer, C.J. Cassara, L.A. Gurski, NCCN Guidelines\u0026reg; Insights: Bladder Cancer, Version 2.2022, J Natl Compr Canc Netw 20(8) (2022) 866-878. http://doi.org/10.6004/jnccn.2022.0041.\u003c/li\u003e\n \u003cli\u003eM. Ferro, G. Di Lorenzo, C. Buonerba, G. Lucarelli, G.I. Russo, F. Cantiello, A.R.A. Farhan, S. Di Stasi, G. Musi, R. Hurle, S. Vincenzo, G.M. Busetto, E. De Berardinis, S. Perdon\u0026agrave;, M. Borghesi, R. Schiavina, G.L. Almeida, P. Bove, E. Lima, G. Grimaldi, D.V. Matei, F.A. Mistretta, N. Crisan, D. Terracciano, V. Paolo, M. Battaglia, G. Guazzoni, R. Autorino, G. Morgia, R. Damiano, M. Muto, R. Rocca, V. Mirone, O. de Cobelli, M.D. Vartolomei, Predictors of Residual T1 High Grade on Re-Transurethral Resection in a Large Multi-Institutional Cohort of Patients with Primary T1 High-Grade/Grade 3 Bladder Cancer, J Cancer 9(22) (2018) 4250-4254. http://doi.org/10.7150/jca.26129.\u003c/li\u003e\n \u003cli\u003eZ. Ping, X. Zhan, T. Chen, Y. Zheng, M. Jiang, Y. Li, B. Fu, Survival Outcome of Partial Cystectomy versus Transurethral Bladder Tumor Resection in T1 High-Grade Bladder Cancer Patients: A Propensity Score Matching Study, J Oncol 2022 (2022) 3016725. http://doi.org/10.1155/2022/3016725.\u003c/li\u003e\n \u003cli\u003eT.W. Flaig, P.E. Spiess, N. Agarwal, R. Bangs, S.A. Boorjian, M.K. Buyyounouski, S. Chang, T.M. Downs, J.A. Efstathiou, T. Friedlander, R.E. Greenberg, K.A. Guru, T. Guzzo, H.W. Herr, J. Hoffman-Censits, C. Hoimes, B.A. Inman, M. Jimbo, A.K. Kader, S.M. Lele, J. Michalski, J.S. Montgomery, L. Nandagopal, L.C. Pagliaro, S.K. Pal, A. Patterson, E.R. Plimack, K.S. Pohar, M.A. Preston, W.J. Sexton, A.O. Siefker-Radtke, J. Tward, J.L. Wright, L.A. Gurski, A. Johnson-Chilla, Bladder Cancer, Version 3.2020, NCCN Clinical Practice Guidelines in Oncology, J Natl Compr Canc Netw 18(3) (2020) 329-354. http://doi.org/10.6004/jnccn.2020.0011.\u003c/li\u003e\n \u003cli\u003eQ. Cai, Z. Wen, Y. Huang, M. Li, L. Ouyang, J. Ling, L. Qian, Y. Guo, H. Wang, Investigation of Synthetic Magnetic Resonance Imaging Applied in the Evaluation of the Tumor Grade of Bladder Cancer, Journal of Magnetic Resonance Imaging 54(6) (2021) 1989-1997. http://doi.org/10.1002/jmri.27770.\u003c/li\u003e\n \u003cli\u003eH. Wang, D. Hu, H. Yao, M. Chen, S. Li, H. Chen, J. Luo, Y. Feng, Y. Guo, Radiomics analysis of multiparametric MRI for the preoperative evaluation of pathological grade in bladder cancer tumors, Eur Radiol 29(11) (2019) 6182-6190. http://doi.org/10.1007/s00330-019-06222-8.\u003c/li\u003e\n \u003cli\u003eY. Song, D. Jin, N. Ou, Z. Luo, G. Chen, J. Chen, Y. Yang, X. Liu, Gene Expression Profiles Identified Novel Urine Biomarkers for Diagnosis and Prognosis of High-Grade Bladder Urothelial Carcinoma, Frontiers in Oncology 10 (2020). http://doi.org/10.3389/fonc.2020.00394.\u003c/li\u003e\n \u003cli\u003eV. Panebianco, Y. Narumi, G. Barchetti, R. Montironi, J.W.F. Catto, Should We Perform Multiparametric Magnetic Resonance Imaging of the Bladder Before Transurethral Resection of Bladder? Time to Reconsider the Rules, Eur Urol 76(1) (2019) 57-58. http://doi.org/10.1016/j.eururo.2019.03.046.\u003c/li\u003e\n \u003cli\u003eG.J. Netto, M.B. Amin, D.M. Berney, E.M. Comp\u0026eacute;rat, A.J. Gill, A. Hartmann, S. Menon, M.R. Raspollini, M.A. Rubin, J.R. Srigley, P. Hoon Tan, S.K. Tickoo, T. Tsuzuki, S. Turajlic, I. Cree, H. Moch, The 2022 World Health Organization Classification of Tumors of the Urinary System and Male Genital Organs-Part B: Prostate and Urinary Tract Tumors, Eur Urol 82(5) (2022) 469-482. http://doi.org/10.1016/j.eururo.2022.07.002.\u003c/li\u003e\n \u003cli\u003eA. Heinrich, S. Schenkl, D. Buckreus, F.V. G\u0026uuml;ttler, U.K.M. Teichgr\u0026auml;ber, CT-based thermometry with virtual monoenergetic images by dual-energy of fat, muscle and bone using FBP, iterative and deep learning\u0026ndash;based reconstruction, European Radiology 32(1) (2021) 424-431. http://doi.org/10.1007/s00330-021-08206-z.\u003c/li\u003e\n \u003cli\u003eE. Bicci, M. Mastrorosato, G. Danti, L. Lattavo, E. Bertelli, D. Cozzi, S. Pradella, S. Agostini, V. Miele, Dual-Energy CT applications in urinary tract cancers: an update, Tumori Journal 109(2) (2022) 148-156. http://doi.org/10.1177/03008916221088883.\u003c/li\u003e\n \u003cli\u003eM. Kozikowski, R. Suarez-Ibarrola, R. Osiecki, K. Bilski, C. Gratzke, S.F. Shariat, A. Miernik, J. Dobruch, Role of Radiomics in the Prediction of Muscle-invasive Bladder Cancer: A Systematic Review and Meta-analysis, European Urology Focus 8(3) (2022) 728-738. http://doi.org/10.1016/j.euf.2021.05.005.\u003c/li\u003e\n \u003cli\u003eD. Han, Y. Yu, N. Yu, S. Dang, H. Wu, R. Jialiang, T. He, Prediction models for clear cell renal cell carcinoma ISUP/WHO grade: comparison between CT radiomics and conventional contrast-enhanced CT, Br J Radiol 93(1114) (2020) 20200131. http://doi.org/10.1259/bjr.20200131.\u003c/li\u003e\n \u003cli\u003eJ. Choe, S.M. Lee, K.-H. Do, J.B. Lee, S.M. Lee, J.-G. Lee, J.B. Seo, Prognostic value of radiomic analysis of iodine overlay maps from dual-energy computed tomography in patients with resectable lung cancer, European Radiology 29(2) (2018) 915-923. http://doi.org/10.1007/s00330-018-5639-0.\u003c/li\u003e\n \u003cli\u003eW. Zhang, J. Liu, W. Jin, R. Li, X. Xie, W. Zhao, S. Xia, D. Han, Radiomics from dual-energy CT-derived iodine maps predict lymph node metastasis in head and neck squamous cell carcinoma, La radiologia medica (2023). http://doi.org/10.1007/s11547-023-01750-2.\u003c/li\u003e\n \u003cli\u003eY. Wan, H. Hao, Y. Chen, Y. Zhang, Q. Yue, Z. Li, Application of spectral CT combined with perfusion scan in diagnosis of pancreatic neuroendocrine tumors, Insights into Imaging 13(1) (2022). http://doi.org/10.1186/s13244-022-01282-9.\u003c/li\u003e\n \u003cli\u003eJ. Li, D. Dong, M. Fang, R. Wang, J. Tian, H. Li, J. Gao, Dual-energy CT\u0026ndash;based deep learning radiomics can improve lymph node metastasis risk prediction for gastric cancer, European Radiology 30(4) (2020) 2324-2333. http://doi.org/10.1007/s00330-019-06621-x.\u003c/li\u003e\n \u003cli\u003eM. Nakagawa, T. Naiki, A. Naiki-Ito, Y. Ozawa, M. Shimohira, M. Ohnishi, Y. Shibamoto, Usefulness of advanced monoenergetic reconstruction technique in dual-energy computed tomography for detecting bladder cancer, Japanese Journal of Radiology 40(2) (2021) 177-183. http://doi.org/10.1007/s11604-021-01195-5.\u003c/li\u003e\n \u003cli\u003eJ. Schreck, K.R. Laukamp, J.H. Niehoff, A.E. Michael, J. Boriesosdick, M.M. W\u0026ouml;ltjen, J.R. Kr\u0026ouml;ger, R.P. Reimer, J.-P. Grunz, J. Borggrefe, S. Lennartz, Metal artifact reduction in patients with total hip replacements: evaluation of clinical photon counting CT using virtual monoenergetic images, European Radiology 33(12) (2023) 9286-9295. http://doi.org/10.1007/s00330-023-09879-4.\u003c/li\u003e\n \u003cli\u003eJ.S. Sung, L. Lebron, D. Keating, D. D\u0026rsquo;Alessio, C.E. Comstock, C.H. Lee, M.C. Pike, M. Ayhan, C.S. Moskowitz, E.A. Morris, M.S. Jochelson, Performance of Dual-Energy Contrast-enhanced Digital Mammography for Screening Women at Increased Risk of Breast Cancer, Radiology 293(1) (2019) 81-88. http://doi.org/10.1148/radiol.2019182660.\u003c/li\u003e\n \u003cli\u003eA. Chen, A. Liu, J. Liu, S. Tian, H. Wang, Y. Liu, Application of dual-energy spectral CT imaging in differential diagnosis of bladder cancer and benign prostate hyperplasia, Medicine 95(52) (2016). http://doi.org/10.1097/md.0000000000005705.\u003c/li\u003e\n \u003cli\u003eE. Al Ajmi, B. Forghani, C. Reinhold, M. Bayat, R. Forghani, Spectral multi-energy CT texture analysis with machine learning for tissue classification: an investigation using classification of benign parotid tumours as a testing paradigm, European Radiology 28(6) (2018) 2604-2611. http://doi.org/10.1007/s00330-017-5214-0.\u003c/li\u003e\n \u003cli\u003eR. Forghani, A. Chatterjee, C. Reinhold, A. P\u0026eacute;rez-Lara, G. Romero-Sanchez, Y. Ueno, M. Bayat, J.W.M. Alexander, L. Kadi, J. Chankowsky, J. Seuntjens, B. Forghani, Head and neck squamous cell carcinoma: prediction of cervical lymph node metastasis by dual-energy CT texture analysis with machine learning, European Radiology 29(11) (2019) 6172-6181. http://doi.org/10.1007/s00330-019-06159-y.\u003c/li\u003e\n \u003cli\u003eX. Zhang, X. Xu, Q. Tian, B. Li, Y. Wu, Z. Yang, Z. Liang, Y. Liu, G. Cui, H. Lu, Radiomics assessment of bladder cancer grade using texture features from diffusion‐weighted imaging, Journal of Magnetic Resonance Imaging 46(5) (2017) 1281-1288. http://doi.org/10.1002/jmri.25669.\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":"abdominal-radiology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"aima","sideBox":"Learn more about [Abdominal Radiology](http://link.springer.com/journal/261)","snPcode":"261","submissionUrl":"https://submission.springernature.com/new-submission/261/3","title":"Abdominal Radiology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Bladder cancer, Radiomics, Multi -parameter images, Dual-energy CT","lastPublishedDoi":"10.21203/rs.3.rs-4722594/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4722594/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eObjective\u003c/strong\u003e: To investigate the effect of radiomics models obtained from dual-energy CT (DECT) material decomposition images and virtual monoenergetic images (VMIs) in predicting the pathological grading of bladder urothelial carcinoma (BUC).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMaterials and Methods\u003c/strong\u003e: Preoperative Energy-Spectrum CT images were retrospectively collected from 112 pathologically confirmed cases of BUC patients, including 76 cases of high-grade urothelial carcinoma and 36 cases of low-grade urothelial carcinoma. Enhanced CT venous phase images of all patients were reconstructed at 40 to 140 keV VMIs (interval 10 keV), Iodine maps, and Water maps, and a total of 13 sets of images were obtained, and imaging features were extracted in each of the 13 sets of images. The best features related to BUC were identified by recursive feature elimination (RFE), the Minimum Redundancy Maximum Relevance (mRMR), and the Least Absolute Shrinkage and Selection Operator (LASSO) in order. A five-fold cross-validation method was used to divide the samples into training and testing sets, and models for pathological prediction of BUC grading were constructed by a random forest (RF) classifier. ROC curves were plotted to evaluate the performance of 13 models obtained from reconstructed images.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults: \u003c/strong\u003eThere were no statistically significant differences in the area under the curve (AUC) between the training set and the testing set for all 13 models, with the AUC ranging from 0.91 to 0.96 in the training set and 0.84 to 0.90 in the testing set for each group of reconstructed images. Although the features selected for the reconstructed images were very different among the groups, all the features selected from 40 to 100 keV VMIs had dependencevariance of the GLDM feature set.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion:\u003c/strong\u003e The variation of spectral CT parameters did no effect on the radiomics-based prediction of the pathological grading of BUC and did not affect the accuracy of the model even if the relevant features differed between reconstructed images.\u003c/p\u003e","manuscriptTitle":"Impact of Multi-Parameter Images Obtained from Dual-Energy CT on Radiomicis to Predict Pathological Grading of Bladder Urothelial Carcinoma","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-08-11 11:44:44","doi":"10.21203/rs.3.rs-4722594/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2024-07-13T17:48:40+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-07-13T14:15:07+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-07-13T02:31:22+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"258989427587327955368324309685698622846","date":"2024-07-12T22:03:01+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"290591745903675490867258900002923151745","date":"2024-07-12T21:50:55+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2024-07-12T21:49:19+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-07-12T10:11:32+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-07-12T10:10:34+00:00","index":"","fulltext":""},{"type":"submitted","content":"Abdominal Radiology","date":"2024-07-11T07:46:11+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"abdominal-radiology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"aima","sideBox":"Learn more about [Abdominal Radiology](http://link.springer.com/journal/261)","snPcode":"261","submissionUrl":"https://submission.springernature.com/new-submission/261/3","title":"Abdominal Radiology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"dece23cb-0869-48d6-a247-59900f61528d","owner":[],"postedDate":"August 11th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2024-08-22T19:32:17+00:00","versionOfRecord":{"articleIdentity":"rs-4722594","link":"https://doi.org/10.1007/s00261-024-04516-0","journal":{"identity":"abdominal-radiology","isVorOnly":false,"title":"Abdominal Radiology"},"publishedOn":"2024-08-12 15:57:12","publishedOnDateReadable":"August 12th, 2024"},"versionCreatedAt":"2024-08-11 11:44:44","video":"","vorDoi":"10.1007/s00261-024-04516-0","vorDoiUrl":"https://doi.org/10.1007/s00261-024-04516-0","workflowStages":[]},"version":"v1","identity":"rs-4722594","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4722594","identity":"rs-4722594","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}