Innovative AI Model for Bladder Cancer Diagnosis

Innovative AI Model for Bladder Cancer Diagnosis | 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 Innovative AI Model for Bladder Cancer Diagnosis Lei Jiang, Wenyu Ge, Ruijiao Feng, Ji Liu, Jingru Huo, Shijie Li, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7760584/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 11 You are reading this latest preprint version Abstract Objectives Bladder cancer is one of the most common malignancies of the urinary system, and early diagnosis and treatment are crucial for improving patient prognosis. Methods In this study, we developed an artificial intelligence (AI) model for bladder cancer diagnosis using CT imaging data from our hospital and The Cancer Imaging Archive (TCIA). The model was trained and validated using a large dataset of CT images, and its diagnostic accuracy was assessed through various performance metrics. The AI model was constructed using deep learning techniques, which can automatically learn and extract features from CT images. Retrospective CT data from our hospital and TCIA were used to train the AI model. Performance metrics were evaluated, and the Grad-CAM results were reviewed by radiologists. Results The test - set accuracy exceeded 0.90, and Grad - CAM correctly highlighted tumors. Conclusions The AI model offers accurate and transparent bladder - cancer detection on routine CT scans and merits prospective validation. artificial intelligence bladder cancer diagnosis Figures Figure 1 Figure 2 Figure 3 Figure 4 1. Introduction Bladder cancer, a prevalent malignancy within the urinary system, poses significant threats to human health [ 1 , 2 ] . Early and accurate diagnosis is paramount for improving patient prognosis and reducing mortality rates. Computed Tomography (CT) imaging has emerged as a crucial tool in the detection and diagnosis of bladder cancer. It offers detailed anatomical information and can visualize the size, location, and extent of tumors. However, interpreting CT images relies heavily on the expertise and experience of radiologists, and subtle lesions or complex cases may lead to variability in diagnostic outcomes. While artificial intelligence (AI) has demonstrated potential in bladder cancer research, this field remains in its early stages, particularly concerning the interpretability of AI applications [ 3 , 4 ] . Recent studies have examined AI's role in various aspects of bladder cancer management, such as diagnosis, treatment response evaluation, and prognosis prediction [ 3 – 16 ] . For example, the use of AI-generated targets and inter-observer variation in online adaptive radiotherapy for bladder cancer was investigated by Åström LM et al, underscoring the challenges and opportunities in this domain [ 3 ] . Additionally, Baressi Šegota S et al. proposed a transfer learning method for the semantic segmentation of urinary bladder cancer masses in CT images, illustrating AI's potential to enhance diagnostic precision [ 4 ] . Furthermore, several investigations have centered on the application of radiomics and deep learning in bladder cancer [ 6 , 11 ] . A CT image-based radiomic model was developed by Cao Y et al. to detect PD-L1 expression status in bladder cancer patients, which may inform personalized treatment approaches [ 5 ] . The assessment of bladder cancer treatment response in CT using radiomics combined with deep learning was conducted by Cha KH et al., demonstrating AI's potential in monitoring treatment effectiveness [ 6 ] . Moreover, machine learning was applied by Garapati SS et al. to stage urinary bladder cancer in CT urography, showcasing AI's capacity to aid clinical decision-making [ 8 ] .Other research has focused on AI's utility in predicting prognosis and detecting metastasis [ 10 , 12 ] . A machine learning-based combination of criteria was proposed by Girard A et al. to detect bladder cancer lymph node metastasis, which may improve metastasis detection accuracy [ 8 ] . A CT-based deep learning model was developed by Wei Z et al. to predict overall survival in muscle-invasive bladder cancer patients post-radical cystectomy, emphasizing AI's prognostic value [ 10 ] . Additionally, Xiong S et al. demonstrated that machine learning-based CT radiomics improves bladder cancer staging predictions, suggesting AI's potential to enhance staging accuracy and treatment planning [ 11 ] . AI has also been employed to distinguish between muscle-invasive and non-muscle-invasive bladder cancer. It was shown by Yang Y et al. that deep learning can serve as a noninvasive tool to differentiate these cancer types using CT, potentially reducing the need for invasive procedures [ 12 ] . Automated AI-based analysis of skeletal muscle volume was utilized by Ying T et al. (2021) to predict overall survival after cystectomy for urinary bladder cancer, highlighting AI's potential in identifying prognostic indicators [ 13 ] . Although AI research in bladder cancer is still in its nascent stages, especially regarding interpretability. As the field progresses, further research is required to address interpretability challenges and validate AI applications' clinical utility in bladder cancer. Our study aims to develop an AI model for bladder cancer diagnosis using CT imaging data from our hospital and The Cancer Imaging Archive (TCIA) [ 17 ] . By leveraging deep learning techniques and incorporating Grad-CAM for region of interest identification [ 18 – 23 ] , we strive to enhance diagnostic accuracy and provide a reliable auxiliary tool for clinicians. Specifically, recent advancements in explainable AI (XAI) have demonstrated the potential to improve diagnostic transparency and clinical trust. For instance, Chen et al. [ 18 ] combined deep learning with Grad-CAM and radiomics to automatically localize and diagnose architectural distortion on digital breast tomosynthesis (DBT), achieving high diagnostic accuracy while providing visual explanations for clinical decision-making. Similarly, Fanizzi et al. [ 19 ] developed an explainable prediction model for human papillomavirus (HPV) status in oropharyngeal squamous cell carcinoma patients using CNN on CT images, which not only predicted HPV status with high accuracy but also highlighted critical image features contributing to the predictions. In the context of breast cancer, Huang et al. [ 20 ] proposed an explainable AI model for predicting molecular expressions using multi-task deep learning based on 3D whole breast ultrasound, which provided insights into the biological underpinnings of imaging features. Shah et al. [ 21 ] integrated local and global attention mechanisms to enhance oral cancer detection, with explainability maps helping clinicians understand the model's focus during diagnosis. Furthermore, Talaat et al. [ 22 ] utilized Grad-CAM with a 3D Inception-ResNet V2 for breast cancer classification on mammograms, empowering radiologists with explainable insights into the model's decision-making process. Lastly, Xia et al. [ 23 ] enhanced cancer classification using Raman spectroscopy and CNN, with visualization of critical features aiding in the interpretation of spectral data. Although Grad-CAM has demonstrated intriguing potential for enhancing the interpretability of AI models in many cancers, its application and efficacy in bladder cancer remain to be more thoroughly investigated [ 18 – 23 ] . While there have been notable advancements in the application of AI to bladder cancer diagnosis, our understanding of how Grad-CAM can be effectively utilized in this specific context is less. This gap in research might, in part, reflect the inherent complexities associated with bladder cancer imaging, as well as the fact that AI applications in this area are still evolving compared to more established fields. By leveraging deep learning techniques and incorporating Grad-CAM for region of interest identification, we strive to enhance diagnostic accuracy and provide a reliable auxiliary tool for clinicians. 2. Materials and Methods The study utilized a combination of our hospital data and publicly available datasets from The Cancer Imaging Archive (TCIA) to compile a comprehensive dataset of CT images for bladder cancer diagnosis. Patients diagnosed with bladder cancer and those confirmed to be free of the disease were included. The CT images were acquired using standardized protocols, ensuring consistency in image quality and resolution. Each image underwent preliminary inspection and annotation by experienced radiologists to identify regions of interest and potential artifacts. Data preprocessing steps included noise reduction, image enhancement, and normalization to standardize the input for the AI model. The flowchart illustrating the clinical data collection and modeling process was presented in Fig. 1 . Model Construction A deep learning-based AI model was developed using convolutional neural networks (CNNs), which are particularly effective for image recognition tasks [ 24 – 28 ] . The model architecture was designed to process CT images and extract relevant features automatically. Transfer learning techniques were employed, leveraging pre-trained models on large-scale image datasets to accelerate training and improve performance. The training process involved feeding the preprocessed CT images into the model, with bladder cancer diagnoses serving as labels. The model learned to associate specific imaging patterns with diagnostic outcomes through multiple iterations. To prevent overfitting, regularization techniques such as dropout layers and data augmentation were applied. Grad-CAM (Gradient-weighted Class Activation Mapping) was integrated into the model to identify regions of interest within the CT images that are most relevant to bladder cancer diagnosis. This technique helps visualize the model's decision-making process, highlighting areas that contribute significantly to the diagnostic prediction. By focusing on these regions, the model's accuracy and reliability were enhanced. Statistical Methods The performance of the AI model was evaluated using a variety of statistical metrics. Accuracy and Intersection over Union (IoU) were calculated to assess the model's diagnostic capability. A confusion matrix was generated to provide a detailed breakdown of true positives, true negatives, false positives, and false negatives in percentages. To ensure the robustness of the results, cross-validation techniques were applied. The dataset was divided into training, validation, and test sets, with the validation set used to tune hyperparameters and the test set reserved for final evaluation. 3. Results 3.1 Quantitative Results Figure 2 was presented to show the entire process from data collection to AI model building, explainable analysis, and further application. In the data collection phase, multi-source data, including patient basic information, medical records, and clinical examination results, was collected and strictly filtered and pre-processed to ensure quality and usability. Next, in the AI model building phase, machine-learning algorithms were employed to train the processed data, and models with high predictive and diagnostic power were obtained. The explainable analysis part involved a deep investigation into the model's internal structure and decision-making mechanism, with visualization techniques used to clarify the complex model's operation and enhance transparency and credibility. Finally, in the further application stage, the model was widely applied in clinical practice, and doctors were assisted in disease diagnosis, treatment planning, and patient prognosis assessment. 3.2 Analysis of Results Based on the results from Table 1 and Table 2 , we analyzed the AI model's performance in terms of IoU and accuracy. In terms of IoU, the background class had shown consistent performance, while the bladder and bladder cancer classes exhibited slight variations between the two datasets. This may suggest that the model encounters certain challenges when identifying the bladder and bladder cancer, or that the validation set contains more challenging samples. As for accuracy, the background class still maintains a very high level, and the accuracy of the bladder and bladder cancer classes has increased in the test set. This further confirms the potential issues the model faces when processing these two classes. Table 1 Performance Evaluation of the Validation Set Class IoU Acc Background 99.63 99.79 Bladder 91.25 98.21 Bladder cancer 84.53 91.44 Table 2 Performance Evaluation of the Test Set Class IoU Acc Background 99.58 99.8 Bladder 92.04 98.65 Bladder cancer 84.61 90.18 3.3 Confusion Matrices for Validation and Test Sets in Model Evaluation Figure 3 clearly showed the confusion matrices for the validation and test sets. During the model evaluation, these two confusion matrices offered rich data, helping us fully understand the model's performance. As could be seen from the figure, each confusion matrix contained a comparison between the model's predicted results and the actual labels. 3.4 Visualization of Bladder Tumor Labeling and AI-Identified Locations via Grad-CAM In Fig. 4 A, the labeling of bladder tumors was clearly displayed, with different colors and markers indicating various tumor types and stages. The labels were meticulously placed, ensuring accurate correspondence to the tumor regions. In Fig. 4 B, the Grad-CAM AI model's identified tumor locations were presented. The heat maps generated by the AI model highlighted areas of interest with varying intensity levels, reflecting the model's confidence in its predictions. These visualizations were offered, giving valuable insights into the AI's decision-making process and its alignment with actual tumor regions. 4. Discussion Bladder cancer remains a prevalent malignancy affecting the urinary system, with a rising incidence globally [ 1 ] . This disease significantly impacts patients' quality of life and poses considerable economic burdens due to high treatment costs [ 1 , 2 ] . Traditional diagnostic approaches, such as cystoscopy and biopsy, are invasive and time-consuming, often leading to patient discomfort and potential complications [ 29 , 30 ] . Consequently, the need for more efficient and non-invasive diagnostic methods is urgent to enhance early detection and improve patient outcomes [ 31 , 32 ] . Advances in imaging technologies and AI present promising avenues for the development of innovative diagnostic tools that can facilitate earlier diagnosis and better management of bladder cancer [ 33 – 36 ] . This study focuses on leveraging deep learning algorithms to construct an artificial intelligence diagnostic model based on CT imaging data for bladder cancer. By extracting and analyzing imaging features, the model aims to improve diagnostic accuracy and consistency, addressing the limitations of traditional methods. The findings of this research provide a foundation for the model's potential application in clinical settings, enhancing early diagnostic capabilities and offering decision support for healthcare professionals. The discussion will delve into the implications of the primary results, emphasizing the importance of AI in refining bladder cancer diagnostics and the need for further validation in diverse clinical populations. This study highlights the significant advancements made in the application of deep learning techniques for the diagnosis of bladder cancer through CT imaging. The extraction of radiomic features from CT scans allows for a complicated understanding of tumor characteristics that traditional imaging methods may overlook. This underscores the potential of artificial intelligence to not only enhance diagnostic precision but also provide a reliable framework for clinical decision-making in bladder cancer management. The findings align with previous studies that have shown similar efficacy of deep learning algorithms in other cancer types, suggesting that AI can be a transformative tool in oncology. Moreover, the study's robust design, which included a diverse dataset for training and validation, strengthens its findings. The high sensitivity and specificity reported for the model indicate its practical applicability in clinical settings, potentially leading to earlier detection of bladder cancer and improved patient outcomes. This is particularly crucial given the often asymptomatic nature of early-stage bladder cancer, which can lead to delayed diagnoses and poorer prognoses. The integration of deep learning in identifying critical features such as tumor texture and shape further enhances the model's capability to predict malignancy. Future research should focus on refining these algorithms and expanding their validation across larger, multi-center cohorts to ensure their generalizability and effectiveness in real-world clinical scenarios. The implications of these findings extend beyond mere diagnostic accuracy; they pave the way for personalized medicine approaches in bladder cancer treatment. By incorporating AI-driven insights into the tumor's biological behavior, clinicians can tailor therapeutic strategies that align more closely with individual patient profiles. For instance, the association between specific imaging features and molecular characteristics of tumors could inform decisions regarding the use of targeted therapies or immunotherapies, thus optimizing treatment outcomes and minimizing unnecessary interventions. In conclusion, this study not only demonstrates the feasibility of using deep learning for bladder cancer diagnosis but also emphasizes the need for ongoing research to harness the full potential of AI in enhancing cancer care. The combination of imaging data with advanced analytical techniques represents a promising frontier in the quest for more effective, precision-driven oncology. The limitations of this study primarily stem from the relatively small sample size, which may impact the generalizability of the developed model. A larger cohort is essential to validate the model's performance across diverse populations and clinical settings. Additionally, the lack of extensive clinical validation data poses challenges in confirming the model's robustness and applicability in real-world scenarios. Variability in imaging protocols and equipment among different institutions could introduce inter-institutional differences, further complicating the model's reliability. Addressing these limitations through multi-center collaborations and larger, more diverse datasets will be crucial for enhancing the model's predictive power and ensuring its clinical utility. In conclusion, this research successfully developed an artificial intelligence diagnostic model based on CT imaging for bladder cancer, demonstrating its potential for improving diagnostic accuracy and prognostic assessment. The integration of radiomics features into clinical practice can provide valuable insights for personalized treatment strategies. Future efforts should focus on expanding the sample size and conducting rigorous clinical validation to establish the model's effectiveness in early diagnosis and tailored therapies for bladder cancer patients. The findings pave the way for advancements in non-invasive diagnostic methodologies that could significantly impact patient outcomes. Limitations Model Limitations Although the deep learning model in this study shows promise, it has several limitations. Firstly, despite high accuracy on training data, it may overfit and lack generalizability to diverse populations and imaging conditions due to the limited sample size and specific training data characteristics. Data Quality Concerns The model's performance heavily depends on the quality of training data. Issues like noise, artifacts, and inconsistent imaging protocols can negatively impact accuracy. The dataset may not fully capture the diversity of bladder cancer presentations, limiting its real-world applicability. Future Directions To overcome these limitations, future research should focus on expanding datasets, improving model interpretability, enhancing data preprocessing techniques, exploring multi-modal approaches, and implementing continuous model updating mechanisms. These efforts will help refine the AI model, making it a more reliable tool for bladder cancer diagnosis in clinical practice. 5. Conclusions In conclusion, this research demonstrates the feasibility and potential impact of AI-driven diagnostic tools for bladder cancer. With continued refinement and validation, such models could become integral to bladder cancer detection and personalized treatment planning, ultimately improving patient outcomes and advancing the field of precision medicine. Abbreviations The following abbreviations are used in this manuscript: CT Computed Tomography IOU Intersection over Union TCIA The Cancer Imaging Archive Acc Grad-CAM CNNs Accuracy Gradient-weighted Class Activation Mapping convolutional neural networks Declarations Author Contributions: Lei Jiang and Wenyu Ge drafted the manuscript. Lei Jiang and Wenyu Ge contributed equally. Jingru Huo and Shijie Li drew the pictures. Ruijiao Feng and Liu Ji analyzed the data. Tingting Fan reviewed and revised the manuscript. All the authors have read and approved the final version of the manuscript. All the authors contributed to the manuscript and approved the submitted version. Funding: Please add: This study was supported by the scientific research of the Heilongjiang Provincial Health Commission (grant number 20240909010037), Heilongjiang Postdoctoral Fund (LBH-Z24304) and Heilongjiang provincial scientific research project of traditional Chinese Medicine(ZHY2024-041) Institutional Review Board Statement: The study was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board of Heilong Provincial Hospital. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Data Availability Statement: The original contributions of this study are included in the article and supplementary material. Further inquiries can be directed to the corresponding authors. The data is collected in https://www.cancerimagingarchive.net/collection/tcga-blca/. Acknowledgments: The authors would like to express their sincere gratitude to the TCGA Research Network for their invaluable contributions to cancer genomics research. Conflicts of Interest: The authors declare no conflicts of interest. Clinical trial number: not applicable. References Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, Jemal A. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024;74(3):229–63. Siegel RL, Kratzer TB, Giaquinto AN, Sung H, Jemal A. Cancer statistics, 2025. CA Cancer J Clin. 2025;75(1):10–45. Åström LM, Sibolt P, Chamberlin H, Serup-Hansen E, Andersen CE, van Herk M, Mouritsen LS, Aznar MC, Behrens CP. Artificial intelligence-generated targets and inter-observer variation in online adaptive radiotherapy of bladder cancer. Phys Imaging Radiat Oncol. 2024;31:100640. Baressi Šegota S, Lorencin I, Smolić K, Anđelić N, Markić D, Mrzljak V, Štifanić D, Musulin J, Španjol J, Car Z. Semantic Segmentation of Urinary Bladder Cancer Masses from CT Images: A Transfer Learning Approach. Biology (Basel). 2021;10(11). Cao Y, Zhu H, Li Z, Liu C, Ye J. CT Image-Based Radiomic Analysis for Detecting PD-L1 Expression Status in Bladder Cancer Patients. Acad Radiol. 2024;31(9):3678–87. Cha KH, Hadjiiski L, Chan HP, Weizer AZ, Alva A, Cohan RH, Caoili EM, Paramagul C, Samala RK. Bladder Cancer Treatment Response Assessment in CT using Radiomics with Deep-Learning. Sci Rep. 2017;7(1):8738. Garapati SS, Hadjiiski L, Cha KH, Chan HP, Caoili EM, Cohan RH, Weizer A, Alva A, Paramagul C, Wei J, Zhou C. Urinary bladder cancer staging in CT urography using machine learning. Med Phys. 2017;44(11):5814–23. Girard A, Dercle L, Vila-Reyes H, Schwartz LH, Girma A, Bertaux M, Radulescu C, Lebret T, Delcroix O, Rouanne M. A machine-learning-based combination of criteria to detect bladder cancer lymph node metastasis on [(18)F]FDG PET/CT: a pathology-controlled study. Eur Radiol. 2023;33(4):2821–9. Palazzo G, Mangili P, Deantoni C, Fodor A, Broggi S, Castriconi R, Ubeira Gabellini MG, Del Vecchio A, Di Muzio NG, Fiorino C. Real-world validation of Artificial Intelligence-based Computed Tomography auto-contouring for prostate cancer radiotherapy planning. Phys Imaging Radiat Oncol. 2023;28:100501. Wei Z, Xv Y, Liu H, Li Y, Yin S, Xie Y, Chen Y, Lv F, Jiang Q, Li F, Xiao M. A CT-based deep learning model predicts overall survival in patients with muscle invasive bladder cancer after radical cystectomy: a multicenter retrospective cohort study. Int J Surg. 2024;110(5):2922–32. Xiong S, Fu Z, Deng Z, Li S, Zhan X, Zheng F, Yang H, Liu X, Xu S, Liu H, Fan B, Dong W, Song Y, Fu B. Machine learning-based CT radiomics enhances bladder cancer staging predictions: A comparative study of clinical, radiomics, and combined models. Med Phys. 2024;51(9):5965–77. Yang Y, Zou X, Wang Y, Ma X. Application of deep learning as a noninvasive tool to differentiate muscle-invasive bladder cancer and non-muscle-invasive bladder cancer with CT. Eur J Radiol. 2021;139:109666. Ying T, Borrelli P, Edenbrandt L, Enqvist O, Kaboteh R, Trägårdh E, Ulén J, Kjölhede H. Automated artificial intelligence-based analysis of skeletal muscle volume predicts overall survival after cystectomy for urinary bladder cancer. Eur Radiol Exp. 2021;5(1):50. Zhang G, Wu Z, Xu L, Zhang X, Zhang D, Mao L, Li X, Xiao Y, Guo J, Ji Z, Sun H, Jin Z. Deep Learning on Enhanced CT Images Can Predict the Muscular Invasiveness of Bladder Cancer. Front Oncol. 2021;11:654685. Zhang-Yin J, Girard A, Marchal E, Lebret T, Homo Seban M, Uhl M, Bertaux M. PET Imaging in Bladder Cancer: An Update and Future Direction. Pharmaceuticals (Basel). 2023;16(4). Zheng Y, Shi H, Hai B, Zhang J. A commentary on 'A CT-based deep learning model predicts overall survival in patients with muscle invasive bladder cancer after radical cystectomy: a multicenter retrospective cohort study'. Int J Surg. 2024;110(8):5200–1. Kirk S, Lee Y, Lucchesi FR, Aredes ND, Gruszauskas N, Catto J, Garcia K, Jarosz R, Duddalwar V, Varghese B, Rieger-Christ K, Lemmerman J. The Cancer Genome Atlas Urothelial Bladder Carcinoma Collection (TCGA-BLCA) (Version 8) [Data set]. 2016. Chen X, Zhang Y, Zhou J, Pan Y, Xu H, Shen Y, Cao G, Su MY, Wang M. Combination of Deep Learning Grad-CAM and Radiomics for Automatic Localization and Diagnosis of Architectural Distortion on DBT. Acad Radiol. 2025;32(3):1287–96. Fanizzi A, Comes MC, Bove S, Cavalera E, de Franco P, Di Rito A, Errico A, Lioce M, Pati F, Portaluri M, Saponaro C, Scognamillo G, Troiano I, Troiano M, Zito FA, Massafra R. Explainable prediction model for the human papillomavirus status in patients with oropharyngeal squamous cell carcinoma using CNN on CT images. Sci Rep. 2024;14(1):14276. Huang Z, Zhang X, Ju Y, Zhang G, Chang W, Song H, Gao Y. Explainable breast cancer molecular expression prediction using multi-task deep-learning based on 3D whole breast ultrasound. Insights Imaging. 2024;15(1):227. Shah SJH, Albishri A, Wang R, Lee Y. Integrating local and global attention mechanisms for enhanced oral cancer detection and explainability. Comput Biol Med. 2025;189:109841. Talaat FM, Gamel SA, El-Balka RM, Shehata M, ZainEldin H. Grad-CAM Enabled Breast Cancer Classification with a 3D Inception-ResNet V2: Empowering Radiologists with Explainable Insights. Cancers (Basel). 2024;16(21). Xia J, Li J, Wang X, Li Y, Li J. Enhanced cancer classification and critical feature visualization using Raman spectroscopy and convolutional neural networks. Spectrochim Acta Mol Biomol Spectrosc. 2025;326:125242. Alazwari S, Alsamri J, Alamgeer M, Alotaibi SS, Obayya M, Salama AS. Automated gall bladder cancer detection using artificial gorilla troops optimizer with transfer learning on ultrasound images. Sci Rep. 2024;14(1):21845. Bozkurt B, Coskun K, Bakal G. Building a challenging medical dataset for comparative evaluation of classifier capabilities. Comput Biol Med. 2024;178:108721. Jian Y, Zhang N, Bi Y, Liu X, Fan J, Wu W, Liu T. TC-Sniffer: A Transformer-CNN Bibranch Framework Leveraging Auxiliary VOCs for Few-Shot UBC Diagnosis via Electronic Noses. ACS Sens. 2025;10(1):213–24. Kurata Y, Nishio M, Moribata Y, Otani S, Himoto Y, Takahashi S, Kusakabe J, Okura R, Shimizu M, Hidaka K, Nishio N, Furuta A, Kido A, Masui K, Onishi H, Segawa T, Kobayashi T, Nakamoto Y. Development of deep learning model for diagnosing muscle-invasive bladder cancer on MRI with vision transformer. Heliyon. 2024;10(16):e36144. Shalata AT, Alksas A, Shehata M, Khater S, Ezzat O, Ali KM, Gondim D, Mahmoud A, El-Gendy EM, Mohamed MA, Alghamdi NS, Ghazal M, El-Baz A. Precise grading of non-muscle invasive bladder cancer with multi-scale pyramidal CNN. Sci Rep. 2024;14(1):25131. Holzbeierlein JM, Bixler BR, Buckley DI, Chang SS, Holmes R, James AC, Kirkby E, McKiernan JM, Schuckman AK. Diagnosis and Treatment of Non-Muscle Invasive Bladder Cancer: AUA/SUO Guideline: 2024 Amendment. J Urol. 2024;211(4):533–8. Yang T, Luo W, Yu J, Zhang H, Hu M, Tian J. Bladder cancer immune-related markers: diagnosis, surveillance, and prognosis. Front Immunol. 2024;15:1481296. Fan J, Chen B, Luo Q, Li J, Huang Y, Zhu M, Chen Z, Li J, Wang J, Liu L, Wei Q, Cao D. Potential molecular biomarkers for the diagnosis and prognosis of bladder cancer. Biomed Pharmacother. 2024;173:116312. Flaig TW, Spiess PE, Abern M, Agarwal N, Bangs R, Buyyounouski MK, Chan K, Chang SS, Chang P, Friedlander T, Greenberg RE, Guru KA, Herr HW, Hoffman-Censits J, Kaimakliotis H, Kishan AU, Kundu S, Lele SM, Mamtani R, Mian OY, Michalski J, Montgomery JS, Parikh M, Patterson A, Peyton C, Plimack ER, Preston MA, Richards K, Sexton WJ, Siefker-Radtke AO, Stewart T, Sundi D, Tollefson M, Tward J, Wright JL, Cassara CJ, Gurski LA. NCCN Guidelines® Insights: Bladder Cancer, Version 3.2024. J Natl Compr Canc Netw. 2024;22(4):216–25. Borna MR, Sepehri MM, Shadpour P, Khaleghi Mehr F. Enhancing bladder cancer diagnosis through transitional cell carcinoma polyp detection and segmentation: an artificial intelligence powered deep learning solution. Front Artif Intell. 2024;7:1406806. Lotan Y, Krishna V, Abuzeid WM, Launer B, Chang SS, Krishna V, Shingi S, Gordetsky JB, Gerald T, Woldu S, Shkolyar E, Hayne D, Redfern A, Spalding L, Stewart C, Eyzaguirre E, Imtiaz S, Narayan VM, Packiam VT, O'Donnell MA, Li R, Baekelandt L, Joniau S, Zuiverloon T, Fernandez MI, Schultz M, Hensley PJ, Allison D, Taylor JA, Hamza A, Kamat A, Nimgaonkar V, Sonawane S, Miller DL, Watson D, Vrabac D, Joshi A, Shah JB, Williams SB. Predicting Response to Intravesical Bacillus Calmette-Guérin in High-Risk Nonmuscle-Invasive Bladder Cancer Using an Artificial Intelligence-Powered Pathology Assay: Development and Validation in an International 12-Center Cohort. J Urol. 2025;213(2):192–204. Ma X, Zhang Q, He L, Liu X, Xiao Y, Hu J, Cai S, Cai H, Yu B. Artificial intelligence application in the diagnosis and treatment of bladder cancer: advance, challenges, and opportunities. Front Oncol. 2024;14:1487676. Zhu M, Gu Z, Chen F, Chen X, Wang Y, Zhao G. Application of artificial intelligence in the diagnosis and treatment of urinary tumors. Front Oncol. 2024;14:1440626. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 13 Nov, 2025 Reviews received at journal 05 Nov, 2025 Reviews received at journal 03 Nov, 2025 Reviewers agreed at journal 02 Nov, 2025 Reviewers agreed at journal 02 Nov, 2025 Reviewers agreed at journal 02 Nov, 2025 Reviewers invited by journal 02 Nov, 2025 Editor invited by journal 31 Oct, 2025 Editor assigned by journal 07 Oct, 2025 Submission checks completed at journal 07 Oct, 2025 First submitted to journal 01 Oct, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7760584","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":541198525,"identity":"89bc2b27-c0e1-491c-aa90-b78d89047cc0","order_by":0,"name":"Lei Jiang","email":"","orcid":"","institution":"The Second Affiliated Hospital of Harbin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Lei","middleName":"","lastName":"Jiang","suffix":""},{"id":541198526,"identity":"c4af48b8-ac65-4dc2-a308-00a911507cff","order_by":1,"name":"Wenyu Ge","email":"","orcid":"","institution":"Harbin Institute of Technology, Heilongjiang Provincial Hospital","correspondingAuthor":false,"prefix":"","firstName":"Wenyu","middleName":"","lastName":"Ge","suffix":""},{"id":541198527,"identity":"6c9eba87-9bde-450d-b027-2a4c12ce6b49","order_by":2,"name":"Ruijiao Feng","email":"","orcid":"","institution":"Heilongjiang University Of Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Ruijiao","middleName":"","lastName":"Feng","suffix":""},{"id":541198528,"identity":"c2180def-3a18-4b65-b2af-f44111e57657","order_by":3,"name":"Ji Liu","email":"","orcid":"","institution":"The Second Affiliated Hospital of Harbin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Ji","middleName":"","lastName":"Liu","suffix":""},{"id":541198529,"identity":"9e82d51c-4711-41f6-8920-76de1afcb50e","order_by":4,"name":"Jingru Huo","email":"","orcid":"","institution":"The Second Affiliated Hospital of Harbin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Jingru","middleName":"","lastName":"Huo","suffix":""},{"id":541198530,"identity":"128c7ff6-948f-45a8-a2d8-47ee5b6adc30","order_by":5,"name":"Shijie Li","email":"","orcid":"","institution":"Third Affiliated Hospital of Harbin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Shijie","middleName":"","lastName":"Li","suffix":""},{"id":541198531,"identity":"0d7edc0a-dfcd-465d-8f4c-0af32ab808d0","order_by":6,"name":"Tingting Fan","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABBklEQVRIie3QsU7DMBCA4YsiJcvRrkZF9BWuslQYqoZHcRQpXQLqmLFdnCUSKyw8BrMrr+YJYMjEQgePGRDC2VhwGJHqf7jB8mfZBgiF/mPxMAgg3UHUiZpdJmmjvAIHohxBBTFZs+ITNMJPhjGc6khy/ijL/IndkJdk6Zn+sNu3dZu+qBkmeiMZCOjrZ8/FJuW1oveixTvBEfWtnO1V1JpXD8ElKdJFBhUVyBy5UCKO5B8ITo+kkfQmYYLGCO8cWSOrFvsHUYpxonEJhrRAduRg1Woh3ScffG9J7w239afOcFpd9fkXm8+b5tD19e/ElTA38t3PJeXb74qtG9nIplAoFDrlvgGPRVdQq1hcUAAAAABJRU5ErkJggg==","orcid":"","institution":"The Second Affiliated Hospital of Harbin Medical University","correspondingAuthor":true,"prefix":"","firstName":"Tingting","middleName":"","lastName":"Fan","suffix":""}],"badges":[],"createdAt":"2025-10-01 14:57:00","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7760584/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7760584/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":95806821,"identity":"48b23874-f40a-4243-a6ad-652f458755fa","added_by":"auto","created_at":"2025-11-13 08:47:56","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2967157,"visible":true,"origin":"","legend":"","description":"","filename":"artificialintelligence.docx","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/9b7a647dc772807bf1d9f73d.docx"},{"id":95807250,"identity":"e0b0d845-9ddf-4421-9ad2-0a3d302e0fac","added_by":"auto","created_at":"2025-11-13 08:48:15","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7714,"visible":true,"origin":"","legend":"","description":"","filename":"b604e6c80d3344c7b7620adda927e04e.json","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/1f745537ecb2cf42c3ad7e46.json"},{"id":95806983,"identity":"94e11e66-6aa0-4fe2-9342-eaa9ef53d5db","added_by":"auto","created_at":"2025-11-13 08:48:01","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":108331,"visible":true,"origin":"","legend":"","description":"","filename":"b604e6c80d3344c7b7620adda927e04e1enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/faace09e794ad20a6413cc56.xml"},{"id":95807326,"identity":"b2e28176-2c60-4b0f-b0f3-31b2db88e77b","added_by":"auto","created_at":"2025-11-13 08:48:21","extension":"eps","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":234511,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage1.eps","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/91f45f61becc0dd3f0062fb1.eps"},{"id":95807294,"identity":"535ccbb1-91e5-46de-83a4-e9aa54d40e3b","added_by":"auto","created_at":"2025-11-13 08:48:17","extension":"eps","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":327815,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage2.eps","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/58e9ea591f3d8c52f3f7f7e3.eps"},{"id":95807170,"identity":"fcc92b06-abf2-4057-b0a4-5059d76969f5","added_by":"auto","created_at":"2025-11-13 08:48:10","extension":"eps","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":169177,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage3.eps","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/a012963bcfdca8f162c7f437.eps"},{"id":95807070,"identity":"13605488-51d7-48fd-93a1-0ebfebc87c0a","added_by":"auto","created_at":"2025-11-13 08:48:05","extension":"eps","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":147995,"visible":true,"origin":"","legend":"","description":"","filename":"drawingimage4.eps","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/6c2391e474f6e63313aa5db9.eps"},{"id":95806990,"identity":"fefac1d5-1905-4e3a-bc72-08c73e90f1a5","added_by":"auto","created_at":"2025-11-13 08:48:01","extension":"jpeg","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3002692,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/26e4a78c602221de724d62c5.jpeg"},{"id":95807327,"identity":"dc35253d-323b-4f95-94b9-e87c45bbe9d5","added_by":"auto","created_at":"2025-11-13 08:48:21","extension":"jpeg","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":135896,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/7274b608bd557fdc09238147.jpeg"},{"id":95807341,"identity":"61a2df58-45a3-45b4-b3bd-556ec74a651c","added_by":"auto","created_at":"2025-11-13 08:48:25","extension":"jpeg","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5275930,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/381e5911ee908fcbb61d9593.jpeg"},{"id":95807330,"identity":"644cd260-0435-499f-942a-b8f29c8e0042","added_by":"auto","created_at":"2025-11-13 08:48:21","extension":"jpeg","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":354204,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/566ec74b10e1f561d4b41ed0.jpeg"},{"id":95806947,"identity":"d2f3968d-e4ab-4435-9336-21d7d06924af","added_by":"auto","created_at":"2025-11-13 08:48:01","extension":"jpeg","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3854144,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/f287aa5283abf16f844be0d0.jpeg"},{"id":95807209,"identity":"a03c1345-3750-4e8d-99e8-6b99ba0bcbca","added_by":"auto","created_at":"2025-11-13 08:48:12","extension":"jpeg","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":157650,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/514b4d99b43a27aeed97b75e.jpeg"},{"id":95807216,"identity":"3a6b08f1-85f9-4332-8e7e-ac9d10a0dbdb","added_by":"auto","created_at":"2025-11-13 08:48:13","extension":"jpeg","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3479610,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/5126b6971a6d843a6371bff4.jpeg"},{"id":95807262,"identity":"a37cb373-fa7f-4fcf-9ac6-32aef7bf1b61","added_by":"auto","created_at":"2025-11-13 08:48:15","extension":"jpeg","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":128460,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage8.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/20e7ad447ca8f7f8ecd8c6f6.jpeg"},{"id":95806987,"identity":"22a6d1af-0a35-4e83-8502-6f5fe1441160","added_by":"auto","created_at":"2025-11-13 08:48:01","extension":"png","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":385160,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/0608bc26cfcecac8b65c09ee.png"},{"id":95807329,"identity":"b88a9646-3b8e-417c-9499-d3d58350e550","added_by":"auto","created_at":"2025-11-13 08:48:21","extension":"png","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":27206,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/640c891f5dbed4d955631210.png"},{"id":95807365,"identity":"bea1f035-2c39-4f5f-bf71-e39b5aff97e8","added_by":"auto","created_at":"2025-11-13 08:48:30","extension":"png","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":44960,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/09da6b090ab5da373dfadefa.png"},{"id":95807137,"identity":"887a5288-f1da-4507-9180-1b2b5addf33a","added_by":"auto","created_at":"2025-11-13 08:48:08","extension":"png","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":70751,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/6395dc311c1975bf36ac9ce8.png"},{"id":95807201,"identity":"3df79d4e-1edf-4c43-852f-a77e7b9ed1ba","added_by":"auto","created_at":"2025-11-13 08:48:11","extension":"png","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":510412,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/6622695cee802a5c978b88fc.png"},{"id":95807044,"identity":"0c373d6d-52d1-485d-bf11-8555abe0fc43","added_by":"auto","created_at":"2025-11-13 08:48:04","extension":"png","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":33144,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/c5cd06ff0e2109927b9ecb1b.png"},{"id":95806937,"identity":"731abde0-876b-44d1-8a96-68b95272e945","added_by":"auto","created_at":"2025-11-13 08:48:00","extension":"png","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":79537,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/0440a9d913883e1642e6efa6.png"},{"id":95807202,"identity":"b4235e31-3cc9-4b8b-a794-0f9ae45eec5a","added_by":"auto","created_at":"2025-11-13 08:48:12","extension":"png","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":28109,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage8.png","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/ea1c9a2c21de69f88bcd0d3a.png"},{"id":95807147,"identity":"010fb2ea-1c7a-49eb-91da-a74417895aa1","added_by":"auto","created_at":"2025-11-13 08:48:09","extension":"xml","order_by":23,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":107144,"visible":true,"origin":"","legend":"","description":"","filename":"b604e6c80d3344c7b7620adda927e04e1structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/342ac0634d65d41750305e82.xml"},{"id":95807156,"identity":"af988243-79ce-45e6-b570-50563afc7e3b","added_by":"auto","created_at":"2025-11-13 08:48:09","extension":"html","order_by":24,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":113893,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/7b9d00a3a0b986b84159ca79.html"},{"id":95807180,"identity":"aea853bc-ba59-4c0b-a151-c019313c580c","added_by":"auto","created_at":"2025-11-13 08:48:11","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":75748,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTitle: \u003c/strong\u003eFlowchart of Clinical Data Collection and Modeling Process\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLegend: \u003c/strong\u003eThe process involves collecting patient data, preprocessing it, building a predictive model via machine learning, and validating the model's performance.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/a239810ec8fec82115e5e122.png"},{"id":95806901,"identity":"5717cda7-d47c-49e3-bf4f-699f6e9578d9","added_by":"auto","created_at":"2025-11-13 08:47:59","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":144878,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTitle:\u003c/strong\u003e Comprehensive Workflow from Data Collection to AI Model Application\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLegend: \u003c/strong\u003eThis workflow encompassed data collection, AI model building, explainable analysis, and further application. In data collection, multi-source data was gathered and preprocessed. The AI model building phase used machine-learning algorithms to train data, generating models with high predictive and diagnostic power. Explainable analysis delved into the model's internal structure and decision-making mechanism, employing visualization techniques. In further application, the model was implemented clinically to aid doctors in diagnosis, treatment planning, and patient prognosis assessment.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/a67fb0b947e1159559b12f46.png"},{"id":95807335,"identity":"52258cd2-68c3-4e88-9867-a287a74c0404","added_by":"auto","created_at":"2025-11-13 08:48:23","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":40925,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTitle: \u003c/strong\u003eConfusion Matrices for AI model Evaluation on Validation and Test Sets\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLegend: \u003c/strong\u003eBoth the validation and test set confusion matrices were displayed, showing the comparison between the model's predicted results and actual labels, and offering insights into the model's performance during development and real-world application.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/db9009aecdf113272ab89f02.png"},{"id":95807090,"identity":"3f537c0f-1510-4faf-acfc-729374a01fbb","added_by":"auto","created_at":"2025-11-13 08:48:05","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":388490,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTitle:\u003c/strong\u003e Visualization of Bladder Tumor Labeling and AI-Predicted Locations\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLegend: \u003c/strong\u003eThese visualizations provide insights into the AI's decision-making process and its alignment with actual tumor regions, aiding in the evaluation of the model's accuracy and identification of areas for improvement.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/2703bf8aa4774ee5f4d23696.png"},{"id":95810362,"identity":"16892bf9-097d-4b57-84f2-9e5f98258a98","added_by":"auto","created_at":"2025-11-13 08:52:26","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1326532,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7760584/v1/96b43f50-8b07-4352-bdbf-908eda2e38e8.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Innovative AI Model for Bladder Cancer Diagnosis","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eBladder cancer, a prevalent malignancy within the urinary system, poses significant threats to human health\u003csup\u003e[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]\u003c/sup\u003e. Early and accurate diagnosis is paramount for improving patient prognosis and reducing mortality rates. Computed Tomography (CT) imaging has emerged as a crucial tool in the detection and diagnosis of bladder cancer. It offers detailed anatomical information and can visualize the size, location, and extent of tumors. However, interpreting CT images relies heavily on the expertise and experience of radiologists, and subtle lesions or complex cases may lead to variability in diagnostic outcomes.\u003c/p\u003e\u003cp\u003eWhile artificial intelligence (AI) has demonstrated potential in bladder cancer research, this field remains in its early stages, particularly concerning the interpretability of AI applications\u003csup\u003e[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]\u003c/sup\u003e. Recent studies have examined AI's role in various aspects of bladder cancer management, such as diagnosis, treatment response evaluation, and prognosis prediction\u003csup\u003e[\u003cspan additionalcitationids=\"CR4 CR5 CR6 CR7 CR8 CR9 CR10 CR11 CR12 CR13 CR14 CR15\" citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]\u003c/sup\u003e. For example, the use of AI-generated targets and inter-observer variation in online adaptive radiotherapy for bladder cancer was investigated by \u0026Aring;str\u0026ouml;m LM et al, underscoring the challenges and opportunities in this domain\u003csup\u003e[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]\u003c/sup\u003e. Additionally, Baressi Šegota S et al. proposed a transfer learning method for the semantic segmentation of urinary bladder cancer masses in CT images, illustrating AI's potential to enhance diagnostic precision\u003csup\u003e[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eFurthermore, several investigations have centered on the application of radiomics and deep learning in bladder cancer\u003csup\u003e[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]\u003c/sup\u003e. A CT image-based radiomic model was developed by Cao Y et al. to detect PD-L1 expression status in bladder cancer patients, which may inform personalized treatment approaches\u003csup\u003e[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]\u003c/sup\u003e. The assessment of bladder cancer treatment response in CT using radiomics combined with deep learning was conducted by Cha KH et al., demonstrating AI's potential in monitoring treatment effectiveness\u003csup\u003e[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]\u003c/sup\u003e. Moreover, machine learning was applied by Garapati SS et al. to stage urinary bladder cancer in CT urography, showcasing AI's capacity to aid clinical decision-making\u003csup\u003e[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]\u003c/sup\u003e.Other research has focused on AI's utility in predicting prognosis and detecting metastasis\u003csup\u003e[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]\u003c/sup\u003e. A machine learning-based combination of criteria was proposed by Girard A et al. to detect bladder cancer lymph node metastasis, which may improve metastasis detection accuracy\u003csup\u003e[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]\u003c/sup\u003e. A CT-based deep learning model was developed by Wei Z et al. to predict overall survival in muscle-invasive bladder cancer patients post-radical cystectomy, emphasizing AI's prognostic value\u003csup\u003e[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]\u003c/sup\u003e. Additionally, Xiong S et al. demonstrated that machine learning-based CT radiomics improves bladder cancer staging predictions, suggesting AI's potential to enhance staging accuracy and treatment planning\u003csup\u003e[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eAI has also been employed to distinguish between muscle-invasive and non-muscle-invasive bladder cancer. It was shown by Yang Y et al. that deep learning can serve as a noninvasive tool to differentiate these cancer types using CT, potentially reducing the need for invasive procedures\u003csup\u003e[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]\u003c/sup\u003e. Automated AI-based analysis of skeletal muscle volume was utilized by Ying T et al. (2021) to predict overall survival after cystectomy for urinary bladder cancer, highlighting AI's potential in identifying prognostic indicators\u003csup\u003e[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eAlthough AI research in bladder cancer is still in its nascent stages, especially regarding interpretability. As the field progresses, further research is required to address interpretability challenges and validate AI applications' clinical utility in bladder cancer.\u003c/p\u003e\u003cp\u003eOur study aims to develop an AI model for bladder cancer diagnosis using CT imaging data from our hospital and The Cancer Imaging Archive (TCIA) \u003csup\u003e[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]\u003c/sup\u003e. By leveraging deep learning techniques and incorporating Grad-CAM for region of interest identification \u003csup\u003e[\u003cspan additionalcitationids=\"CR19 CR20 CR21 CR22\" citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]\u003c/sup\u003e, we strive to enhance diagnostic accuracy and provide a reliable auxiliary tool for clinicians. Specifically, recent advancements in explainable AI (XAI) have demonstrated the potential to improve diagnostic transparency and clinical trust. For instance, Chen et al. \u003csup\u003e[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]\u003c/sup\u003e combined deep learning with Grad-CAM and radiomics to automatically localize and diagnose architectural distortion on digital breast tomosynthesis (DBT), achieving high diagnostic accuracy while providing visual explanations for clinical decision-making. Similarly, Fanizzi et al. \u003csup\u003e[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]\u003c/sup\u003e developed an explainable prediction model for human papillomavirus (HPV) status in oropharyngeal squamous cell carcinoma patients using CNN on CT images, which not only predicted HPV status with high accuracy but also highlighted critical image features contributing to the predictions.\u003c/p\u003e\u003cp\u003eIn the context of breast cancer, Huang et al. \u003csup\u003e[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]\u003c/sup\u003e proposed an explainable AI model for predicting molecular expressions using multi-task deep learning based on 3D whole breast ultrasound, which provided insights into the biological underpinnings of imaging features. Shah et al. \u003csup\u003e[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]\u003c/sup\u003e integrated local and global attention mechanisms to enhance oral cancer detection, with explainability maps helping clinicians understand the model's focus during diagnosis. Furthermore, Talaat et al. \u003csup\u003e[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]\u003c/sup\u003e utilized Grad-CAM with a 3D Inception-ResNet V2 for breast cancer classification on mammograms, empowering radiologists with explainable insights into the model's decision-making process. Lastly, Xia et al. \u003csup\u003e[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]\u003c/sup\u003e enhanced cancer classification using Raman spectroscopy and CNN, with visualization of critical features aiding in the interpretation of spectral data.\u003c/p\u003e\u003cp\u003eAlthough Grad-CAM has demonstrated intriguing potential for enhancing the interpretability of AI models in many cancers, its application and efficacy in bladder cancer remain to be more thoroughly investigated\u003csup\u003e[\u003cspan additionalcitationids=\"CR19 CR20 CR21 CR22\" citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]\u003c/sup\u003e. While there have been notable advancements in the application of AI to bladder cancer diagnosis, our understanding of how Grad-CAM can be effectively utilized in this specific context is less. This gap in research might, in part, reflect the inherent complexities associated with bladder cancer imaging, as well as the fact that AI applications in this area are still evolving compared to more established fields.\u003c/p\u003e\u003cp\u003eBy leveraging deep learning techniques and incorporating Grad-CAM for region of interest identification, we strive to enhance diagnostic accuracy and provide a reliable auxiliary tool for clinicians.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e"},{"header":"2. Materials and Methods","content":"\u003cp\u003eThe study utilized a combination of our hospital data and publicly available datasets from The Cancer Imaging Archive (TCIA) to compile a comprehensive dataset of CT images for bladder cancer diagnosis. Patients diagnosed with bladder cancer and those confirmed to be free of the disease were included.\u003c/p\u003e\u003cp\u003eThe CT images were acquired using standardized protocols, ensuring consistency in image quality and resolution. Each image underwent preliminary inspection and annotation by experienced radiologists to identify regions of interest and potential artifacts. Data preprocessing steps included noise reduction, image enhancement, and normalization to standardize the input for the AI model. The flowchart illustrating the clinical data collection and modeling process was presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\u003cp\u003e\u003cb\u003eModel Construction\u003c/b\u003e\u003c/p\u003e\u003cp\u003eA deep learning-based AI model was developed using convolutional neural networks (CNNs), which are particularly effective for image recognition tasks\u003csup\u003e[\u003cspan additionalcitationids=\"CR25 CR26 CR27\" citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]\u003c/sup\u003e. The model architecture was designed to process CT images and extract relevant features automatically. Transfer learning techniques were employed, leveraging pre-trained models on large-scale image datasets to accelerate training and improve performance.\u003c/p\u003e\u003cp\u003eThe training process involved feeding the preprocessed CT images into the model, with bladder cancer diagnoses serving as labels. The model learned to associate specific imaging patterns with diagnostic outcomes through multiple iterations. To prevent overfitting, regularization techniques such as dropout layers and data augmentation were applied.\u003c/p\u003e\u003cp\u003eGrad-CAM (Gradient-weighted Class Activation Mapping) was integrated into the model to identify regions of interest within the CT images that are most relevant to bladder cancer diagnosis. This technique helps visualize the model's decision-making process, highlighting areas that contribute significantly to the diagnostic prediction. By focusing on these regions, the model's accuracy and reliability were enhanced.\u003c/p\u003e\u003cp\u003e\u003cb\u003eStatistical Methods\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe performance of the AI model was evaluated using a variety of statistical metrics. Accuracy and Intersection over Union (IoU) were calculated to assess the model's diagnostic capability. A confusion matrix was generated to provide a detailed breakdown of true positives, true negatives, false positives, and false negatives in percentages. To ensure the robustness of the results, cross-validation techniques were applied. The dataset was divided into training, validation, and test sets, with the validation set used to tune hyperparameters and the test set reserved for final evaluation.\u003c/p\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\u003ch2\u003e3.1 Quantitative Results\u003c/h2\u003e\u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eFigure \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e was presented to show the entire process from data collection to AI model building, explainable analysis, and further application. In the data collection phase, multi-source data, including patient basic information, medical records, and clinical examination results, was collected and strictly filtered and pre-processed to ensure quality and usability. Next, in the AI model building phase, machine-learning algorithms were employed to train the processed data, and models with high predictive and diagnostic power were obtained. The explainable analysis part involved a deep investigation into the model's internal structure and decision-making mechanism, with visualization techniques used to clarify the complex model's operation and enhance transparency and credibility. Finally, in the further application stage, the model was widely applied in clinical practice, and doctors were assisted in disease diagnosis, treatment planning, and patient prognosis assessment.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003e3.2 Analysis of Results\u003c/h2\u003e\u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eBased on the results from Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, we analyzed the AI model's performance in terms of IoU and accuracy.\u003c/p\u003e\u003cp\u003eIn terms of IoU, the background class had shown consistent performance, while the bladder and bladder cancer classes exhibited slight variations between the two datasets. This may suggest that the model encounters certain challenges when identifying the bladder and bladder cancer, or that the validation set contains more challenging samples.\u003c/p\u003e\u003cp\u003eAs for accuracy, the background class still maintains a very high level, and the accuracy of the bladder and bladder cancer classes has increased in the test set. This further confirms the potential issues the model faces when processing these two classes.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003ePerformance Evaluation of the Validation Set\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"3\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eClass\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eIoU\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eAcc\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBackground\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e99.63\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e99.79\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBladder\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e91.25\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e98.21\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBladder cancer\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e84.53\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e91.44\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003e\u003cb\u003ePerformance Evaluation of the Test Set\u003c/b\u003e\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"3\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eClass\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eIoU\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eAcc\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBackground\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e99.58\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e99.8\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBladder\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e92.04\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e98.65\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBladder cancer\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e84.61\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e90.18\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e3.3 Confusion Matrices for Validation and Test Sets in Model Evaluation\u003c/h2\u003e\u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eFigure \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e clearly showed the confusion matrices for the validation and test sets. During the model evaluation, these two confusion matrices offered rich data, helping us fully understand the model's performance. As could be seen from the figure, each confusion matrix contained a comparison between the model's predicted results and the actual labels.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003e3.4 Visualization of Bladder Tumor Labeling and AI-Identified Locations via Grad-CAM\u003c/h2\u003e\u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eIn Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA, the labeling of bladder tumors was clearly displayed, with different colors and markers indicating various tumor types and stages. The labels were meticulously placed, ensuring accurate correspondence to the tumor regions. In Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB, the Grad-CAM AI model's identified tumor locations were presented. The heat maps generated by the AI model highlighted areas of interest with varying intensity levels, reflecting the model's confidence in its predictions. These visualizations were offered, giving valuable insights into the AI's decision-making process and its alignment with actual tumor regions.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e"},{"header":"4. Discussion","content":"\u003cp\u003e\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eBladder cancer remains a prevalent malignancy affecting the urinary system, with a rising incidence globally\u003csup\u003e[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]\u003c/sup\u003e. This disease significantly impacts patients' quality of life and poses considerable economic burdens due to high treatment costs\u003csup\u003e[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]\u003c/sup\u003e. Traditional diagnostic approaches, such as cystoscopy and biopsy, are invasive and time-consuming, often leading to patient discomfort and potential complications\u003csup\u003e[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]\u003c/sup\u003e. Consequently, the need for more efficient and non-invasive diagnostic methods is urgent to enhance early detection and improve patient outcomes\u003csup\u003e[\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]\u003c/sup\u003e. Advances in imaging technologies and AI present promising avenues for the development of innovative diagnostic tools that can facilitate earlier diagnosis and better management of bladder cancer\u003csup\u003e[\u003cspan additionalcitationids=\"CR34 CR35\" citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eThis study focuses on leveraging deep learning algorithms to construct an artificial intelligence diagnostic model based on CT imaging data for bladder cancer. By extracting and analyzing imaging features, the model aims to improve diagnostic accuracy and consistency, addressing the limitations of traditional methods. The findings of this research provide a foundation for the model's potential application in clinical settings, enhancing early diagnostic capabilities and offering decision support for healthcare professionals. The discussion will delve into the implications of the primary results, emphasizing the importance of AI in refining bladder cancer diagnostics and the need for further validation in diverse clinical populations.\u003c/p\u003e\u003cp\u003eThis study highlights the significant advancements made in the application of deep learning techniques for the diagnosis of bladder cancer through CT imaging. The extraction of radiomic features from CT scans allows for a complicated understanding of tumor characteristics that traditional imaging methods may overlook. This underscores the potential of artificial intelligence to not only enhance diagnostic precision but also provide a reliable framework for clinical decision-making in bladder cancer management. The findings align with previous studies that have shown similar efficacy of deep learning algorithms in other cancer types, suggesting that AI can be a transformative tool in oncology.\u003c/p\u003e\u003cp\u003eMoreover, the study's robust design, which included a diverse dataset for training and validation, strengthens its findings. The high sensitivity and specificity reported for the model indicate its practical applicability in clinical settings, potentially leading to earlier detection of bladder cancer and improved patient outcomes. This is particularly crucial given the often asymptomatic nature of early-stage bladder cancer, which can lead to delayed diagnoses and poorer prognoses. The integration of deep learning in identifying critical features such as tumor texture and shape further enhances the model's capability to predict malignancy.\u003c/p\u003e\u003cp\u003eFuture research should focus on refining these algorithms and expanding their validation across larger, multi-center cohorts to ensure their generalizability and effectiveness in real-world clinical scenarios. The implications of these findings extend beyond mere diagnostic accuracy; they pave the way for personalized medicine approaches in bladder cancer treatment. By incorporating AI-driven insights into the tumor's biological behavior, clinicians can tailor therapeutic strategies that align more closely with individual patient profiles. For instance, the association between specific imaging features and molecular characteristics of tumors could inform decisions regarding the use of targeted therapies or immunotherapies, thus optimizing treatment outcomes and minimizing unnecessary interventions.\u003c/p\u003e\u003cp\u003eIn conclusion, this study not only demonstrates the feasibility of using deep learning for bladder cancer diagnosis but also emphasizes the need for ongoing research to harness the full potential of AI in enhancing cancer care. The combination of imaging data with advanced analytical techniques represents a promising frontier in the quest for more effective, precision-driven oncology.\u003c/p\u003e\u003cp\u003eThe limitations of this study primarily stem from the relatively small sample size, which may impact the generalizability of the developed model. A larger cohort is essential to validate the model's performance across diverse populations and clinical settings. Additionally, the lack of extensive clinical validation data poses challenges in confirming the model's robustness and applicability in real-world scenarios. Variability in imaging protocols and equipment among different institutions could introduce inter-institutional differences, further complicating the model's reliability. Addressing these limitations through multi-center collaborations and larger, more diverse datasets will be crucial for enhancing the model's predictive power and ensuring its clinical utility.\u003c/p\u003e\u003cp\u003eIn conclusion, this research successfully developed an artificial intelligence diagnostic model based on CT imaging for bladder cancer, demonstrating its potential for improving diagnostic accuracy and prognostic assessment. The integration of radiomics features into clinical practice can provide valuable insights for personalized treatment strategies. Future efforts should focus on expanding the sample size and conducting rigorous clinical validation to establish the model's effectiveness in early diagnosis and tailored therapies for bladder cancer patients. The findings pave the way for advancements in non-invasive diagnostic methodologies that could significantly impact patient outcomes.\u003c/p\u003e\u003cp\u003e\u003cb\u003eLimitations\u003c/b\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eModel Limitations\u003c/b\u003e\u003c/p\u003e\u003cp\u003eAlthough the deep learning model in this study shows promise, it has several limitations. Firstly, despite high accuracy on training data, it may overfit and lack generalizability to diverse populations and imaging conditions due to the limited sample size and specific training data characteristics.\u003c/p\u003e\u003cp\u003e\u003cb\u003eData Quality Concerns\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe model's performance heavily depends on the quality of training data. Issues like noise, artifacts, and inconsistent imaging protocols can negatively impact accuracy. The dataset may not fully capture the diversity of bladder cancer presentations, limiting its real-world applicability.\u003c/p\u003e\u003cp\u003e\u003cb\u003eFuture Directions\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTo overcome these limitations, future research should focus on expanding datasets, improving model interpretability, enhancing data preprocessing techniques, exploring multi-modal approaches, and implementing continuous model updating mechanisms. These efforts will help refine the AI model, making it a more reliable tool for bladder cancer diagnosis in clinical practice.\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e"},{"header":"5. Conclusions","content":"\u003cp\u003eIn conclusion, this research demonstrates the feasibility and potential impact of AI-driven diagnostic tools for bladder cancer. With continued refinement and validation, such models could become integral to bladder cancer detection and personalized treatment planning, ultimately improving patient outcomes and advancing the field of precision medicine.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eThe following abbreviations are used in this manuscript:\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"524\" class=\"fr-table-selection-hover\" style=\"margin-right: calc(32%); width: 68%;\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 20.0166%;\"\u003e\n \u003cp\u003eCT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 79.7926%;\"\u003e\n \u003cp\u003eComputed Tomography\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 20.0166%;\"\u003e\n \u003cp\u003eIOU\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 79.7926%;\"\u003e\n \u003cp\u003eIntersection over Union\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 20.0166%;\"\u003e\n \u003cp\u003eTCIA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 79.7926%;\"\u003e\n \u003cp\u003eThe Cancer Imaging Archive\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 20.0166%;\"\u003e\n \u003cp\u003eAcc\u003c/p\u003e\n \u003cp\u003eGrad-CAM\u003c/p\u003e\n \u003cp\u003eCNNs\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 79.7926%;\"\u003e\n \u003cp\u003eAccuracy\u003c/p\u003e\n \u003cp\u003eGradient-weighted Class Activation Mapping\u003c/p\u003e\n \u003cp\u003econvolutional neural networks\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAuthor Contributions:\u003c/strong\u003e Lei Jiang and Wenyu Ge drafted the manuscript. Lei Jiang and Wenyu Ge contributed equally. Jingru Huo and Shijie Li drew the pictures. Ruijiao Feng and Liu Ji analyzed the data. Tingting Fan reviewed and revised the manuscript. All the authors have read and approved the final version of the manuscript. All the authors contributed to the manuscript and approved the submitted version.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding:\u003c/strong\u003e Please add: This study was supported by the scientific research of the Heilongjiang Provincial Health Commission (grant number 20240909010037), Heilongjiang Postdoctoral Fund (LBH-Z24304) and Heilongjiang provincial scientific research project of traditional Chinese Medicine(ZHY2024-041)\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eInstitutional Review Board Statement:\u0026nbsp;\u003c/strong\u003eThe study was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board of Heilong Provincial Hospital.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eInformed Consent Statement:\u0026nbsp;\u003c/strong\u003eInformed consent was obtained from all subjects involved in the study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability Statement:\u003c/strong\u003e The original contributions of this study are included in the article and supplementary material. Further inquiries can be directed to the corresponding authors. The data is collected in \u003cstrong\u003e\u003cem\u003ehttps://www.cancerimagingarchive.net/collection/tcga-blca/.\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments:\u003c/strong\u003e The authors would like to express their sincere gratitude to the TCGA Research Network for their invaluable contributions to cancer genomics research.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflicts of Interest:\u003c/strong\u003e The authors declare no conflicts of interest.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical trial number:\u003c/strong\u003e not applicable.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eBray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, Jemal A. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024;74(3):229\u0026ndash;63.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSiegel RL, Kratzer TB, Giaquinto AN, Sung H, Jemal A. Cancer statistics, 2025. CA Cancer J Clin. 2025;75(1):10\u0026ndash;45.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003e\u0026Aring;str\u0026ouml;m LM, Sibolt P, Chamberlin H, Serup-Hansen E, Andersen CE, van Herk M, Mouritsen LS, Aznar MC, Behrens CP. Artificial intelligence-generated targets and inter-observer variation in online adaptive radiotherapy of bladder cancer. Phys Imaging Radiat Oncol. 2024;31:100640.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBaressi Šegota S, Lorencin I, Smolić K, Anđelić N, Markić D, Mrzljak V, Štifanić D, Musulin J, Španjol J, Car Z. Semantic Segmentation of Urinary Bladder Cancer Masses from CT Images: A Transfer Learning Approach. Biology (Basel). 2021;10(11).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCao Y, Zhu H, Li Z, Liu C, Ye J. CT Image-Based Radiomic Analysis for Detecting PD-L1 Expression Status in Bladder Cancer Patients. Acad Radiol. 2024;31(9):3678\u0026ndash;87.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCha KH, Hadjiiski L, Chan HP, Weizer AZ, Alva A, Cohan RH, Caoili EM, Paramagul C, Samala RK. Bladder Cancer Treatment Response Assessment in CT using Radiomics with Deep-Learning. Sci Rep. 2017;7(1):8738.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGarapati SS, Hadjiiski L, Cha KH, Chan HP, Caoili EM, Cohan RH, Weizer A, Alva A, Paramagul C, Wei J, Zhou C. Urinary bladder cancer staging in CT urography using machine learning. Med Phys. 2017;44(11):5814\u0026ndash;23.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGirard A, Dercle L, Vila-Reyes H, Schwartz LH, Girma A, Bertaux M, Radulescu C, Lebret T, Delcroix O, Rouanne M. A machine-learning-based combination of criteria to detect bladder cancer lymph node metastasis on [(18)F]FDG PET/CT: a pathology-controlled study. Eur Radiol. 2023;33(4):2821\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePalazzo G, Mangili P, Deantoni C, Fodor A, Broggi S, Castriconi R, Ubeira Gabellini MG, Del Vecchio A, Di Muzio NG, Fiorino C. Real-world validation of Artificial Intelligence-based Computed Tomography auto-contouring for prostate cancer radiotherapy planning. Phys Imaging Radiat Oncol. 2023;28:100501.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWei Z, Xv Y, Liu H, Li Y, Yin S, Xie Y, Chen Y, Lv F, Jiang Q, Li F, Xiao M. A CT-based deep learning model predicts overall survival in patients with muscle invasive bladder cancer after radical cystectomy: a multicenter retrospective cohort study. Int J Surg. 2024;110(5):2922\u0026ndash;32.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eXiong S, Fu Z, Deng Z, Li S, Zhan X, Zheng F, Yang H, Liu X, Xu S, Liu H, Fan B, Dong W, Song Y, Fu B. Machine learning-based CT radiomics enhances bladder cancer staging predictions: A comparative study of clinical, radiomics, and combined models. Med Phys. 2024;51(9):5965\u0026ndash;77.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYang Y, Zou X, Wang Y, Ma X. Application of deep learning as a noninvasive tool to differentiate muscle-invasive bladder cancer and non-muscle-invasive bladder cancer with CT. Eur J Radiol. 2021;139:109666.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYing T, Borrelli P, Edenbrandt L, Enqvist O, Kaboteh R, Tr\u0026auml;g\u0026aring;rdh E, Ul\u0026eacute;n J, Kj\u0026ouml;lhede H. Automated artificial intelligence-based analysis of skeletal muscle volume predicts overall survival after cystectomy for urinary bladder cancer. Eur Radiol Exp. 2021;5(1):50.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang G, Wu Z, Xu L, Zhang X, Zhang D, Mao L, Li X, Xiao Y, Guo J, Ji Z, Sun H, Jin Z. Deep Learning on Enhanced CT Images Can Predict the Muscular Invasiveness of Bladder Cancer. Front Oncol. 2021;11:654685.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang-Yin J, Girard A, Marchal E, Lebret T, Homo Seban M, Uhl M, Bertaux M. PET Imaging in Bladder Cancer: An Update and Future Direction. Pharmaceuticals (Basel). 2023;16(4).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZheng Y, Shi H, Hai B, Zhang J. A commentary on 'A CT-based deep learning model predicts overall survival in patients with muscle invasive bladder cancer after radical cystectomy: a multicenter retrospective cohort study'. Int J Surg. 2024;110(8):5200\u0026ndash;1.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKirk S, Lee Y, Lucchesi FR, Aredes ND, Gruszauskas N, Catto J, Garcia K, Jarosz R, Duddalwar V, Varghese B, Rieger-Christ K, Lemmerman J. The Cancer Genome Atlas Urothelial Bladder Carcinoma Collection (TCGA-BLCA) (Version 8) [Data set]. 2016.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChen X, Zhang Y, Zhou J, Pan Y, Xu H, Shen Y, Cao G, Su MY, Wang M. Combination of Deep Learning Grad-CAM and Radiomics for Automatic Localization and Diagnosis of Architectural Distortion on DBT. Acad Radiol. 2025;32(3):1287\u0026ndash;96.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFanizzi A, Comes MC, Bove S, Cavalera E, de Franco P, Di Rito A, Errico A, Lioce M, Pati F, Portaluri M, Saponaro C, Scognamillo G, Troiano I, Troiano M, Zito FA, Massafra R. Explainable prediction model for the human papillomavirus status in patients with oropharyngeal squamous cell carcinoma using CNN on CT images. Sci Rep. 2024;14(1):14276.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHuang Z, Zhang X, Ju Y, Zhang G, Chang W, Song H, Gao Y. Explainable breast cancer molecular expression prediction using multi-task deep-learning based on 3D whole breast ultrasound. Insights Imaging. 2024;15(1):227.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eShah SJH, Albishri A, Wang R, Lee Y. Integrating local and global attention mechanisms for enhanced oral cancer detection and explainability. Comput Biol Med. 2025;189:109841.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTalaat FM, Gamel SA, El-Balka RM, Shehata M, ZainEldin H. Grad-CAM Enabled Breast Cancer Classification with a 3D Inception-ResNet V2: Empowering Radiologists with Explainable Insights. Cancers (Basel). 2024;16(21).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eXia J, Li J, Wang X, Li Y, Li J. Enhanced cancer classification and critical feature visualization using Raman spectroscopy and convolutional neural networks. Spectrochim Acta Mol Biomol Spectrosc. 2025;326:125242.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAlazwari S, Alsamri J, Alamgeer M, Alotaibi SS, Obayya M, Salama AS. Automated gall bladder cancer detection using artificial gorilla troops optimizer with transfer learning on ultrasound images. Sci Rep. 2024;14(1):21845.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBozkurt B, Coskun K, Bakal G. Building a challenging medical dataset for comparative evaluation of classifier capabilities. Comput Biol Med. 2024;178:108721.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJian Y, Zhang N, Bi Y, Liu X, Fan J, Wu W, Liu T. TC-Sniffer: A Transformer-CNN Bibranch Framework Leveraging Auxiliary VOCs for Few-Shot UBC Diagnosis via Electronic Noses. ACS Sens. 2025;10(1):213\u0026ndash;24.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKurata Y, Nishio M, Moribata Y, Otani S, Himoto Y, Takahashi S, Kusakabe J, Okura R, Shimizu M, Hidaka K, Nishio N, Furuta A, Kido A, Masui K, Onishi H, Segawa T, Kobayashi T, Nakamoto Y. Development of deep learning model for diagnosing muscle-invasive bladder cancer on MRI with vision transformer. Heliyon. 2024;10(16):e36144.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eShalata AT, Alksas A, Shehata M, Khater S, Ezzat O, Ali KM, Gondim D, Mahmoud A, El-Gendy EM, Mohamed MA, Alghamdi NS, Ghazal M, El-Baz A. Precise grading of non-muscle invasive bladder cancer with multi-scale pyramidal CNN. Sci Rep. 2024;14(1):25131.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHolzbeierlein JM, Bixler BR, Buckley DI, Chang SS, Holmes R, James AC, Kirkby E, McKiernan JM, Schuckman AK. Diagnosis and Treatment of Non-Muscle Invasive Bladder Cancer: AUA/SUO Guideline: 2024 Amendment. J Urol. 2024;211(4):533\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYang T, Luo W, Yu J, Zhang H, Hu M, Tian J. Bladder cancer immune-related markers: diagnosis, surveillance, and prognosis. Front Immunol. 2024;15:1481296.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFan J, Chen B, Luo Q, Li J, Huang Y, Zhu M, Chen Z, Li J, Wang J, Liu L, Wei Q, Cao D. Potential molecular biomarkers for the diagnosis and prognosis of bladder cancer. Biomed Pharmacother. 2024;173:116312.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFlaig TW, Spiess PE, Abern M, Agarwal N, Bangs R, Buyyounouski MK, Chan K, Chang SS, Chang P, Friedlander T, Greenberg RE, Guru KA, Herr HW, Hoffman-Censits J, Kaimakliotis H, Kishan AU, Kundu S, Lele SM, Mamtani R, Mian OY, Michalski J, Montgomery JS, Parikh M, Patterson A, Peyton C, Plimack ER, Preston MA, Richards K, Sexton WJ, Siefker-Radtke AO, Stewart T, Sundi D, Tollefson M, Tward J, Wright JL, Cassara CJ, Gurski LA. NCCN Guidelines\u0026reg; Insights: Bladder Cancer, Version 3.2024. J Natl Compr Canc Netw. 2024;22(4):216\u0026ndash;25.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBorna MR, Sepehri MM, Shadpour P, Khaleghi Mehr F. Enhancing bladder cancer diagnosis through transitional cell carcinoma polyp detection and segmentation: an artificial intelligence powered deep learning solution. Front Artif Intell. 2024;7:1406806.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLotan Y, Krishna V, Abuzeid WM, Launer B, Chang SS, Krishna V, Shingi S, Gordetsky JB, Gerald T, Woldu S, Shkolyar E, Hayne D, Redfern A, Spalding L, Stewart C, Eyzaguirre E, Imtiaz S, Narayan VM, Packiam VT, O'Donnell MA, Li R, Baekelandt L, Joniau S, Zuiverloon T, Fernandez MI, Schultz M, Hensley PJ, Allison D, Taylor JA, Hamza A, Kamat A, Nimgaonkar V, Sonawane S, Miller DL, Watson D, Vrabac D, Joshi A, Shah JB, Williams SB. Predicting Response to Intravesical Bacillus Calmette-Gu\u0026eacute;rin in High-Risk Nonmuscle-Invasive Bladder Cancer Using an Artificial Intelligence-Powered Pathology Assay: Development and Validation in an International 12-Center Cohort. J Urol. 2025;213(2):192\u0026ndash;204.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMa X, Zhang Q, He L, Liu X, Xiao Y, Hu J, Cai S, Cai H, Yu B. Artificial intelligence application in the diagnosis and treatment of bladder cancer: advance, challenges, and opportunities. Front Oncol. 2024;14:1487676.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhu M, Gu Z, Chen F, Chen X, Wang Y, Zhao G. Application of artificial intelligence in the diagnosis and treatment of urinary tumors. Front Oncol. 2024;14:1440626.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"discover-oncology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"dion","sideBox":"Learn more about [Discover Oncology](https://www.springer.com/12672)","snPcode":"","submissionUrl":"","title":"Discover Oncology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Discover Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"artificial intelligence, bladder cancer, diagnosis","lastPublishedDoi":"10.21203/rs.3.rs-7760584/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7760584/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eObjectives\u003c/h2\u003e\u003cp\u003eBladder cancer is one of the most common malignancies of the urinary system, and early diagnosis and treatment are crucial for improving patient prognosis.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e\u003cp\u003eIn this study, we developed an artificial intelligence (AI) model for bladder cancer diagnosis using CT imaging data from our hospital and The Cancer Imaging Archive (TCIA). The model was trained and validated using a large dataset of CT images, and its diagnostic accuracy was assessed through various performance metrics. The AI model was constructed using deep learning techniques, which can automatically learn and extract features from CT images. Retrospective CT data from our hospital and TCIA were used to train the AI model. Performance metrics were evaluated, and the Grad-CAM results were reviewed by radiologists.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eThe test - set accuracy exceeded 0.90, and Grad - CAM correctly highlighted tumors.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e\u003cp\u003eThe AI model offers accurate and transparent bladder - cancer detection on routine CT scans and merits prospective validation.\u003c/p\u003e","manuscriptTitle":"Innovative AI Model for Bladder Cancer Diagnosis","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-13 08:03:22","doi":"10.21203/rs.3.rs-7760584/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-11-13T18:25:26+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-11-05T15:23:31+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-11-03T12:18:32+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"216001808651637417694622850719804450285","date":"2025-11-02T23:19:03+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"67797238009874377463961679653900757525","date":"2025-11-02T20:54:45+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"34248039060810031890236471926703961943","date":"2025-11-02T20:18:18+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-11-02T19:43:25+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-10-31T12:47:51+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-10-07T09:33:53+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-10-07T09:33:35+00:00","index":"","fulltext":""},{"type":"submitted","content":"Discover Oncology","date":"2025-10-01T14:49:17+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"discover-oncology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"dion","sideBox":"Learn more about [Discover Oncology](https://www.springer.com/12672)","snPcode":"","submissionUrl":"","title":"Discover Oncology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Discover Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"e37a199e-315e-4d52-a10c-be460aff255a","owner":[],"postedDate":"November 13th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2025-12-10T14:09:23+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-13 08:03:22","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7760584","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7760584","identity":"rs-7760584","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}