CT based deep learning model for differentiating primary renal sarcomas from large renal cell carcinomas

CT based deep learning model for differentiating primary renal sarcomas from large renal cell carcinomas | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article CT based deep learning model for differentiating primary renal sarcomas from large renal cell carcinomas Haishan Lin, Shurong Li, Yuhang Chen, Zhenhua Chen, Jinhuan Wei, and 7 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4478575/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background Due to the prognosis and treatment differences between primary renal sarcomas and renal cell carcinoma, preoperative differentiation between them is important but challenging. This study aims to explore and develop a diagnostic method based on computed tomography (CT) and clinical data for preoperatively differentiating primary renal sarcomas from large renal cell carcinomas. Methods Patients pathologically diagnosed with primary renal sarcoma from two center between 2009–2021 were retrospectively included, and large renal cell carcinomas were probably 2:1 compared to renal sarcomas as the control group. Clinical data, standard contrast-enhanced CT images and histological findings were obtained. A clinical model was established with independent indicators based on logistic regression analysis. The region of interest was outlined in each three modal CT images (unenhanced phase [UP], corticomedullary phase [CMP] and nephrographic phase [NP]) and formed 7 modal imaging datasets for deep learning models’ development. Reported performance metrics included accuracy and areas under the receiver operating characteristic curves (AUC). Results Totally, 27 renal sarcomas and 58 large RCCs were enrolled. Multivariate logistic regression showed that the independent indicators of renal sarcoma were intratumoral artery and Gerota’s fascia invasion ( P < 0.05). The AUC of clinical model was 0.77 (95% confidence interval [CI]: 0.67–0.87), sensitivity 0.74, specificity 0.67, positive predictive value 0.51, and negative predictive value 0.85. The deep learning models yielded effective discrimination. The unenhanced phase model yielded AUC = 0.95±0.09 and accuracy (ACC) = 0.94±0.07, and UP + NP model yielded nearly AUC = 0.95±0.06, and ACC = 0.94±0.07. Conclusion The deep learning models based on multimodal CT images show good performance for differentiating renal sarcomas from large renal cell carcinomas, which assist in individualized management. Renal sarcoma Renal cell carcinoma Computed tomography Deep learning Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Adult primary renal sarcomas are rare renal malignancies derived from the mesenchymal tissue of the kidney. In previous studies, the prevalence of renal sarcoma was only 0.8–1.1%( 1 , 2 ). The epidemiological data of renal sarcomas in China accounted for 0.6–1.9% of renal tumor cases( 3 ). While, renal cancer is one of the top ten most common cancers worldwide, accounting for 2.2% of all malignant tumors. As the most common solid tumor of the kidney, renal cell carcinoma (RCC) accounts for about 80–90% of primary malignant tumors of the kidney( 4 , 5 ). For early-stage patients such as localized and locally advanced RCCs, nephron sparing surgery is the preferred treatment to preserved kidney function better( 6 – 8 ). The prognosis of patients with renal sarcomas is worse than RCCs due to the rapid growth and high malignancy( 3 ). The treatment strategy for primary renal sarcoma typically involves radical nephrectomy instead of nephron sparing surgery in order to reduce positive surgical margins( 9 – 11 ). Therefore, accurately diagnosing primary renal sarcoma can assist in preoperative clinical decision making and improve patient prognosis, which is an important issue to be solved in clinical practice. But the preoperative differential diagnosis between primary renal sarcoma and RCC before surgery is difficult, especially for the lager diameter tumors. In recent years, abdominal imaging has been widely used in health examinations, and the detective rate and diagnostic accuracy of renal tumors have significantly improved( 12 ). Computed tomography (CT) is the cornerstone of imaging method for screening and diagnosis of renal mass. The CT imaging features of renal sarcoma have been described in three reviews( 13 – 15 ). Based on the imaging features of sarcoma summarized by previous articles, Johannes Uhlig and et al. found that the imaging features associated with renal sarcoma were: tumor laterality (right side), larger maximum diameter, irregular shape, ill-defined margins, vascular invasion, necrosis, and organ invasion( 16 ). However, the studies above did not propose standard imaging evaluation method and neglected the differentiation between large RCC and renal sarcoma. Moreover, the high-throughput information of images needs further exploration. The application of artificial intelligence and radiomics in medical image analysis has developed rapidly. Radiomics can be used for disease screening, diagnosis, prognosis analysis and decision making( 17 , 18 ). Many studies have demonstrated that deep learning can be useful for image interpretation tasks and the models showed expert-level performance( 19 – 21 ). Compared with traditional feature engineering-based method, deep learning radiomics can learn features automatically without precise tumor annotation and distinguish the features adaptive to specific task though self-learning strategy. Therefore, we aimed to establish a deep learning model to distinguish adult primary renal sarcoma and large RCC preoperatively, which can assist in formulating individualized treatment strategies for adult primary renal sarcoma patients. Material and Methods The study was reviewed and approved by the Ethics Committee of the First Affiliated Hospital of Sun Yat-sen University (FAH-SYSU) (Nr [2022] 155) for retrospective case collection and analysis. Patient cohort Patients pathologically diagnosed with renal sarcoma at FAH-SYSU and Sun Yat-sen University Cancer Center (SYSUCC) from 2009 to 2021 were retrospectively included in this study. The pathological subtypes of renal sarcoma were based on world health organization classification of renal tumors( 22 ). Participants were excluded if they had previous history of sarcoma, presented with renal metastasis of non-renal originated sarcomas, with unsatisfied images quality, and if their age less than 18 or more than 80. The control group randomly included from consecutive patients who were initially diagnosed from 2018 to 2020, with a pathological diagnosis of RCC and a diameter of ≥ 7 cm (in order to match the sample sizes between two groups). Cystic renal mass was excluded. The flowchart of patients’ enrollment is shown in Fig. 1 . Data Collection and analysis Demographic data and pathological findings were collected from the medical records of all patients. The estimated glomerular filtration rate (eGFR, ml/min/1.73m 2 ), were calculated by the Cockcroft-Gault formula to assess the renal function( 23 ). CT images within 1 month before intervention were collected. The CT examinations were performed in two centers with the same protocol. CT image data included both unenhanced phase and three standard enhanced phases: corticomedullary phase (CMP), nephrographic phase (NP) and excretory phase (EP). Due to the focus of this study was on the tumor area in CT images, we took UP, CMP and NP into analysis. Image analysis was performed by two physicians with more than 3 years of experience in abdominal imaging. The readers were blind to patient’s inclusion process, clinical and histopathological information. Renal tumor radiological features were evaluated according to the 13 features provided in Table 2 , including the intratumoral arteries, defined as visible arteries of the mass, which were summarized according to previous studies( 13 – 16 ). Image preprocessing The preprocessing steps performed before ML model construction including format conversion, pixel resizing, tumor segmentation, normalization and augmentation. MicroDicom (Version 3.8.1) software was used to convert image files into consecutive joint photographic experts group (JPEG) images. We extracted images at the corresponding tumor levels from the 3-phase CT images of the same patient, while discarding any additional images at the upper or lower region of the tumor. This ensured that we maintained the balance of the 3-phase CT image datasets. Pixel resizing aims to standardize the resolution size of input images. Tumor segmentation with the region of Interest (ROI) outline applied to renal mass regions of the images, removing meaningless backgrounds to improve the model performance. The image data normalization and augmentation such as image flipping were performed to improve data quality and enrich the data sets. Model development and validation In this study, we used residual neural network 34 (ResNet-34) for model development( 24 ). The deep learning (DL) models was trained with the image data sets obtained after image preprocessing. To evaluate the performance of the model across different CT modalities, we input CT images from different dataset either individually or in combination. Hence, 7 models were established: ( 1 ) 3 single-mode models trained with image dataset separately. ( 2 ) 3 bimodal models trained with two randomly combined data sets. ( 3 ) A trimodal model trained with all 3 CT data sets. The study design and basic network structure is showed in Fig. 2 . The neural network in our models was an end-to-end images classifier. Since each patient has multiple consecutive CT slices from whole tumors, we took the mean of all the image-level results from each patient as a patient-level prediction. We trained and validated the DL model according to the 5-fold cross-validation (CV) strategy. The diagnostic performance of DL models was evaluated by Receiver operating characteristic curve (ROC), area under the curve (AUC). Statistical analysis All analyses were performed using SPSS (version 25.0) and R studio (version 2022.02.0 + 443). The Shapiro-Wilk test was used to assessing the normality of the continuous variables. Continuous variables were tested using two independent samples t test and Mann Whitney U test. While categorical variables were compared using Chi-square test or Fisher's exact test. Univariate and multivariate logistic regression was used to filter independent indicators of sarcoma and RCC, and a clinical model was constructed. The maximized Youden Index was used to determine the optimal threshold and calculate sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) from the ROC curve. The difference was considered statistically significant at a two-side P < 0.05. Results Patient characteristics In total, 85 patients were retrospectively included in this study, including 27 renal sarcomas and 58 renal cell carcinomas from the FAH-SYSU and the SYSUCC. The baseline characteristics were shown in Table 1 , the most common histological types of renal sarcoma were liposarcoma (n = 9, 33.3%), followed by leiomyosarcoma (LMS, n = 8, 29.6%). The most common pathological type of RCCs was clear cell renal carcinoma (ccRCC; n = 37, 63.8%), followed by chromophobe renal cell carcinoma (chRCC; n = 11, 18.9%). As detailed in Table 2 , there was no statistical difference in the demographic distribution of renal sarcoma and large RCC patients. Table 1 Baseline characteristics of patients Characteristics Total Age (year, \(\stackrel{-}{x}\pm s\) ) 52.4±13.8 Sex Male 45 (52.9%) Female 40 (47.1%) Laterality Left 39 (45.9%) Right 46 (54.1%) Tumor size (cm) 9.8 (8.0, 12.4) Histologic subtype Renal sarcomas Liposarcoma 9 (33.3%) LMS 8 (29.6%) Dedifferentiated sarcoma 2 (7.4%) Synovial sarcoma 2 (7.4%) Others 6 (22.2%) RCCs ccRCC 37 (63.8%) pRCC 10 (17.3%) chRCC 11 (18.9%) Continuous variables were expressed in median and interquartile range (IQR) or mean ± standard deviations (SD). LMS: leiomyosarcoma; RCC: renal cell carcinoma; ccRCC: clear cell renal cell carcinoma; pRCC: papillary renal cell carcinoma; chRCC: chromophobe renal cell carcinoma. Table 2 Clinical and CT radiological parameters of renal sarcoma and renal cell carcinoma Parameter Renal sarcoma Renal cell carcinoma P value Age (year, \(\stackrel{-}{x}\pm s\) ) 49.4±15.5 53.7±12.9 0.177 Sex 0.284 Female 15 (55.6%) 25 (43.1%) Male 12 (44.4%) 33 (56.9%) WBC(×10 9 /L) 7.6(6.5-9.0) 6.4(5.6–8.4) 0.025 Hb (g/L, \(\stackrel{-}{x}\pm s\) ) 119.3±19.6 125.7±21.6 0.195 PLT (×10 9 /L, \(\stackrel{-}{x}\pm s\) ) 319.1±113.7 282.2±88.1 0.106 eGFR(ml/min/1.73m 2 ) 72.1(65.3–79.4) 79.4(60.8–99.7) 0.235 Laterality 0.222 Right 12 (44.4%) 34 (58.6%) Left 15 (55.6%) 24 (41.4%) Tumor size(cm) 11.5(9.2–15.0) 9.2(7.9–11.7) 0.012 Tumor shape 1.000 Irregular 24 (88.9%) 50 (86.2%) Regular 3 (11.1%) 8 (13.8%) Tumor margins 0.183 Ill-defined 19 (70.4%) 32 (55.2%) Well-defined 8 (29.6%) 26 (44.8%) Enhancement 0.318 No 1 (3.7%) 0 (0.0%) Yes 26 (96.3%) 58 (100%) Intratumoral arteries 0.029 No 20 (74.1%) 28 (48.3%) Yes 7 (25.9%) 30 (51.7%) Tumor necrosis 0.519 No 4 (14.8%) 12 (20.7%) Yes 23 (85.2%) 46 (79.3%) Gerota’s fascia invasion 0.001 No 7 (25.9%) 39 (67.2%) Yes 20 (74.1%) 19 (32.8%) Cystonephrosis 0.801 No 25 (92.6%) 56 (96.6%) Yes 2 (7.4%) 2 (3.4%) Pelvic invasion 0.479 No 18 (66.7%) 34 (58.6%) Yes 9 (33.3%) 24 (41.4%) Perinephric invasion 0.001 No 14 (51.9%) 50 (86.2%) Yes 13 (48.1%) 8 (13.8%) Lymph node metastasis 0.367 No 22 (81.5%) 42 (72.4%) Yes 5 (18.5%) 16 (27.6%) Vascular invasion 0.465 No 20 (74.1%) 47 (81.0%) Yes 7 (25.9%) 11 (19.0%) Continuous variables were expressed in median and interquartile range (IQR) or mean ± standard deviations (SD). WBC: white blood cell count; Hb: hemoglobin; PLT: platelet; eGFR: estimated glomerular filtration rate. Clinical characteristics and radiological features The distribution of preoperative clinical data and radiological features among different groups were shown in Table 2 . Between renal sarcoma and large RCC cohorts, white blood cell count (WBC, P = 0.025), tumor size ( P = 0.012), intratumoral artery ( P = 0.029), Gerota’s fascia invasion ( P = 0.001), perinephric invasion ( P = 0.001) showed significant difference. Development and validation of models According to the multivariate analysis in Fig. 3 A, intratumoral artery (OR = 0.27, 95%CI: 0.08–0.93, P = 0.038) and Gerota’s fascia invasion (OR = 4.47, 95%CI: 1.37–14.57, P = 0.013) were independent associated with renal sarcoma. When intratumoral artery used, the diagnostic model yielded an AUC = 0.63 (95%CI: 0.50–0.75), and another model developed by Gerota’s fascia invasion yielded an AUC = 0.71 (95%CI: 0.59–0.83). While the indicators above were combined to develop a clinical model, the AUC was 0.77 (95%CI: 0.66–0.87). The ROC curve analysis was showed in Fig. 3 B. To test the robustness of the model, we used 5-fold CV, yielding AUC of 0.79±0.11. The cutoff value of the clinical model was 0.28, the sensitivity, specificity, PPV and NPV of the model was 0.74, 0.67, 0.51 and 0.85 respectively. The performance of the 7 different DL models were validated by 5-fold CV, and the results were presented in Table 3 . The scatter plots were utilized to visualize the accuracy (ACC) and AUC distribution in each fold (Fig. 4 ). Overall, the AUC results of 7 models were in the range of 0.90 to 0.95. The overall ACC of the 7 models for differentiating renal sarcoma and RCC were in the range of 0.85 to 0.95. Table 3 Performance of seven deep learning models Models AUC Accuracy Sensitivity Specificity UP 0.95±0.09 0.94±0.07 0.95±0.05 0.93±0.15 CMP 0.94±0.07 0.93±0.06 0.93±0.10 0.93±0.15 NP 0.91±0.13 0.87±0.16 0.82±0.27 0.96±0.09 UP + CMP 0.93±0.09 0.92±0.08 0.93±0.10 0.90±0.22 UP + NP 0.95±0.06 0.94±0.07 0.93±0.08 0.97±0.07 CMP + NP 0.94±0.10 0.94±0.08 0.96±0.08 0.90±0.22 UP + CMP + NP 0.94±0.11 0.93±0.11 0.93±0.10 0.93±0.15 The performance was showed with mean±standard deviations (SD) under 5-fold cross-validation. AUC: area under the receiver operating characteristic curve; UP: unenhanced phase; CMP: corticomedullary phase; NP: nephrographic phase. Regarding single-mode models, The UP model could reach a satisfied diagnostic performance, yielding AUC of 0.95±0.09, with ACC of 0.94±0.07, sensitivity of 0.95±0.05 and specificity of 0.93±0.15. Contrary to our expectations, the models trained by contrast-enhanced CT images did not demonstrate improved classification performance. The CMP and NP model produced AUC of 0.94±0.07 and 0.91±0.13 respectively. There were no significant differences between the mean of three models ( P = 0.796). Regarding bimodal and trimodal models, the UP + NP model yielded nearly AUC = 0.95±0.06, and ACC = 0.94±0.07. Although the mean of the model was not statistically different from UP model ( P = 1.000), the combined data seemed to enhance the stability and robustness of the model. The other three kind of combination did not show satisfied improved effects, with the best AUC of 0.94±0.10 and ACC of 0.94±0.08 (CMP + NP model). Overall, Utilization of enhanced phase CT images or combinations of different CT modalities did not significantly improve the performance. To better understand the classification process and enhance the interpretability of DL model, we visualized the features learned by the model with heat map techniques. The heat map revealed the attention of the model. We chose the UP + CMP model and used the method of gradient class activation mapping (Grad-CAM) to generate heat maps for images input. We found that the values within the tumor were higher than the surrounding area for both renal sarcoma and large RCC, with renal sarcoma having higher values. Examples of typical CT slides and heat maps of renal sarcoma and large RCC are illustrated in Fig. 5 . Discussion Renal sarcoma is a rare malignant tumor, with an incidence rate approximately 1% of all renal tumors( 1 – 3 ). Compared to RCC, renal sarcomas are characterized by high malignancy and poor prognosis, requiring more aggressive surgical intervenes. Therefore, it is vital to differentiate renal sarcoma from renal cell carcinoma, especially the larger diameter RCC, preoperatively. However, there is limited systemic studies on the imaging characteristics and diagnostic methods for renal sarcoma. Hence, a novel method for differential diagnosis of renal sarcoma is needed. In this study, we collected clinical and CT imaging data from 27 renal sarcomas and 58 large RCCs, obtaining a total of 7482 kidney tumor CT images. We analyzed the clinical and imaging data to investigate the differences in clinical factors and radiological features between renal sarcoma and RCC, and construct and validate a diagnostic model for renal sarcoma. The clinical model showed a good performance in differentiating renal sarcoma from large RCC, with an AUC of 0.799. The multimodal DL models further improved the diagnostic performance, with AUC results in range of 0.90–0.95 by 5-fold CV. The results indicated the good discrimination of the DL models. We found that the presence of intratumoral artery and Gerota’s fascia invasion were independent indicators and highly suggestive of renal sarcoma. Absence of intratumoral arteries was associated with renal sarcoma, which may due to the fact that the blood supply of renal sarcomas often comes from extracapsular vessels, which is consistent with the previous studies( 1 ). The rapid growth rate of renal sarcoma causing relatively lack of blood supply, resulting in hypoxia and starve of the tumor cells, may be the reason of neovascularization outside the tumor. The Gerota’s fascia invasion as an indicator showed biological behavior of infiltrating surrounding tissues, which be attributable to highly malignancy of renal sarcoma. Currently, there are no non-invasive clinical or radiomics methods available that can provide similar diagnostic performance in this study. In previous studies, Johannes Uhlig et al. used machine learning to analyze clinical and radiological features to distinguish renal sarcoma from all other renal tumors( 16 ). However, the model is limited due to deficiency of deep-level exploration in CT images and the incomplete inclusion of CT modes. Our study used DL algorithms to explore the roles of different modalities of CT scans and further investigated the integration of multi-modalities of CT images to make the study more comprehensive. The utilization of end-to-end DL algorithm reduce the need for manual feature engineering and allows for more accurate and robust representations of the data. In multimodal CT models, the enhanced CT model did not have higher AUC median than unenhanced model. The reason may be the unstable quality of the contrast enhanced phase in CT examinations, influenced by various factors, such as renal vascular status, renal function, contrast agent concentration speed and difference between institutions. Compared to single-mode models, the combination of multimodal CT images did not significantly improve the diagnostic performance of the models. In previous renal radiomics research, the selection of CT images was often based on a single modality. For example, Gomes et al. found that a single-mode model trained on CMP images had comparable accuracy to models built using images from all phases in the classification task of RCC subtypes( 25 ). Yan et al. also proposed that the CMP was the best choice for model construction in the task of classifying ccRCC and pRCC( 26 ). This study has several limitations. First, this study was exploratory study due to lacking of research on factors related to renal sarcoma, and the results of this study need further validation. Further, the sample size of this study was small, which can lead to overfitting of the models. Therefore, a 5-fold cross-validation was used to reduce the potential overfitting. Finally, the clinical application of the model in this study remained uncertainly, which required further prospective and independent test. In conclusion, this study achieved effective differential diagnosis between renal sarcoma and renal cell carcinoma. Despite the small sample size in this study, the ResNet-34 network models based on multimodal CT images showed strong performance and generalization ability by 5-fold cross-validation. The DL models established in this study might aid in accurate preoperative diagnosis of renal sarcoma and assist in managing the treatment of renal cancer patients. Abbreviations AI artificial intelligence AUC areas under the receiver operating characteristic curve CI confidence interval CMP corticomedullary phase CT computed tomography DL deep learning NP nephrographic phase RCC renal cell carcinoma UP unenhanced phase Declarations Ethics approval and consent to participate The study was reviewed and approved by the Ethics Committee of the First Affiliated Hospital of Sun Yat-sen University. Informed consent to participate was waived by the Institutional Review Board. The study was constructed following ethical guidelines of World Medical Association (WMA) Declaration of Helsinki. Consent for publication Not applicable. Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Competing interests The authors declare that they have no competing interests. Funding This research was supported in part by the National Natural Science Foundation of China (award number: 82373433, 81725016, 81872094, 82272862, 81902576), the National Key Research and Development Program of China (award number: 2016YFC0902600), Guangdong Provincial Department of Finance Project in 2022 (award number: KS0120220267) and the Guangzhou Science and Technology Projects (award number: 202201010910). Authors' contributions H Lin, S Li and Y Chen designed the study, interpreted the literature and completed the writing. Z Chen, J Wei, J Liu, Y Guo, W Chen, H Wang collected and analyzed the data. J Luo, L Tian, H Yao supervised the study and revised the manuscript. Acknowledgements Not applicable. References Srinivas V, Sogani PC, Hajdu SI, Whitmore WJ. Sarcomas of the kidney. J Urol [Journal Article]. 1984 1984-07-01;132(1):13–6. Vogelzang NJ, Fremgen AM, Guinan PD, Chmiel JS, Sylvester JL, Sener SF. Primary renal sarcoma in adults. A natural history and management study by the American Cancer Society, Illinois Division. CANCER-AM CANCER SOC. [Comparative Study; Journal Article]. 1993 1993-02-01;71(3):804–10. Wang X, Xu R, Yan L, Zhuang J, Wei B, Kang D et al. Adult renal sarcoma: clinical features and survival in a series of patients treated at a high-volume institution. Urol [Journal Article]. 2011 2011-04-01;77(4):836–41. Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, Bray F. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021 Feb 4. doi: 10.3322/caac.21660. Epub ahead of print. PMID: 33538338. Siegel RL, Miller KD, Wagle NS, Jemal A. Cancer statistics, 2023. CA-CANCER J CLIN. [Journal Article]. 2023 2023-01-01;73(1):17–48. Huang WC, Elkin EB, Levey AS, Jang TL, Russo P. Partial nephrectomy versus radical nephrectomy in patients with small renal tumors–is there a difference in mortality and cardiovascular outcomes? J UROLOGY. [Comparative Study; Journal Article]. 2009 2009-01-01;181(1):55–61, 61 – 2. Van Poppel H, Da PL, Albrecht W, Matveev V, Bono A, Borkowski A, A prospective, randomised EORTC intergroup phase 3 study comparing the oncologic outcome of elective nephron-sparing surgery and radical nephrectomy for low-stage renal cell carcinoma. EUR UROL. [Clinical Trial, Phase III et al. Comparative Study; Journal Article; Randomized Controlled Trial; Research Support, N.I.H., Extramural; Research Support, Non-U.S. Gov't]. 2011 2011-04-01;59(4):543 – 52. Ljungberg B, Albiges L, Abu-Ghanem Y, Bedke J, Capitanio U, Dabestani S et al. European Association of Urology Guidelines on Renal Cell Carcinoma: The 2022 Update. EUR UROL. [Journal Article; Practice Guideline]. 2022 2022-10-01;82(4):399–410. Alvarenga JC, Ball AB, Fisher C, Fryatt I, Jones L, Thomas JM. Limitations of surgery in the treatment of retroperitoneal sarcoma. BRIT J SURG [Journal Article]. 1991 1991-08-01;78(8):912–6. Bevilacqua RG, Rogatko A, Hajdu SI, Brennan MF. Prognostic factors in primary retroperitoneal soft-tissue sarcomas. Arch Surg. [Journal Article; Research Support, Non-U.S. Gov't; Research Support, U.S. Gov't, P.H.S.]. 1991 1991-03-01;126(3):328–34. Heslin MJ, Lewis JJ, Nadler E, Newman E, Woodruff JM, Casper ES et al. Prognostic factors associated with long-term survival for retroperitoneal sarcoma: implications for management. J CLIN ONCOL. [Journal Article; Research Support, U.S. Gov't, P.H.S.]. 1997 1997-08-01;15(8):2832–9. Roy C, Buy X, Ghali SE. Imaging in Renal Cell Cancer. Eau Update. 2003;1(4):209–14. Lalwani N, Prasad SR, Vikram R, Katabathina V, Shanbhogue A, Restrepo C. Pediatric and adult primary sarcomas of the kidney: a cross-sectional imaging review. ACTA RADIOL [Journal Article]. 2011 2011-05-01;52(4):448–57. Katabathina VS, Vikram R, Nagar AM, Tamboli P, Menias CO, Prasad SR. Mesenchymal neoplasms of the kidney in adults: imaging spectrum with radiologic-pathologic correlation. RADIOGRAPHICS [Journal Article; Review]. 2010 2010-10-01;30(6):1525–40. Karaosmanoğlu AD, Onur MR, Shirkhoda A, Ozmen M, Hahn PF. Unusual Malignant Solid Neoplasms of the Kidney: Cross-Sectional Imaging Findings. KOREAN J RADIOL [Journal Article]. 2015 2015-07-01;16(4):853–9. Uhlig J, Uhlig A, Bachanek S, Onur MR, Kinner S, Geisel D et al. Primary renal sarcomas: imaging features and discrimination from non-sarcoma renal tumors. EUR RADIOL [Journal Article]. 2022 2022-02-01;32(2):981–9. Kumar V, Gu Y, Basu S, Berglund A, Eschrich SA, Schabath MB et al. Radiomics: the process and the challenges. MAGN RESON IMAGING. [Journal Article; Research Support, N.I.H., Extramural; Review]. 2012 2012-11-01;30(9):1234–48. Lambin P, Rios-Velazquez E, Leijenaar R, Carvalho S, van Stiphout RG, Granton P et al. Radiomics: extracting more information from medical images using advanced feature analysis. EUR J CANCER. [Evaluation Study; Journal Article; Research Support, Non-U.S. Gov't; Review]. 2012 2012-03-01;48(4):441–6. Esteva A, Kuprel B, Novoa RA, Ko J, Swetter SM, Blau HM et al. Dermatologist-level classification of skin cancer with deep neural networks. NATURE. [Journal Article; Research Support, N.I.H., Extramural; Research Support, Non-U.S. Gov't; Validation Study]. 2017 2017-02-02;542(7639):115–8. Ting D, Cheung CY, Lim G, Tan G, Quang ND, Gan A et al. Development and Validation of a Deep Learning System for Diabetic Retinopathy and Related Eye Diseases Using Retinal Images From Multiethnic Populations With Diabetes. JAMA-J AM MED ASSOC. [Comparative Study; Journal Article; Validation Study]. 2017 2017-12-12;318(22):2211–23. Wang K, Lu X, Zhou H, Gao Y, Zheng J, Tong M et al. Deep learning Radiomics of shear wave elastography significantly improved diagnostic performance for assessing liver fibrosis in chronic hepatitis B: a prospective multicentre study. GUT. [Journal Article; Multicenter Study; Research Support, Non-U.S. Gov't]. 2019 2019-04-01;68(4):729–41. Moch H, Cubilla AL, Humphrey PA, Reuter VE, Ulbright TM. The 2016 WHO Classification of Tumours of the Urinary System and Male Genital Organs-Part A: Renal, Penile, and Testicular Tumours. EUR UROL. [Journal Article; Review]. 2016 2016-07-01;70(1):93–105. Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. NEPHRON [Journal Article]. 1976 1976-01-19;16(1):31–41. Z KHX, R S. J. S, ^editors. Deep Residual Learning for Image Recognition. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR); 2016 2016-01-01. Pub Place; Year Published. Veloso GF, Matos AP, Palas J, Mascarenhas V, Herédia V, Duarte S et al. Renal cell carcinoma subtype differentiation using single-phase corticomedullary contrast-enhanced CT. CLIN IMAG. [Journal Article]. 2015 2015-03-01;39(2):273–7. Yan L, Liu Z, Wang G, Huang Y, Liu Y, Yu Y et al. Angiomyolipoma with minimal fat: differentiation from clear cell renal cell carcinoma and papillary renal cell carcinoma by texture analysis on CT images. ACAD RADIOL. [Comparative Study; Journal Article; Research Support, Non-U.S. Gov't]. 2015 2015-09-01;22(9):1115–21. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted 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-4478575","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":314626001,"identity":"9f629410-43f0-4027-b116-706e073f5d65","order_by":0,"name":"Haishan Lin","email":"","orcid":"","institution":"First Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Haishan","middleName":"","lastName":"Lin","suffix":""},{"id":314626002,"identity":"e7bcab39-ba1f-4189-822c-2e2301751899","order_by":1,"name":"Shurong Li","email":"","orcid":"","institution":"First Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Shurong","middleName":"","lastName":"Li","suffix":""},{"id":314626003,"identity":"d53a8b8d-bedd-447e-a261-76ae9ce460cf","order_by":2,"name":"Yuhang Chen","email":"","orcid":"","institution":"First Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Yuhang","middleName":"","lastName":"Chen","suffix":""},{"id":314626004,"identity":"d2f8e5f2-e1c6-4235-b22f-cae4b1c78a4f","order_by":3,"name":"Zhenhua Chen","email":"","orcid":"","institution":"First Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Zhenhua","middleName":"","lastName":"Chen","suffix":""},{"id":314626005,"identity":"f1314e59-dbbe-4925-8d6c-d583d639e4a8","order_by":4,"name":"Jinhuan Wei","email":"","orcid":"","institution":"First Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Jinhuan","middleName":"","lastName":"Wei","suffix":""},{"id":314626006,"identity":"39c1260c-a9e1-4a28-b06e-a061309a2baa","order_by":5,"name":"Jiumin Liu","email":"","orcid":"","institution":"Guangdong Provincial People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"Jiumin","middleName":"","lastName":"Liu","suffix":""},{"id":314626007,"identity":"f5a44884-d4c0-4548-a457-7ba1b76f5e38","order_by":6,"name":"Yan Guo","email":"","orcid":"","institution":"First Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Yan","middleName":"","lastName":"Guo","suffix":""},{"id":314626008,"identity":"c6259faf-6c05-444e-811e-d3eb5c42d084","order_by":7,"name":"Wei Chen","email":"","orcid":"","institution":"First Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Wei","middleName":"","lastName":"Chen","suffix":""},{"id":314626009,"identity":"bb390a5c-a69c-46a1-afd0-1aac85610439","order_by":8,"name":"Huanjun Wang","email":"","orcid":"","institution":"First Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Huanjun","middleName":"","lastName":"Wang","suffix":""},{"id":314626010,"identity":"ff9afd48-392c-45ac-89b7-c0269b7b29a0","order_by":9,"name":"Junhang Luo","email":"","orcid":"","institution":"First Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":false,"prefix":"","firstName":"Junhang","middleName":"","lastName":"Luo","suffix":""},{"id":314626011,"identity":"936b2aa1-461a-47e7-8290-f7366e063c42","order_by":10,"name":"Li Tian","email":"","orcid":"","institution":"Sun Yat-sen University Cancer Center","correspondingAuthor":false,"prefix":"","firstName":"Li","middleName":"","lastName":"Tian","suffix":""},{"id":314626012,"identity":"cd8667e3-a96c-4e62-ade9-ee752a740361","order_by":11,"name":"Haohua Yao","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABC0lEQVRIiWNgGAWjYHACNmbGPzYMDMzMjQ9gQhKEtTSkAbUwNhuQouUwkGZsg6vEq8Xg+OFjjwt3nI/mb2dskGD4c9je4ADzwds8DHZ5OLWcSUs3nnnmdu6Mw4wNBoxthxM3HGBLtuZhSC7GqeVAjpk0D9vt3AaglgSgCxMMDvAARRgOJDbg0nL+DUjLudz5QC0HIA7j/4Zfyw2gLbxtB3I3HGZsbGBgO8y44QAPG14tkjeepUnznEnO3XiYsZkhsS09ceZhNmPLOQbJOLXwnU8+Js1TYZc77/zh4z8+/LG25zve/PDGmwo7nFoUDiDzEhiagXEKdjAO9UAgj2ZWHW6lo2AUjIJRMGIBAI5CW5Qyx7wUAAAAAElFTkSuQmCC","orcid":"","institution":"First Affiliated Hospital of Sun Yat-sen University","correspondingAuthor":true,"prefix":"","firstName":"Haohua","middleName":"","lastName":"Yao","suffix":""}],"badges":[],"createdAt":"2024-05-26 04:08:23","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4478575/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4478575/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":59122189,"identity":"808ec398-cf15-4489-a41f-0bd34133150a","added_by":"auto","created_at":"2024-06-26 15:03:19","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":77318,"visible":true,"origin":"","legend":"\u003cp\u003eFlowchart of the patient enrollment. RCC: renal cell carcinoma.\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-4478575/v1/38604910aecb9a1797f78cc2.jpeg"},{"id":59123083,"identity":"ca1245be-5f19-4f17-afda-1b5d70ec20a6","added_by":"auto","created_at":"2024-06-26 15:11:19","extension":"jpeg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":266490,"visible":true,"origin":"","legend":"\u003cp\u003eStudy design and basic structure of the deep learning models. Cov: convolutional layer; fc: full connecting layer.\u003c/p\u003e","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-4478575/v1/0a860161c45efafdf9afd138.jpeg"},{"id":59123082,"identity":"7a2be060-cd20-4c07-99cc-e023864049a5","added_by":"auto","created_at":"2024-06-26 15:11:19","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":144766,"visible":true,"origin":"","legend":"\u003cp\u003eIndicators and diagnostic performance of the clinical model. (A) Multivariate Logistic regression for screening independent indicators of renal sarcoma. (B) Receiver operating characteristics curve (ROC), area under ROC (AUC) and 95% confidence interval (CI). WBC: white blood cell count; OR: odds ratio.\u003c/p\u003e","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-4478575/v1/7d6dc6bc8564bd4da42eab91.jpeg"},{"id":59122190,"identity":"349cea8f-2f2f-46b8-8fcc-9b2c070415d8","added_by":"auto","created_at":"2024-06-26 15:03:19","extension":"jpeg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":61500,"visible":true,"origin":"","legend":"\u003cp\u003ePerformance of deep learning models trained on multimodal CT images based on 5-fold cross-validation. Scatter plot showed the ACC (A) and AUC (B) results in each fold. ACC: accuracy; UP: unenhanced phase; CMP: corticomedullary phase; NP: nephrographic phase.\u003c/p\u003e","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-4478575/v1/67869129943c34aa2963aac1.jpeg"},{"id":59122193,"identity":"8d4df2c7-61d0-4b8c-a928-d250ac01a438","added_by":"auto","created_at":"2024-06-26 15:03:19","extension":"jpeg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":910996,"visible":true,"origin":"","legend":"\u003cp\u003eExamples of typical CT images and corresponding heat maps of renal sarcoma and RCC. RS: renal sarcoma; RCC: renal cell carcinoma; ROI: region of interest.\u003c/p\u003e","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-4478575/v1/636508ec1496706b110c4b90.jpeg"},{"id":73937925,"identity":"d8f82e2d-f4f2-478d-bc05-c38975d615c7","added_by":"auto","created_at":"2025-01-16 07:24:39","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":826164,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4478575/v1/b1172fd9-3bc4-4b6f-ae1e-236447a9c33c.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"CT based deep learning model for differentiating primary renal sarcomas from large renal cell carcinomas","fulltext":[{"header":"Introduction","content":"\u003cp\u003eAdult primary renal sarcomas are rare renal malignancies derived from the mesenchymal tissue of the kidney. In previous studies, the prevalence of renal sarcoma was only 0.8\u0026ndash;1.1%(\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). The epidemiological data of renal sarcomas in China accounted for 0.6\u0026ndash;1.9% of renal tumor cases(\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eWhile, renal cancer is one of the top ten most common cancers worldwide, accounting for 2.2% of all malignant tumors. As the most common solid tumor of the kidney, renal cell carcinoma (RCC) accounts for about 80\u0026ndash;90% of primary malignant tumors of the kidney(\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e). For early-stage patients such as localized and locally advanced RCCs, nephron sparing surgery is the preferred treatment to preserved kidney function better(\u003cspan additionalcitationids=\"CR7\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe prognosis of patients with renal sarcomas is worse than RCCs due to the rapid growth and high malignancy(\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e). The treatment strategy for primary renal sarcoma typically involves radical nephrectomy instead of nephron sparing surgery in order to reduce positive surgical margins(\u003cspan additionalcitationids=\"CR10\" citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). Therefore, accurately diagnosing primary renal sarcoma can assist in preoperative clinical decision making and improve patient prognosis, which is an important issue to be solved in clinical practice. But the preoperative differential diagnosis between primary renal sarcoma and RCC before surgery is difficult, especially for the lager diameter tumors.\u003c/p\u003e \u003cp\u003eIn recent years, abdominal imaging has been widely used in health examinations, and the detective rate and diagnostic accuracy of renal tumors have significantly improved(\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e). Computed tomography (CT) is the cornerstone of imaging method for screening and diagnosis of renal mass. The CT imaging features of renal sarcoma have been described in three reviews(\u003cspan additionalcitationids=\"CR14\" citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e). Based on the imaging features of sarcoma summarized by previous articles, Johannes Uhlig and et al. found that the imaging features associated with renal sarcoma were: tumor laterality (right side), larger maximum diameter, irregular shape, ill-defined margins, vascular invasion, necrosis, and organ invasion(\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e). However, the studies above did not propose standard imaging evaluation method and neglected the differentiation between large RCC and renal sarcoma. Moreover, the high-throughput information of images needs further exploration.\u003c/p\u003e \u003cp\u003eThe application of artificial intelligence and radiomics in medical image analysis has developed rapidly. Radiomics can be used for disease screening, diagnosis, prognosis analysis and decision making(\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e). Many studies have demonstrated that deep learning can be useful for image interpretation tasks and the models showed expert-level performance(\u003cspan additionalcitationids=\"CR20\" citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e). Compared with traditional feature engineering-based method, deep learning radiomics can learn features automatically without precise tumor annotation and distinguish the features adaptive to specific task though self-learning strategy.\u003c/p\u003e \u003cp\u003eTherefore, we aimed to establish a deep learning model to distinguish adult primary renal sarcoma and large RCC preoperatively, which can assist in formulating individualized treatment strategies for adult primary renal sarcoma patients.\u003c/p\u003e"},{"header":"Material and Methods","content":"\u003cp\u003e The study was reviewed and approved by the Ethics Committee of the First Affiliated Hospital of Sun Yat-sen University (FAH-SYSU) (Nr [2022] 155) for retrospective case collection and analysis.\u003c/p\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003ePatient cohort\u003c/h2\u003e \u003cp\u003ePatients pathologically diagnosed with renal sarcoma at FAH-SYSU and Sun Yat-sen University Cancer Center (SYSUCC) from 2009 to 2021 were retrospectively included in this study. The pathological subtypes of renal sarcoma were based on world health organization classification of renal tumors(\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e). Participants were excluded if they had previous history of sarcoma, presented with renal metastasis of non-renal originated sarcomas, with unsatisfied images quality, and if their age less than 18 or more than 80. The control group randomly included from consecutive patients who were initially diagnosed from 2018 to 2020, with a pathological diagnosis of RCC and a diameter of \u0026ge;\u0026thinsp;7 cm (in order to match the sample sizes between two groups). Cystic renal mass was excluded. The flowchart of patients\u0026rsquo; enrollment is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003eData Collection and analysis\u003c/h2\u003e \u003cp\u003eDemographic data and pathological findings were collected from the medical records of all patients. The estimated glomerular filtration rate (eGFR, ml/min/1.73m\u003csup\u003e2\u003c/sup\u003e), were calculated by the Cockcroft-Gault formula to assess the renal function(\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eCT images within 1 month before intervention were collected. The CT examinations were performed in two centers with the same protocol. CT image data included both unenhanced phase and three standard enhanced phases: corticomedullary phase (CMP), nephrographic phase (NP) and excretory phase (EP). Due to the focus of this study was on the tumor area in CT images, we took UP, CMP and NP into analysis. Image analysis was performed by two physicians with more than 3 years of experience in abdominal imaging. The readers were blind to patient\u0026rsquo;s inclusion process, clinical and histopathological information. Renal tumor radiological features were evaluated according to the 13 features provided in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e2\u003c/span\u003e, including the intratumoral arteries, defined as visible arteries of the mass, which were summarized according to previous studies(\u003cspan additionalcitationids=\"CR14 CR15\" citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eImage preprocessing\u003c/h2\u003e \u003cp\u003eThe preprocessing steps performed before ML model construction including format conversion, pixel resizing, tumor segmentation, normalization and augmentation. MicroDicom (Version 3.8.1) software was used to convert image files into consecutive joint photographic experts group (JPEG) images. We extracted images at the corresponding tumor levels from the 3-phase CT images of the same patient, while discarding any additional images at the upper or lower region of the tumor. This ensured that we maintained the balance of the 3-phase CT image datasets. Pixel resizing aims to standardize the resolution size of input images. Tumor segmentation with the region of Interest (ROI) outline applied to renal mass regions of the images, removing meaningless backgrounds to improve the model performance. The image data normalization and augmentation such as image flipping were performed to improve data quality and enrich the data sets.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003eModel development and validation\u003c/h2\u003e \u003cp\u003eIn this study, we used residual neural network 34 (ResNet-34) for model development(\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e). The deep learning (DL) models was trained with the image data sets obtained after image preprocessing. To evaluate the performance of the model across different CT modalities, we input CT images from different dataset either individually or in combination. Hence, 7 models were established: (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) 3 single-mode models trained with image dataset separately. (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) 3 bimodal models trained with two randomly combined data sets. (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) A trimodal model trained with all 3 CT data sets. The study design and basic network structure is showed in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe neural network in our models was an end-to-end images classifier. Since each patient has multiple consecutive CT slices from whole tumors, we took the mean of all the image-level results from each patient as a patient-level prediction. We trained and validated the DL model according to the 5-fold cross-validation (CV) strategy. The diagnostic performance of DL models was evaluated by Receiver operating characteristic curve (ROC), area under the curve (AUC).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003eStatistical analysis\u003c/h2\u003e \u003cp\u003eAll analyses were performed using SPSS (version 25.0) and R studio (version 2022.02.0\u0026thinsp;+\u0026thinsp;443). The Shapiro-Wilk test was used to assessing the normality of the continuous variables. Continuous variables were tested using two independent samples \u003cem\u003et\u003c/em\u003e test and Mann Whitney U test. While categorical variables were compared using Chi-square test or Fisher's exact test. Univariate and multivariate logistic regression was used to filter independent indicators of sarcoma and RCC, and a clinical model was constructed. The maximized Youden Index was used to determine the optimal threshold and calculate sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) from the ROC curve. The difference was considered statistically significant at a two-side \u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003ePatient characteristics\u003c/h2\u003e \u003cp\u003eIn total, 85 patients were retrospectively included in this study, including 27 renal sarcomas and 58 renal cell carcinomas from the FAH-SYSU and the SYSUCC. The baseline characteristics were shown in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e1\u003c/span\u003e, the most common histological types of renal sarcoma were liposarcoma (n\u0026thinsp;=\u0026thinsp;9, 33.3%), followed by leiomyosarcoma (LMS, n\u0026thinsp;=\u0026thinsp;8, 29.6%). The most common pathological type of RCCs was clear cell renal carcinoma (ccRCC; n\u0026thinsp;=\u0026thinsp;37, 63.8%), followed by chromophobe renal cell carcinoma (chRCC; n\u0026thinsp;=\u0026thinsp;11, 18.9%). As detailed in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e2\u003c/span\u003e, there was no statistical difference in the demographic distribution of renal sarcoma and large RCC patients.\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 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eBaseline characteristics of patients\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\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 \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCharacteristics\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTotal\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge (year, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\stackrel{-}{x}\\pm s\\)\u003c/span\u003e\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e52.4\u0026plusmn;13.8\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSex\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e45 (52.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e40 (47.1%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLaterality\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLeft\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e39 (45.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRight\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e46 (54.1%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTumor size (cm)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e9.8 (8.0, 12.4)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHistologic subtype\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRenal sarcomas\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLiposarcoma\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e9 (33.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLMS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e8 (29.6%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDedifferentiated sarcoma\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2 (7.4%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSynovial sarcoma\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2 (7.4%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOthers\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6 (22.2%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRCCs\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eccRCC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e37 (63.8%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003epRCC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e10 (17.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003echRCC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e11 (18.9%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"2\"\u003eContinuous variables were expressed in median and interquartile range (IQR) or mean\u0026thinsp;\u0026plusmn;\u0026thinsp;standard deviations (SD). LMS: leiomyosarcoma; RCC: renal cell carcinoma; ccRCC: clear cell renal cell carcinoma; pRCC: papillary renal cell carcinoma; chRCC: chromophobe renal cell carcinoma.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e\u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eClinical and CT radiological parameters of renal sarcoma and renal cell carcinoma\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\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=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eParameter\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRenal sarcoma\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eRenal cell carcinoma\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003eP\u003c/em\u003e value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge (year, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\stackrel{-}{x}\\pm s\\)\u003c/span\u003e\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e49.4\u0026plusmn;15.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e53.7\u0026plusmn;12.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.177\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSex\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.284\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e15 (55.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e25 (43.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e12 (44.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e33 (56.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWBC(\u0026times;10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7.6(6.5-9.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e6.4(5.6\u0026ndash;8.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u003cb\u003e0.025\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHb (g/L, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\stackrel{-}{x}\\pm s\\)\u003c/span\u003e\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e119.3\u0026plusmn;19.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e125.7\u0026plusmn;21.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.195\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePLT (\u0026times;10\u003csup\u003e9\u003c/sup\u003e/L, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\stackrel{-}{x}\\pm s\\)\u003c/span\u003e\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e319.1\u0026plusmn;113.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e282.2\u0026plusmn;88.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.106\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eeGFR(ml/min/1.73m\u003csup\u003e2\u003c/sup\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e72.1(65.3\u0026ndash;79.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e79.4(60.8\u0026ndash;99.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.235\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLaterality\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.222\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRight\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e12 (44.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e34 (58.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLeft\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e15 (55.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e24 (41.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTumor size(cm)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e11.5(9.2\u0026ndash;15.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e9.2(7.9\u0026ndash;11.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u003cb\u003e0.012\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTumor shape\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIrregular\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e24 (88.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e50 (86.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRegular\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3 (11.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e8 (13.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTumor margins\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.183\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIll-defined\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e19 (70.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e32 (55.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWell-defined\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e8 (29.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e26 (44.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEnhancement\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.318\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1 (3.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0 (0.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e26 (96.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e58 (100%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIntratumoral arteries\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u003cb\u003e0.029\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e20 (74.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e28 (48.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7 (25.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e30 (51.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTumor necrosis\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.519\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4 (14.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e12 (20.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e23 (85.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e46 (79.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGerota\u0026rsquo;s fascia invasion\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u003cb\u003e0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7 (25.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e39 (67.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e20 (74.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e19 (32.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCystonephrosis\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.801\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e25 (92.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e56 (96.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2 (7.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2 (3.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePelvic invasion\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.479\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e18 (66.7%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e34 (58.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e9 (33.3%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e24 (41.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePerinephric invasion\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u003cb\u003e0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e14 (51.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e50 (86.2%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e13 (48.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e8 (13.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLymph node metastasis\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.367\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e22 (81.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e42 (72.4%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5 (18.5%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e16 (27.6%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVascular invasion\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=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.465\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e20 (74.1%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e47 (81.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7 (25.9%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e11 (19.0%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"4\"\u003eContinuous variables were expressed in median and interquartile range (IQR) or mean\u0026thinsp;\u0026plusmn;\u0026thinsp;standard deviations (SD). WBC: white blood cell count; Hb: hemoglobin; PLT: platelet; eGFR: estimated glomerular filtration rate.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003eClinical characteristics and radiological features\u003c/h2\u003e \u003cp\u003eThe distribution of preoperative clinical data and radiological features among different groups were shown in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e2\u003c/span\u003e. Between renal sarcoma and large RCC cohorts, white blood cell count (WBC, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.025), tumor size (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.012), intratumoral artery (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.029), Gerota\u0026rsquo;s fascia invasion (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.001), perinephric invasion (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.001) showed significant difference.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eDevelopment and validation of models\u003c/h2\u003e \u003cp\u003eAccording to the multivariate analysis in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA, intratumoral artery (OR\u0026thinsp;=\u0026thinsp;0.27, 95%CI: 0.08\u0026ndash;0.93, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.038) and Gerota\u0026rsquo;s fascia invasion (OR\u0026thinsp;=\u0026thinsp;4.47, 95%CI: 1.37\u0026ndash;14.57, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.013) were independent associated with renal sarcoma. When intratumoral artery used, the diagnostic model yielded an AUC\u0026thinsp;=\u0026thinsp;0.63 (95%CI: 0.50\u0026ndash;0.75), and another model developed by Gerota\u0026rsquo;s fascia invasion yielded an AUC\u0026thinsp;=\u0026thinsp;0.71 (95%CI: 0.59\u0026ndash;0.83). While the indicators above were combined to develop a clinical model, the AUC was 0.77 (95%CI: 0.66\u0026ndash;0.87). The ROC curve analysis was showed in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB. To test the robustness of the model, we used 5-fold CV, yielding AUC of 0.79\u0026plusmn;0.11. The cutoff value of the clinical model was 0.28, the sensitivity, specificity, PPV and NPV of the model was 0.74, 0.67, 0.51 and 0.85 respectively.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe performance of the 7 different DL models were validated by 5-fold CV, and the results were presented in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. The scatter plots were utilized to visualize the accuracy (ACC) and AUC distribution in each fold (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). Overall, the AUC results of 7 models were in the range of 0.90 to 0.95. The overall ACC of the 7 models for differentiating renal sarcoma and RCC were in the range of 0.85 to 0.95.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePerformance of seven deep learning models\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=\"char\" char=\"\u0026plusmn;\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\"\u0026plusmn;\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\"\u0026plusmn;\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\"\u0026plusmn;\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModels\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAUC\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAccuracy\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSpecificity\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e \u003cp\u003e0.95\u0026plusmn;0.09\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e \u003cp\u003e0.94\u0026plusmn;0.07\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e \u003cp\u003e0.95\u0026plusmn;0.05\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c5\"\u003e \u003cp\u003e0.93\u0026plusmn;0.15\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCMP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e \u003cp\u003e0.94\u0026plusmn;0.07\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e \u003cp\u003e0.93\u0026plusmn;0.06\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e \u003cp\u003e0.93\u0026plusmn;0.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c5\"\u003e \u003cp\u003e0.93\u0026plusmn;0.15\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e \u003cp\u003e0.91\u0026plusmn;0.13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e \u003cp\u003e0.87\u0026plusmn;0.16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e \u003cp\u003e0.82\u0026plusmn;0.27\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c5\"\u003e \u003cp\u003e0.96\u0026plusmn;0.09\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUP\u0026thinsp;+\u0026thinsp;CMP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e \u003cp\u003e0.93\u0026plusmn;0.09\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e \u003cp\u003e0.92\u0026plusmn;0.08\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e \u003cp\u003e0.93\u0026plusmn;0.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c5\"\u003e \u003cp\u003e0.90\u0026plusmn;0.22\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUP\u0026thinsp;+\u0026thinsp;NP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e \u003cp\u003e0.95\u0026plusmn;0.06\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e \u003cp\u003e0.94\u0026plusmn;0.07\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e \u003cp\u003e0.93\u0026plusmn;0.08\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c5\"\u003e \u003cp\u003e0.97\u0026plusmn;0.07\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCMP\u0026thinsp;+\u0026thinsp;NP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e \u003cp\u003e0.94\u0026plusmn;0.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e \u003cp\u003e0.94\u0026plusmn;0.08\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e \u003cp\u003e0.96\u0026plusmn;0.08\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c5\"\u003e \u003cp\u003e0.90\u0026plusmn;0.22\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUP\u0026thinsp;+\u0026thinsp;CMP\u0026thinsp;+\u0026thinsp;NP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c2\"\u003e \u003cp\u003e0.94\u0026plusmn;0.11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c3\"\u003e \u003cp\u003e0.93\u0026plusmn;0.11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c4\"\u003e \u003cp\u003e0.93\u0026plusmn;0.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\"\u0026plusmn;\" colname=\"c5\"\u003e \u003cp\u003e0.93\u0026plusmn;0.15\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003eThe performance was showed with mean\u0026plusmn;standard deviations (SD) under 5-fold cross-validation. AUC: area under the receiver operating characteristic curve; UP: unenhanced phase; CMP: corticomedullary phase; NP: nephrographic phase.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eRegarding single-mode models, The UP model could reach a satisfied diagnostic performance, yielding AUC of 0.95\u0026plusmn;0.09, with ACC of 0.94\u0026plusmn;0.07, sensitivity of 0.95\u0026plusmn;0.05 and specificity of 0.93\u0026plusmn;0.15. Contrary to our expectations, the models trained by contrast-enhanced CT images did not demonstrate improved classification performance. The CMP and NP model produced AUC of 0.94\u0026plusmn;0.07 and 0.91\u0026plusmn;0.13 respectively. There were no significant differences between the mean of three models (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.796).\u003c/p\u003e \u003cp\u003eRegarding bimodal and trimodal models, the UP\u0026thinsp;+\u0026thinsp;NP model yielded nearly AUC\u0026thinsp;=\u0026thinsp;0.95\u0026plusmn;0.06, and ACC\u0026thinsp;=\u0026thinsp;0.94\u0026plusmn;0.07. Although the mean of the model was not statistically different from UP model (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;1.000), the combined data seemed to enhance the stability and robustness of the model. The other three kind of combination did not show satisfied improved effects, with the best AUC of 0.94\u0026plusmn;0.10 and ACC of 0.94\u0026plusmn;0.08 (CMP\u0026thinsp;+\u0026thinsp;NP model). Overall, Utilization of enhanced phase CT images or combinations of different CT modalities did not significantly improve the performance.\u003c/p\u003e \u003cp\u003eTo better understand the classification process and enhance the interpretability of DL model, we visualized the features learned by the model with heat map techniques. The heat map revealed the attention of the model. We chose the UP\u0026thinsp;+\u0026thinsp;CMP model and used the method of gradient class activation mapping (Grad-CAM) to generate heat maps for images input. We found that the values within the tumor were higher than the surrounding area for both renal sarcoma and large RCC, with renal sarcoma having higher values. Examples of typical CT slides and heat maps of renal sarcoma and large RCC are illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eRenal sarcoma is a rare malignant tumor, with an incidence rate approximately 1% of all renal tumors(\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e). Compared to RCC, renal sarcomas are characterized by high malignancy and poor prognosis, requiring more aggressive surgical intervenes. Therefore, it is vital to differentiate renal sarcoma from renal cell carcinoma, especially the larger diameter RCC, preoperatively. However, there is limited systemic studies on the imaging characteristics and diagnostic methods for renal sarcoma. Hence, a novel method for differential diagnosis of renal sarcoma is needed.\u003c/p\u003e \u003cp\u003eIn this study, we collected clinical and CT imaging data from 27 renal sarcomas and 58 large RCCs, obtaining a total of 7482 kidney tumor CT images. We analyzed the clinical and imaging data to investigate the differences in clinical factors and radiological features between renal sarcoma and RCC, and construct and validate a diagnostic model for renal sarcoma. The clinical model showed a good performance in differentiating renal sarcoma from large RCC, with an AUC of 0.799. The multimodal DL models further improved the diagnostic performance, with AUC results in range of 0.90\u0026ndash;0.95 by 5-fold CV. The results indicated the good discrimination of the DL models.\u003c/p\u003e \u003cp\u003eWe found that the presence of intratumoral artery and Gerota\u0026rsquo;s fascia invasion were independent indicators and highly suggestive of renal sarcoma. Absence of intratumoral arteries was associated with renal sarcoma, which may due to the fact that the blood supply of renal sarcomas often comes from extracapsular vessels, which is consistent with the previous studies(\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e). The rapid growth rate of renal sarcoma causing relatively lack of blood supply, resulting in hypoxia and starve of the tumor cells, may be the reason of neovascularization outside the tumor. The Gerota\u0026rsquo;s fascia invasion as an indicator showed biological behavior of infiltrating surrounding tissues, which be attributable to highly malignancy of renal sarcoma.\u003c/p\u003e \u003cp\u003eCurrently, there are no non-invasive clinical or radiomics methods available that can provide similar diagnostic performance in this study. In previous studies, Johannes Uhlig et al. used machine learning to analyze clinical and radiological features to distinguish renal sarcoma from all other renal tumors(\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e). However, the model is limited due to deficiency of deep-level exploration in CT images and the incomplete inclusion of CT modes. Our study used DL algorithms to explore the roles of different modalities of CT scans and further investigated the integration of multi-modalities of CT images to make the study more comprehensive. The utilization of end-to-end DL algorithm reduce the need for manual feature engineering and allows for more accurate and robust representations of the data.\u003c/p\u003e \u003cp\u003eIn multimodal CT models, the enhanced CT model did not have higher AUC median than unenhanced model. The reason may be the unstable quality of the contrast enhanced phase in CT examinations, influenced by various factors, such as renal vascular status, renal function, contrast agent concentration speed and difference between institutions. Compared to single-mode models, the combination of multimodal CT images did not significantly improve the diagnostic performance of the models. In previous renal radiomics research, the selection of CT images was often based on a single modality. For example, Gomes et al. found that a single-mode model trained on CMP images had comparable accuracy to models built using images from all phases in the classification task of RCC subtypes(\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e). Yan et al. also proposed that the CMP was the best choice for model construction in the task of classifying ccRCC and pRCC(\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThis study has several limitations. First, this study was exploratory study due to lacking of research on factors related to renal sarcoma, and the results of this study need further validation. Further, the sample size of this study was small, which can lead to overfitting of the models. Therefore, a 5-fold cross-validation was used to reduce the potential overfitting. Finally, the clinical application of the model in this study remained uncertainly, which required further prospective and independent test.\u003c/p\u003e \u003cp\u003eIn conclusion, this study achieved effective differential diagnosis between renal sarcoma and renal cell carcinoma. Despite the small sample size in this study, the ResNet-34 network models based on multimodal CT images showed strong performance and generalization ability by 5-fold cross-validation. The DL models established in this study might aid in accurate preoperative diagnosis of renal sarcoma and assist in managing the treatment of renal cancer patients.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eAI\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eartificial intelligence\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eAUC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eareas under the receiver operating characteristic curve\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eCI\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003econfidence interval\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eCMP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ecorticomedullary phase\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eCT\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ecomputed tomography\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDL\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003edeep learning\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eNP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003enephrographic phase\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eRCC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003erenal cell carcinoma\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eUP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eunenhanced phase\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study was reviewed and approved by the Ethics Committee of the First Affiliated Hospital of Sun Yat-sen University. Informed consent to participate was waived by the Institutional Review Board. The study was constructed following ethical guidelines of World Medical Association (WMA) Declaration of Helsinki.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis research was supported in part by the National Natural Science Foundation of China (award number: 82373433, 81725016, 81872094, 82272862, 81902576), the National Key Research and Development Program of China (award number: 2016YFC0902600), Guangdong Provincial Department of Finance Project in 2022 (award number: KS0120220267) and the Guangzhou Science and Technology Projects (award number: 202201010910).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026apos; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eH Lin, S Li and Y Chen designed the study, interpreted the literature and completed the writing. Z Chen, J Wei, J Liu, Y Guo, W Chen, H Wang collected and analyzed the data. J Luo, L Tian, H Yao supervised the study and revised the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eSrinivas V, Sogani PC, Hajdu SI, Whitmore WJ. Sarcomas of the kidney. J Urol [Journal Article]. 1984 1984-07-01;132(1):13\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVogelzang NJ, Fremgen AM, Guinan PD, Chmiel JS, Sylvester JL, Sener SF. Primary renal sarcoma in adults. A natural history and management study by the American Cancer Society, Illinois Division. CANCER-AM CANCER SOC. [Comparative Study; Journal Article]. 1993 1993-02-01;71(3):804\u0026ndash;10.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang X, Xu R, Yan L, Zhuang J, Wei B, Kang D et al. Adult renal sarcoma: clinical features and survival in a series of patients treated at a high-volume institution. Urol [Journal Article]. 2011 2011-04-01;77(4):836\u0026ndash;41.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, Bray F. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021 Feb 4. doi: 10.3322/caac.21660. Epub ahead of print. PMID: 33538338.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSiegel RL, Miller KD, Wagle NS, Jemal A. Cancer statistics, 2023. CA-CANCER J CLIN. [Journal Article]. 2023 2023-01-01;73(1):17\u0026ndash;48.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang WC, Elkin EB, Levey AS, Jang TL, Russo P. Partial nephrectomy versus radical nephrectomy in patients with small renal tumors\u0026ndash;is there a difference in mortality and cardiovascular outcomes? J UROLOGY. [Comparative Study; Journal Article]. 2009 2009-01-01;181(1):55\u0026ndash;61, 61\u0026thinsp;\u0026ndash;\u0026thinsp;2.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVan Poppel H, Da PL, Albrecht W, Matveev V, Bono A, Borkowski A, A prospective, randomised EORTC intergroup phase 3 study comparing the oncologic outcome of elective nephron-sparing surgery and radical nephrectomy for low-stage renal cell carcinoma. EUR UROL. [Clinical Trial, Phase III et al. Comparative Study; Journal Article; Randomized Controlled Trial; Research Support, N.I.H., Extramural; Research Support, Non-U.S. Gov't]. 2011 2011-04-01;59(4):543\u0026thinsp;\u0026ndash;\u0026thinsp;52.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLjungberg B, Albiges L, Abu-Ghanem Y, Bedke J, Capitanio U, Dabestani S et al. European Association of Urology Guidelines on Renal Cell Carcinoma: The 2022 Update. EUR UROL. [Journal Article; Practice Guideline]. 2022 2022-10-01;82(4):399\u0026ndash;410.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlvarenga JC, Ball AB, Fisher C, Fryatt I, Jones L, Thomas JM. Limitations of surgery in the treatment of retroperitoneal sarcoma. BRIT J SURG [Journal Article]. 1991 1991-08-01;78(8):912\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBevilacqua RG, Rogatko A, Hajdu SI, Brennan MF. Prognostic factors in primary retroperitoneal soft-tissue sarcomas. Arch Surg. [Journal Article; Research Support, Non-U.S. Gov't; Research Support, U.S. Gov't, P.H.S.]. 1991 1991-03-01;126(3):328\u0026ndash;34.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHeslin MJ, Lewis JJ, Nadler E, Newman E, Woodruff JM, Casper ES et al. Prognostic factors associated with long-term survival for retroperitoneal sarcoma: implications for management. J CLIN ONCOL. [Journal Article; Research Support, U.S. Gov't, P.H.S.]. 1997 1997-08-01;15(8):2832\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRoy C, Buy X, Ghali SE. Imaging in Renal Cell Cancer. Eau Update. 2003;1(4):209\u0026ndash;14.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLalwani N, Prasad SR, Vikram R, Katabathina V, Shanbhogue A, Restrepo C. Pediatric and adult primary sarcomas of the kidney: a cross-sectional imaging review. ACTA RADIOL [Journal Article]. 2011 2011-05-01;52(4):448\u0026ndash;57.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKatabathina VS, Vikram R, Nagar AM, Tamboli P, Menias CO, Prasad SR. Mesenchymal neoplasms of the kidney in adults: imaging spectrum with radiologic-pathologic correlation. RADIOGRAPHICS [Journal Article; Review]. 2010 2010-10-01;30(6):1525\u0026ndash;40.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKaraosmanoğlu AD, Onur MR, Shirkhoda A, Ozmen M, Hahn PF. Unusual Malignant Solid Neoplasms of the Kidney: Cross-Sectional Imaging Findings. KOREAN J RADIOL [Journal Article]. 2015 2015-07-01;16(4):853\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUhlig J, Uhlig A, Bachanek S, Onur MR, Kinner S, Geisel D et al. Primary renal sarcomas: imaging features and discrimination from non-sarcoma renal tumors. EUR RADIOL [Journal Article]. 2022 2022-02-01;32(2):981\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKumar V, Gu Y, Basu S, Berglund A, Eschrich SA, Schabath MB et al. Radiomics: the process and the challenges. MAGN RESON IMAGING. [Journal Article; Research Support, N.I.H., Extramural; Review]. 2012 2012-11-01;30(9):1234\u0026ndash;48.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLambin P, Rios-Velazquez E, Leijenaar R, Carvalho S, van Stiphout RG, Granton P et al. Radiomics: extracting more information from medical images using advanced feature analysis. EUR J CANCER. [Evaluation Study; Journal Article; Research Support, Non-U.S. Gov't; Review]. 2012 2012-03-01;48(4):441\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEsteva A, Kuprel B, Novoa RA, Ko J, Swetter SM, Blau HM et al. Dermatologist-level classification of skin cancer with deep neural networks. NATURE. [Journal Article; Research Support, N.I.H., Extramural; Research Support, Non-U.S. Gov't; Validation Study]. 2017 2017-02-02;542(7639):115\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTing D, Cheung CY, Lim G, Tan G, Quang ND, Gan A et al. Development and Validation of a Deep Learning System for Diabetic Retinopathy and Related Eye Diseases Using Retinal Images From Multiethnic Populations With Diabetes. JAMA-J AM MED ASSOC. [Comparative Study; Journal Article; Validation Study]. 2017 2017-12-12;318(22):2211\u0026ndash;23.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang K, Lu X, Zhou H, Gao Y, Zheng J, Tong M et al. Deep learning Radiomics of shear wave elastography significantly improved diagnostic performance for assessing liver fibrosis in chronic hepatitis B: a prospective multicentre study. GUT. [Journal Article; Multicenter Study; Research Support, Non-U.S. Gov't]. 2019 2019-04-01;68(4):729\u0026ndash;41.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMoch H, Cubilla AL, Humphrey PA, Reuter VE, Ulbright TM. The 2016 WHO Classification of Tumours of the Urinary System and Male Genital Organs-Part A: Renal, Penile, and Testicular Tumours. EUR UROL. [Journal Article; Review]. 2016 2016-07-01;70(1):93\u0026ndash;105.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. NEPHRON [Journal Article]. 1976 1976-01-19;16(1):31\u0026ndash;41.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZ KHX, R S. J. S, ^editors. Deep Residual Learning for Image Recognition. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR); 2016 2016-01-01. Pub Place; Year Published.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVeloso GF, Matos AP, Palas J, Mascarenhas V, Her\u0026eacute;dia V, Duarte S et al. Renal cell carcinoma subtype differentiation using single-phase corticomedullary contrast-enhanced CT. CLIN IMAG. [Journal Article]. 2015 2015-03-01;39(2):273\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYan L, Liu Z, Wang G, Huang Y, Liu Y, Yu Y et al. Angiomyolipoma with minimal fat: differentiation from clear cell renal cell carcinoma and papillary renal cell carcinoma by texture analysis on CT images. ACAD RADIOL. [Comparative Study; Journal Article; Research Support, Non-U.S. Gov't]. 2015 2015-09-01;22(9):1115\u0026ndash;21.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Renal sarcoma, Renal cell carcinoma, Computed tomography, Deep learning","lastPublishedDoi":"10.21203/rs.3.rs-4478575/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4478575/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eDue to the prognosis and treatment differences between primary renal sarcomas and renal cell carcinoma, preoperative differentiation between them is important but challenging. This study aims to explore and develop a diagnostic method based on computed tomography (CT) and clinical data for preoperatively differentiating primary renal sarcomas from large renal cell carcinomas.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003ePatients pathologically diagnosed with primary renal sarcoma from two center between 2009\u0026ndash;2021 were retrospectively included, and large renal cell carcinomas were probably 2:1 compared to renal sarcomas as the control group. Clinical data, standard contrast-enhanced CT images and histological findings were obtained. A clinical model was established with independent indicators based on logistic regression analysis. The region of interest was outlined in each three modal CT images (unenhanced phase [UP], corticomedullary phase [CMP] and nephrographic phase [NP]) and formed 7 modal imaging datasets for deep learning models\u0026rsquo; development. Reported performance metrics included accuracy and areas under the receiver operating characteristic curves (AUC).\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eTotally, 27 renal sarcomas and 58 large RCCs were enrolled. Multivariate logistic regression showed that the independent indicators of renal sarcoma were intratumoral artery and Gerota\u0026rsquo;s fascia invasion (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05). The AUC of clinical model was 0.77 (95% confidence interval [CI]: 0.67\u0026ndash;0.87), sensitivity 0.74, specificity 0.67, positive predictive value 0.51, and negative predictive value 0.85. The deep learning models yielded effective discrimination. The unenhanced phase model yielded AUC\u0026thinsp;=\u0026thinsp;0.95\u0026plusmn;0.09 and accuracy (ACC)\u0026thinsp;=\u0026thinsp;0.94\u0026plusmn;0.07, and UP\u0026thinsp;+\u0026thinsp;NP model yielded nearly AUC\u0026thinsp;=\u0026thinsp;0.95\u0026plusmn;0.06, and ACC\u0026thinsp;=\u0026thinsp;0.94\u0026plusmn;0.07.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eThe deep learning models based on multimodal CT images show good performance for differentiating renal sarcomas from large renal cell carcinomas, which assist in individualized management.\u003c/p\u003e","manuscriptTitle":"CT based deep learning model for differentiating primary renal sarcomas from large renal cell carcinomas","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-06-26 15:03:14","doi":"10.21203/rs.3.rs-4478575/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"bfc2ecb4-4222-4e46-b9c6-3753845ae0d1","owner":[],"postedDate":"June 26th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-01-16T07:24:14+00:00","versionOfRecord":[],"versionCreatedAt":"2024-06-26 15:03:14","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4478575","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4478575","identity":"rs-4478575","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}