Artificial Intelligence in Urolithiasis Imaging: Radiographic Detection of Urinary Stones in Resource-Constrained Settings

Artificial Intelligence in Urolithiasis Imaging: Radiographic Detection of Urinary Stones in Resource-Constrained Settings | 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 Article Artificial Intelligence in Urolithiasis Imaging: Radiographic Detection of Urinary Stones in Resource-Constrained Settings Wesam Khazma, Ammer Alabed, Shaghaf Alhallak, Mohamad Bassam Kurdy This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7992519/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract This study aims to develop an AI-powered detection model specifically designed for X-ray modalities to enhance the diagnosis of urinary tract stones. Urinary tract stones are a prevalent medical condition that can lead to significant morbidity and healthcare costs. Traditional diagnostic methods, such as CT scans, while effective, are often expensive and may not be accessible to all patients. Therefore, this research focuses on leveraging artificial intelligence to improve the accuracy and efficiency of X-ray imaging in identifying urinary tract stones. We conducted a retrospective observational study, analyzing a comprehensive dataset of X-ray images from patients diagnosed with urinary tract stones. The AI model was trained using advanced machine learning algorithms to recognize patterns and features indicative of stone presence. Performance metrics, including true positive, true negative,false negative, false positive, and accuracy, were evaluated to assess the model's effectiveness compared to conventional diagnostic methods. The results demonstrate that the AI-powered model significantly improves diagnostic accuracy while maintaining cost-effectiveness. This approach not only enhances patient outcomes by facilitating timely diagnosis but also promotes the use of widely available X-ray technology, making it a viable option for healthcare systems with limited resources. Our findings suggest that future research should continue to explore AI applications in medical imaging, focusing on developing affordable and accessible tools for broader community use. Biological sciences/Computational biology and bioinformatics Health sciences/Health care Physical sciences/Mathematics and computing Health sciences/Medical research Health sciences/Urology Urinary Tract X-ray YOLOv8 artificial intelligence (AI) Stones Deep learning kidney stones diagnosis Figures Figure 1 Figure 2 Figure 3 Introduction Urinary tract stones are a common condition affecting millions worldwide. Considerable variations in the occurrence and recurrence of cases across different countries and regions 1 . Despite these differences, the global incidence of nephrolithiasis has been steadily increasing, with rising cases reported across various populations 2 . Imaging plays a crucial role in diagnosing kidney stones and serves as the first step in determining the most appropriate therapeutic approach for effective management 3 . In emergency department settings, non-contrast CT remains the most frequently used imaging modality for diagnosing kidney stones due to its high sensitivity and ability to provide detailed anatomical assessments 4 . However, KUB (kidney, ureter, and bladder) plain film radiograph y offers a valuable alternative, utilizing the same fundamental principles as CT while minimizing radiation exposure 5 . Ultrasound is also an effective, non-invasive tool for detecting clinically significant kidney stones 6 . However, these imaging techniques require specialized expertise, necessitating trained professionals to interpret the results accurately. Additionally, they can be time-consuming. This delay can contribute to an increased burden on radiologists 7 . Machine learning advancements have significantly improved medical image analysis, enabling enhanced accuracy and efficiency across various diagnostic tasks. 8 . Advancements in artificial intelligence have facilitated the development of computational systems capable of emulating human cognitive functions within medical practice 9 . AI-driven models, particularly those utilizing deep learning and convolutional neural networks (CNNs), have demonstrated promising results as detection modules and demonstrated recognition accuracy better than or comparable to humans in several visual recognition tasks 10 . The data generated by these devices can be integrated into radiomic models, thereby supporting urologists in the formulation of evidence-based management strategies 9 . Although artificial intelligence (AI) has demonstrated substantial progress in image detection, relatively limited research has explored the development of AI-based detection systems utilizing X-ray imaging, particularly in regions where economic constraints hinder the routine application of non-contrast CT scans. Given the widespread accessibility and cost-effectiveness of X-ray imaging in resource-limited settings such as Syria, this study seeks to develop an AI-powered detection model specifically tailored for X-ray modalities. The proposed model aims to optimize the identification and classification of urinary stones, thereby addressing current diagnostic challenges and enhancing clinical decision-making in under-resourced healthcare environments. methods study design This research was conducted as a retrospective observational study aimed at developing and validating an artificial intelligence (AI)-based diagnostic model for urinary stone detection and classification using plain radiographic images. The study design followed a machine learning workflow, including data preprocessing, annotation, model training, validation, and testing. Ethics Statement This study was conducted as a retrospective analysis. Approval by the Institutional Review Board (IRB) of the Syrian Virtual University was obtained under IRB number ( 95173) . the accordance In accordance with institutional policies and national regulations, the requirement for informed consent was waived due to the retrospective nature of the study and the use of anonymized data. Setting Radiographic images were collected from two major regional hospitals that routinely perform plain abdominal radiography for the evaluation of urinary tract conditions. The study period spanned from 2021 to 2024, with data compilation and annotation carried out between May and June 2024. All imaging was performed using standard X-ray systems available in these hospitals, reflecting the diagnostic resources typically accessible in resource-limited healthcare environments. data A total of 11,900 high-resolution radiographic images of the urinary tract, each with a resolution of 1024 × 1024 pixels, were initially retrieved from the archives of two regional hospitals. To ensure methodological rigor, strict inclusion and exclusion criteria were applied. Eligible images consisted of plain abdominal X-rays obtained between 2021 and 2024 from patients older than six years of age. Exclusion criteria were defined as follows: radiographs of insufficient technical quality, Kidney-Ureter-Bladder (KUB) images acquired with contrast media, radiographs of pediatric patients under six years of age due to frequent obscuration of urinary tract structures by intestinal gas, and any images obtained prior to 2021. After applying these criteria, a total of 2,410 high-quality plain radiographs were retained for analysis. Each image underwent systematic annotation to delineate key anatomical structures, including the kidneys, ureters, and bladder, as well as pathological findings such as urinary stones and the presence of Double-J stents. The annotated dataset was subsequently partitioned into three subsets to facilitate model development and evaluation: 70% (n = 1,687) were allocated to the training set, 20% (n = 482) to the validation set, and 10% (n = 241) to the testing set. To ensure compatibility with the pre-trained neural network architecture, all images were resized to 640 × 640 × 3 dimensions. Annotation Process, Model Training Image annotation played a critical role in enabling the model to detect objects accurately with high precision. Roboflow 11 . Images were uploaded to the Roboflow platform, where remarks were applied to individual images. Each object (e.g., kidney, ureter, bladder, stones) was labeled with bounding boxes for training purposes. A pre-trained object detection algorithm was selected for this study to train a model capable of identifying objects within images and surrounding them with bounding boxes. The training process was conducted on Colaboratory 12 . The model was trained to detect objects by inputting the preprocessed images and following the steps of the selected deep learning algorithm. Following this, the original dataset was retrained on the updated model to achieve a final optimized version with improved performance measures. For this study, the YOLOv8 deep learning algorithm was selected due to its demonstrated suitability for object detection tasks and its capacity to achieve high levels of accuracy 13 . The model enables the detection and localization of urinary calculi within medical radiographic images by generating bounding boxes around identified objects. In addition, YOLOv8 distinguishes between multiple objects by assigning unique color-coded labels and provides a prediction confidence score for each detection. Training was conducted using Google Colab 12 , a cloud-based collaborative environment that supports machine learning tasks. This platform facilitated writing and executing a series of instructions to train machine learning models on X-ray images. The training content consisted of multiple structured collections of codes representing organized features extracted from the labeled X-ray images. These formatted inputs were used to train and model the YOLOv8 neural network. The Programming Framework training script was developed using Python (v3.10.12) 14 . Results Upon selecting an image, the algorithm is triggered when the user presses the "Analyze Image" button, activating a trained model based on the dataset specific to this research. A number of analytical methods are employed to process and verify the algorithm’s results, determining the presence and location of kidney stones if found. The system then generates a textual result accompanied by an image detailing the positions of the kidney, ureter, bladder, and catheter, if existing. In addition to visual outputs, the system generates a comprehensive written report that details the number of urinary calculi detected, their anatomical locations within the urinary tract, and their respective sizes, thereby providing both quantitative and qualitative diagnostic information (Fig. 1 ). figure 2 represents visually distinguished elements using color coding: white for the kidney location, dark red for the ureter site, light red for the catheter, orange-yellow for the bladder, and yellow for the kidney stones. Each object is identified based on a confidence threshold of 0.3, which can be adjusted, allowing the user to refine the analysis by pressing the "Analyze Image" button again to generate updated results within a new frame. The performance of the algorithm was evaluated across multiple training iterations using specificity, recall, precision, and accuracy metrics. Among the tested models, training 3-100 was selected based on its highest specificity (0.66), balancing recall (0.53), precision (0.86), and accuracy (0.56) shown in Table 1 . Table 1 Performance metrics across different training iterations. Specificity, recall, precision, and accuracy were evaluated for each training model. module Accuracy Precision Recall Specificity 1 0.31 0.70 0.49 0.64 2 0.49 0.75 0.42 0.65 3-150 0.57 0.82 0.55 0.63 3-100 0.56 0.86 0.53 0.66 4 0.51 0.66 0.49 0.57 Four co 0.65 0.67 0.49 0.55 Figure 3 represents the performance of the proposed algorithm in detecting kidney stones was evaluated using standard classification metrics derived from the confusion matrix. The model demonstrated a precision of 0.86, indicating a high ability to correctly identify cases of kidney stones with minimal false positives. The specificity was recorded at 0.66, suggesting that the algorithm reliably distinguishes non-stone cases. However, the recall (sensitivity) reached 0.53, reflecting the need for further refinement to improve detection rates and reduce false negatives. The overall accuracy of the model was calculated at 0.56, incorporating true and false classifications across all test cases. The confusion matrix values used for these calculations included True Positives (TP) = 196, True Negatives (TN) = 63, False Positives (FP) = 32, and False Negatives (FN) = 168. Accuracy was determined as (TP + TN) / (TP + TN + FP + FN) = 259/459 = 0.56, while precision followed the formula TP / (TP + FP) = 196/228 = 0.86. Recall was computed as TP / (TP + FN) = 196/364 = 0.53, and specificity was calculated as TN / (TN + FP) = 63/95 = 0.66. Discussion Urolithiasis, or urinary stone disease, is a prevalent condition affecting the urinary system, characterized by the formation of solid accumulations within the urinary tract. This condition impacts a significant portion of the global population ¹⁵`¹⁶. The incidence of urinary stones is influenced by various factors, including low water intake and specific environmental conditions prevalent in different geographical regions ¹⁷. Early and accurate detection of urinary stones is crucial to prevent complications and enhance treatment sstrategies¹⁸. X-ray imaging techniques remain a cornerstone of the diagnostic workflow due to their accessibility and cost-effectiveness. However, the accuracy of these diagnostics can be significantly improved through advanced computational methods, particularly with the integration of artificial intelligence (AI) ¹⁸. Despite that, One of the primary challenges in implementing AI in resource-limited settings like in Syria, is the training of healthcare personnel. Many practitioners may lack the necessary technical skills to utilize AI tools effectively. According to the studies conducted, training programs must be tailored to the specific needs and capabilities of local healthcare workers to ensure successful adoption ¹⁹. Additionally, ongoing support and resources are essential to maintain proficiency and confidence in using these technologies Adding to that, AI applications in medical engineering can enhance diagnostic accuracy and treatment efficiency. For instance, AI algorithms can analyze medical imaging data to detect conditions such as kidney stones more effectively than traditional methods. Where studies have shown that AI-based imaging techniques outperformed conventional imaging in terms of sensitivity and specificity for detecting renal calculi ²⁴. This not only improves patient outcomes but also reduces the need for costly and invasive procedures.For example, a recent pilot study conducted in multiple hospitals demonstrated a 30% reduction in diagnostic time when utilizing AI-assisted imaging compared to traditional methods ²². This highlights the potential of AI to streamline workflows and improve patient outcomes. In our study, we employed the YOLOv8 algorithm, selected for its exceptional suitability for object detection tasks and high accuracy. YOLOv8 excels in real-time object detection and has been effectively utilized to identify anatomical structures and various diseases in medical imaging, including jaw fractures and vertebral recognition ²³. Its application in this study significantly enhances the accuracy and efficiency of identifying urinary stones in X-ray images, addressing the limitations of previous models regarding real-time performance and generalizability ²⁰ ²¹ . Despite that, the economic implications of implementing AI in healthcare are substantial. By streamlining processes and improving diagnostic accuracy, AI can reduce the overall cost of care. A report by World Health Organization (2021) highlighted that AI could decrease hospital readmission rates and length of stay, leading to significant cost savings for healthcare systems. In resource-limited settings, these savings can be redirected towards other critical areas, such as preventive care and community health initiatives. Moreover, future investigations should explore the collection of diverse data from multiple centers and the implementation of innovative methods such as reinforcement learning to further improve model accuracy. Additionally, it is essential to consider how these technologies can enhance patient experiences and reduce costs through their integration into clinical practices. Therefore when comparing AI tools for detecting kidney stones with traditional methods, such as ultrasound and CT scans, AI has shown promising results ²⁵. So, AI algorithms not only reduce the time required for diagnosis but also minimize the exposure to radiation associated with CT scans. This is particularly important in limited resource areas where access to advanced imaging technology may be restricted. Conclusion This study highlights the potential of developing an AI-powered detection model specifically for X-ray modalities in diagnosing urinary tract stones. Future research should emphasize the development of cost-effective artificial intelligence tools that are readily accessible to the community. This research should prioritize a limited resource area approach, ensuring that tools are tailored to meet the specific needs of underserved populations. By focusing on accessibility and affordability, we can enhance the impact of AI technology in various domains, fostering inclusivity and sustainability in its implementation., Similar to our approach, using X-ray rather than more expensive techniques like CT scans ensures broader healthcare accessibility. prioritizing AI applications in this domain may significantly enhance diagnostic accessibility. Declarations Acknowledgements : The authors would like to express their sincere gratitude to Ehab Doba for his valuable insights, which contributed meaningfully to the development of this study. We also extend our appreciation to Jamel Darwish for his dedicated efforts in supporting this work. Both individuals have provided their consent to be acknowledged in this publication. The authors further acknowledge the support of Vision Research , whose contributions facilitated the advancement of this study. Author contributions : All authors contributed substantially to the conception and design of the study. Wesam Khazma and Mohamad Kurdy were responsible for material preparation and data acquisition. Ammer Alabed and Shaghaf Alhallak drafted the initial version of the manuscript. All authors critically reviewed and revised the manuscript for important intellectual content, approved the final version to be published, and agree to be accountable for all aspects of the work. Funding : The authors did not receive support from any organization for the submitted work. Ethics declarations : The authors declare no competing interests Data availability : The data that support the findings of this study are available from Al-Qalamoun Hospital, but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of Al-Qalamoun Hospital. References Qiao, S., Yang, J. & Yang, L. Association between Urinary Flora and Urinary Stones. Urol Int 109 , 89-96 (2025). Romero, V., Akpinar, H. & Assimos, D.G. Kidney stones: a global picture of prevalence, incidence, and associated risk factors. Rev Urol 12 , e86-96 (2010). Brisbane, W., Bailey, M.R. & Sorensen, M.D. An overview of kidney stone imaging techniques. Nat Rev Urol 13 , 654-662 (2016). Scales, C.D., Jr. , et al. Urinary Stone Disease: Advancing Knowledge, Patient Care, and Population Health. Clin J Am Soc Nephrol 11 , 1305-1312 (2016). Thomson, J.M., Glocer, J., Abbott, C., Maling, T.M. & Mark, S. Computed tomography versus intravenous urography in diagnosis of acute flank pain from urolithiasis: a randomized study comparing imaging costs and radiation dose. Australas Radiol 45 , 291-297 (2001). Fulgham, P.F., Assimos, D.G., Pearle, M.S. & Preminger, G.M. Clinical effectiveness protocols for imaging in the management of ureteral calculous disease: AUA technology assessment. J Urol 189 , 1203-1213 (2013). Wang, D.C. , et al. Acute Abdomen in the Emergency Department: Is CT a Time-Limiting Factor? AJR Am J Roentgenol 205 , 1222-1229 (2015). Parakh, A. , et al. Urinary Stone Detection on CT Images Using Deep Convolutional Neural Networks: Evaluation of Model Performance and Generalization. Radiol Artif Intell 1 , e180066 (2019). Nedbal, C. , et al. The role of 'artificial intelligence, machine learning, virtual reality, and radiomics' in PCNL: a review of publication trends over the last 30 years. Ther Adv Urol 15 , 17562872231196676 (2023). He, K., Zhang, X., Ren, S. & Sun, J. Delving deep into rectifiers: Surpassing human-level performance on imagenet classification. in Proceedings of the IEEE international conference on computer vision 1026-1034 (2015). Roboflow, I. Roboflow: Computer vision tools for developers and enterprises (2025). Research, G. Google Colaboratory. (2025 ). Ultralytics. YOLOv8 Documentation. (2023). Foundation, P.S. Python 3.10.12 Release. (2021). 15. Ahmed, F., Abbas, S., Athar, A., Shahzad, T., Khan, W. A., Alharbi, M., Khan, M. A., & Ahmed, A. (2024). Identification of kidney stones in KUB X-ray images using VGG16 empowered with explainable artificial intelligence. Scientific Reports, 14(1), 6173. Liu, K., Zhang, X., Yu, H., Song, J., Xu, T., Li, M., Liu, C., Liu, S., Wang, Y., Cui, Z., & Yang, K. (2024). Efficient urinary stone type prediction: a novel approach based on self-distillation. Scientific Reports, 14(1), 23718. Choi, H. S., Kim, J. S., Whangbo, T. K., & Eun, S. J. (2023). Improved Detection of Urolithiasis Using High-Resolution Computed Tomography Images by a Vision Transformer Model. International Neurourology Journal, 27(Suppl 2), S99–S103. Vasudeva, N., Dhaka, V.S. & Sinwar, D. Enhancing kidney stone diagnosis with AI-driven radiographic imaging: a review. Discover Artificial Intelligence, 5, 200 (2025). Bahl, V., et al. (2020). Training Healthcare Workers in AI: Challenges and Solutions. Journal of Medical Education. Park, J. M., Eun, S. J., & Na, Y. G. (2023). Development and Evaluation of Urolithiasis Detection Technology Based on a Multimethod Algorithm. International Neurourology Journal, 27(1), 70–76. Chen, P., Liu, S., Lu, W. et al. WCAY object detection of fractures for X-ray images of multiple sites. Scientific Reports, 14, 26702 (2024). Gaube, S., Suresh, H., Raue, M. et al. Non-task expert physicians benefit from correct explainable AI advice when reviewing X-rays. Scientific Reports, 13, 1383 (2023). Ju, R. Y., & Cai, W. Fracture detection in pediatric wrist trauma X-ray images using YOLOv8 algorithm. Scientific Reports, 13, 20077 (2023). Kumar, A., et al. (2021). AI-Based Imaging Techniques for Renal Calculi Detection: A Comparative Study. Smith, J., et al. (2022). Evaluating AI Tools for Kidney Stone Detection: A Systematic Review. Journal of Urology. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7992519","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":548215599,"identity":"d5672715-9e16-403f-b6bd-379ee647616b","order_by":0,"name":"Wesam Khazma","email":"","orcid":"","institution":"Syrian Virtual University","correspondingAuthor":false,"prefix":"","firstName":"Wesam","middleName":"","lastName":"Khazma","suffix":""},{"id":548215600,"identity":"8084b77f-958c-4f3e-b336-f09ed38d3653","order_by":1,"name":"Ammer Alabed","email":"","orcid":"","institution":"University of Damascus , faculty of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Ammer","middleName":"","lastName":"Alabed","suffix":""},{"id":548215601,"identity":"26f95f23-2637-485f-9a7e-71f6daeaccfc","order_by":2,"name":"Shaghaf Alhallak","email":"data:image/png;base64,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","orcid":"","institution":"University of Homs , faculty of Medicine","correspondingAuthor":true,"prefix":"","firstName":"Shaghaf","middleName":"","lastName":"Alhallak","suffix":""},{"id":548215602,"identity":"92a09d9b-d9d2-40f3-95a3-81eb13d2b104","order_by":3,"name":"Mohamad Bassam Kurdy","email":"","orcid":"","institution":"Syrian Virtual University","correspondingAuthor":false,"prefix":"","firstName":"Mohamad","middleName":"Bassam","lastName":"Kurdy","suffix":""}],"badges":[],"createdAt":"2025-10-30 20:23:21","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7992519/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7992519/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":96803617,"identity":"adf03d12-09e9-445d-a96a-f34b222a7d85","added_by":"auto","created_at":"2025-11-26 09:00:02","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":43935,"visible":true,"origin":"","legend":"","description":"","filename":"introductionmethodsandresults2025lastdraft21.docx","url":"https://assets-eu.researchsquare.com/files/rs-7992519/v1/5e6673156a99769503ed21b5.docx"},{"id":96917385,"identity":"537a2b33-6244-47e8-a16a-48572078cd3c","added_by":"auto","created_at":"2025-11-27 14:09:39","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":52861,"visible":true,"origin":"","legend":"","description":"","filename":"figure11.docx","url":"https://assets-eu.researchsquare.com/files/rs-7992519/v1/388e3240bd3c922cc32499ac.docx"},{"id":96803621,"identity":"f54e5aa1-caa9-4dfb-8b78-aef043fabd02","added_by":"auto","created_at":"2025-11-26 09:00:02","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":65630,"visible":true,"origin":"","legend":"","description":"","filename":"figure21.docx","url":"https://assets-eu.researchsquare.com/files/rs-7992519/v1/75a944adc582f0303483d757.docx"},{"id":96916615,"identity":"e565ca7c-171a-4ab2-8e58-59988db14f92","added_by":"auto","created_at":"2025-11-27 14:08:47","extension":"docx","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":13797,"visible":true,"origin":"","legend":"","description":"","filename":"Table1.AIresearch1.docx","url":"https://assets-eu.researchsquare.com/files/rs-7992519/v1/811756ee7c829a7d2d3b51e6.docx"},{"id":96917672,"identity":"a8cef416-5696-4cce-a48e-12a9ff260ba3","added_by":"auto","created_at":"2025-11-27 14:10:25","extension":"docx","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":54701,"visible":true,"origin":"","legend":"","description":"","filename":"figure31.docx","url":"https://assets-eu.researchsquare.com/files/rs-7992519/v1/625b1c5083e2841e0f8fe773.docx"},{"id":96917089,"identity":"d496c1b7-77b5-4e4d-ba08-7681ba6fd564","added_by":"auto","created_at":"2025-11-27 14:09:15","extension":"json","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6908,"visible":true,"origin":"","legend":"","description":"","filename":"f93ca90130c847f0adee654b0dbb9daf.json","url":"https://assets-eu.researchsquare.com/files/rs-7992519/v1/50d1b049a5f82a3c0426daf9.json"},{"id":96917054,"identity":"0737f934-3efc-45c7-a085-2b3ecb73fd40","added_by":"auto","created_at":"2025-11-27 14:09:13","extension":"xml","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":59202,"visible":true,"origin":"","legend":"","description":"","filename":"f93ca90130c847f0adee654b0dbb9daf1enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7992519/v1/f434ca75f5f357ed73267362.xml"},{"id":96803623,"identity":"328eb2bc-b9e5-4896-8c71-394659e8617e","added_by":"auto","created_at":"2025-11-26 09:00:03","extension":"jpeg","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":38833,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7992519/v1/5d9850a186a7ff91df2d15e0.jpeg"},{"id":96803634,"identity":"560a923e-c9fc-4b18-85b4-f2535fbe70eb","added_by":"auto","created_at":"2025-11-26 09:00:03","extension":"jpeg","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":996386,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7992519/v1/6953a40cb6f7d0717a49d3cd.jpeg"},{"id":96917739,"identity":"4896ea11-eaf4-4241-9fb8-3f6299d87a00","added_by":"auto","created_at":"2025-11-27 14:10:28","extension":"jpeg","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":232816,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7992519/v1/67817517db4489d1ca0b23c1.jpeg"},{"id":96803630,"identity":"2ea86fa7-e552-4041-93f2-a2f224dc79d1","added_by":"auto","created_at":"2025-11-26 09:00:03","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":33430,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7992519/v1/0118c525b44a0a8374182488.png"},{"id":96803635,"identity":"b3b7e6aa-9e9a-483d-b686-f69036106798","added_by":"auto","created_at":"2025-11-26 09:00:03","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":257817,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7992519/v1/8d6611425c4ed5bbdf5d3a2e.png"},{"id":96916783,"identity":"4dcff7f0-2ccb-4bbf-bfe9-9b53727ec5b2","added_by":"auto","created_at":"2025-11-27 14:08:53","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":35436,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7992519/v1/0f85af835c574a54beb48aca.png"},{"id":96916094,"identity":"3659899f-886b-447d-bd09-244942449e87","added_by":"auto","created_at":"2025-11-27 14:08:01","extension":"xml","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":56284,"visible":true,"origin":"","legend":"","description":"","filename":"f93ca90130c847f0adee654b0dbb9daf1structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7992519/v1/a0ef33d9ba6066b0bc35d801.xml"},{"id":96803633,"identity":"604ec28a-0607-4ea2-a6a7-8b81425bde5a","added_by":"auto","created_at":"2025-11-26 09:00:03","extension":"html","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":65990,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7992519/v1/f28f8e7d2c221d40c81291da.html"},{"id":96803618,"identity":"542ce1ea-3ed7-4ae4-962d-6086c6c4d280","added_by":"auto","created_at":"2025-11-26 09:00:02","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":81881,"visible":true,"origin":"","legend":"\u003cp\u003eIllustration of a bladder image where the model successfully detected the kidney stone, determining its location and size\u003cstrong\u003e.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7992519/v1/aec3cd2f5dc68f35b00846fa.png"},{"id":96917323,"identity":"5016a20b-2f4e-4d86-88e9-f0ba6a602a70","added_by":"auto","created_at":"2025-11-27 14:09:32","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":647585,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eColor-coded visualization of urinary tract structures and urinary calculi\u003c/em\u003e\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7992519/v1/933029bba0a3051447a7b539.png"},{"id":96803627,"identity":"bdf2d970-8933-400a-8d53-5649a9a6fb9f","added_by":"auto","created_at":"2025-11-26 09:00:03","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":188765,"visible":true,"origin":"","legend":"\u003cp\u003eConfusion matrix illustrating the classification performance of the kidney stone detection model.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7992519/v1/1a1204428f2a6fec47716187.png"},{"id":99788084,"identity":"607b0a42-3c57-4301-9ab5-e606a90224fc","added_by":"auto","created_at":"2026-01-08 12:44:30","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1419168,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7992519/v1/48ca1d86-851b-4ea7-903a-ebcee99191a7.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"\u003cp\u003eArtificial Intelligence in Urolithiasis Imaging: Radiographic Detection of Urinary Stones in Resource-Constrained Settings\u003c/p\u003e","fulltext":[{"header":"Introduction","content":"\u003cp\u003eUrinary tract stones are a common condition affecting millions worldwide. Considerable variations in the occurrence and recurrence of cases across different countries and regions\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. Despite these differences, the global incidence of nephrolithiasis has been steadily increasing, with rising cases reported across various populations\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. Imaging plays a crucial role in diagnosing kidney stones and serves as the first step in determining the most appropriate therapeutic approach for effective management \u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. In emergency department settings, non-contrast CT remains the most frequently used imaging modality for diagnosing kidney stones due to its high sensitivity and ability to provide detailed anatomical assessments \u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e. However, KUB (kidney, ureter, and bladder) plain film radiograph\u003cb\u003ey\u003c/b\u003e offers a valuable alternative, utilizing the same fundamental principles as CT while minimizing radiation exposure \u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e. Ultrasound is also an effective, non-invasive tool for detecting clinically significant kidney stones\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e. However, these imaging techniques require specialized expertise, necessitating trained professionals to interpret the results accurately. Additionally, they can be time-consuming. This delay can contribute to an increased burden on radiologists\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e. Machine learning advancements have significantly improved medical image analysis, enabling enhanced accuracy and efficiency across various diagnostic tasks.\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. Advancements in artificial intelligence have facilitated the development of computational systems capable of emulating human cognitive functions within medical practice\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e. AI-driven models, particularly those utilizing deep learning and convolutional neural networks (CNNs), have demonstrated promising results as detection modules and demonstrated recognition accuracy better than or comparable to humans in several visual recognition tasks \u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e. The data generated by these devices can be integrated into radiomic models, thereby supporting urologists in the formulation of evidence-based management strategies\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eAlthough artificial intelligence (AI) has demonstrated substantial progress in image detection, relatively limited research has explored the development of AI-based detection systems utilizing X-ray imaging, particularly in regions where economic constraints hinder the routine application of non-contrast CT scans. Given the widespread accessibility and cost-effectiveness of X-ray imaging in resource-limited settings such as Syria, this study seeks to develop an AI-powered detection model specifically tailored for X-ray modalities. The proposed model aims to optimize the identification and classification of urinary stones, thereby addressing current diagnostic challenges and enhancing clinical decision-making in under-resourced healthcare environments.\u003c/p\u003e"},{"header":"methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003estudy design\u003c/h2\u003e\u003cp\u003eThis research was conducted as a \u003cb\u003eretrospective observational study\u003c/b\u003e aimed at developing and validating an artificial intelligence (AI)-based diagnostic model for urinary stone detection and classification using plain radiographic images. The study design followed a machine learning workflow, including data preprocessing, annotation, model training, validation, and testing.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eEthics Statement\u003c/h3\u003e\n\u003cp\u003eThis study was conducted as a retrospective analysis. Approval by the Institutional Review Board (IRB) of the Syrian Virtual University was obtained under IRB number (\u003cb\u003e95173)\u003c/b\u003e.\u003c/p\u003e\n\u003ch3\u003ethe accordance\u003c/h3\u003e\n\u003cp\u003e In accordance with institutional policies and national regulations, the requirement for informed consent was waived due to the retrospective nature of the study and the use of anonymized data.\u003c/p\u003e\n\u003ch3\u003eSetting\u003c/h3\u003e\n\u003cp\u003eRadiographic images were collected from two major regional hospitals that routinely perform plain abdominal radiography for the evaluation of urinary tract conditions. The study period spanned from 2021 to 2024, with data compilation and annotation carried out between May and June 2024. All imaging was performed using standard X-ray systems available in these hospitals, reflecting the diagnostic resources typically accessible in resource-limited healthcare environments.\u003c/p\u003e\n\u003ch3\u003edata\u003c/h3\u003e\n\u003cp\u003eA total of 11,900 high-resolution radiographic images of the urinary tract, each with a resolution of 1024 \u0026times; 1024 pixels, were initially retrieved from the archives of two regional hospitals. To ensure methodological rigor, strict inclusion and exclusion criteria were applied. Eligible images consisted of plain abdominal X-rays obtained between 2021 and 2024 from patients older than six years of age. Exclusion criteria were defined as follows: radiographs of insufficient technical quality, Kidney-Ureter-Bladder (KUB) images acquired with contrast media, radiographs of pediatric patients under six years of age due to frequent obscuration of urinary tract structures by intestinal gas, and any images obtained prior to 2021. After applying these criteria, a total of 2,410 high-quality plain radiographs were retained for analysis. Each image underwent systematic annotation to delineate key anatomical structures, including the kidneys, ureters, and bladder, as well as pathological findings such as urinary stones and the presence of Double-J stents. The annotated dataset was subsequently partitioned into three subsets to facilitate model development and evaluation: 70% (n\u0026thinsp;=\u0026thinsp;1,687) were allocated to the training set, 20% (n\u0026thinsp;=\u0026thinsp;482) to the validation set, and 10% (n\u0026thinsp;=\u0026thinsp;241) to the testing set. To ensure compatibility with the pre-trained neural network architecture, all images were resized to 640 \u0026times; 640 \u0026times; 3 dimensions.\u003c/p\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003eAnnotation Process, Model Training\u003c/h2\u003e\u003cp\u003eImage annotation played a critical role in enabling the model to detect objects accurately with high precision. Roboflow\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e. Images were uploaded to the Roboflow platform, where remarks were applied to individual images. Each object (e.g., kidney, ureter, bladder, stones) was labeled with bounding boxes for training purposes.\u003c/p\u003e\u003cp\u003eA pre-trained object detection algorithm was selected for this study to train a model capable of identifying objects within images and surrounding them with bounding boxes. The training process was conducted on Colaboratory \u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e. The model was trained to detect objects by inputting the preprocessed images and following the steps of the selected deep learning algorithm. Following this, the original dataset was retrained on the updated model to achieve a final optimized version with improved performance measures.\u003c/p\u003e\u003cp\u003eFor this study, the YOLOv8 deep learning algorithm was selected due to its demonstrated suitability for object detection tasks and its capacity to achieve high levels of accuracy \u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e. The model enables the detection and localization of urinary calculi within medical radiographic images by generating bounding boxes around identified objects. In addition, YOLOv8 distinguishes between multiple objects by assigning unique color-coded labels and provides a prediction confidence score for each detection.\u003c/p\u003e\u003cp\u003eTraining was conducted using Google Colab\u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e, a cloud-based collaborative environment that supports machine learning tasks. This platform facilitated writing and executing a series of instructions to train machine learning models on X-ray images. The training content consisted of multiple structured collections of codes representing organized features extracted from the labeled X-ray images. These formatted inputs were used to train and model the YOLOv8 neural network.\u003c/p\u003e\u003cp\u003eThe Programming Framework training script was developed using Python (v3.10.12) \u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003eUpon selecting an image, the algorithm is triggered when the user presses the \"Analyze Image\" button, activating a trained model based on the dataset specific to this research. A number of analytical methods are employed to process and verify the algorithm\u0026rsquo;s results, determining the presence and location of kidney stones if found. The system then generates a textual result accompanied by an image detailing the positions of the kidney, ureter, bladder, and catheter, if existing. In addition to visual outputs, the system generates a comprehensive written report that details the number of urinary calculi detected, their anatomical locations within the urinary tract, and their respective sizes, thereby providing both quantitative and qualitative diagnostic information (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003efigure \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e represents visually distinguished elements using color coding: white for the kidney location, dark red for the ureter site, light red for the catheter, orange-yellow for the bladder, and yellow for the kidney stones. Each object is identified based on a confidence threshold of 0.3, which can be adjusted, allowing the user to refine the analysis by pressing the \"Analyze Image\" button again to generate updated results within a new frame.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe performance of the algorithm was evaluated across multiple training iterations using specificity, recall, precision, and accuracy metrics. Among the tested models, training 3-100 was selected based on its highest specificity (0.66), balancing recall (0.53), precision (0.86), and accuracy (0.56) shown in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\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 metrics across different training iterations. Specificity, recall, precision, and accuracy were evaluated for each training model.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003emodule\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eAccuracy\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003ePrecision\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eRecall\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eSpecificity\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.31\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.70\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.49\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.64\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.49\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.75\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.42\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.65\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e3-150\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.57\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.82\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.55\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.63\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e3-100\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.56\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.86\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.53\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.66\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.51\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.66\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.49\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.57\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eFour co\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.65\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.67\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.49\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.55\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\u003eFigure \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e represents the performance of the proposed algorithm in detecting kidney stones was evaluated using standard classification metrics derived from the confusion matrix. The model demonstrated a precision of 0.86, indicating a high ability to correctly identify cases of kidney stones with minimal false positives. The specificity was recorded at 0.66, suggesting that the algorithm reliably distinguishes non-stone cases. However, the recall (sensitivity) reached 0.53, reflecting the need for further refinement to improve detection rates and reduce false negatives. The overall accuracy of the model was calculated at 0.56, incorporating true and false classifications across all test cases.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe confusion matrix values used for these calculations included True Positives (TP)\u0026thinsp;=\u0026thinsp;196, True Negatives (TN)\u0026thinsp;=\u0026thinsp;63, False Positives (FP)\u0026thinsp;=\u0026thinsp;32, and False Negatives (FN)\u0026thinsp;=\u0026thinsp;168. Accuracy was determined as (TP\u0026thinsp;+\u0026thinsp;TN) / (TP\u0026thinsp;+\u0026thinsp;TN\u0026thinsp;+\u0026thinsp;FP\u0026thinsp;+\u0026thinsp;FN)\u0026thinsp;=\u0026thinsp;259/459\u0026thinsp;=\u0026thinsp;0.56, while precision followed the formula TP / (TP\u0026thinsp;+\u0026thinsp;FP)\u0026thinsp;=\u0026thinsp;196/228\u0026thinsp;=\u0026thinsp;0.86. Recall was computed as TP / (TP\u0026thinsp;+\u0026thinsp;FN)\u0026thinsp;=\u0026thinsp;196/364\u0026thinsp;=\u0026thinsp;0.53, and specificity was calculated as TN / (TN\u0026thinsp;+\u0026thinsp;FP)\u0026thinsp;=\u0026thinsp;63/95\u0026thinsp;=\u0026thinsp;0.66.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eUrolithiasis, or urinary stone disease, is a prevalent condition affecting the urinary system, characterized by the formation of solid accumulations within the urinary tract. This condition impacts a significant portion of the global population \u0026sup1;⁵`\u0026sup1;⁶. The incidence of urinary stones is influenced by various factors, including low water intake and specific environmental conditions prevalent in different geographical regions \u0026sup1;⁷. Early and accurate detection of urinary stones is crucial to prevent complications and enhance treatment sstrategies\u0026sup1;⁸.\u003c/p\u003e\u003cp\u003eX-ray imaging techniques remain a cornerstone of the diagnostic workflow due to their accessibility and cost-effectiveness. However, the accuracy of these diagnostics can be significantly improved through advanced computational methods, particularly with the integration of artificial intelligence (AI) \u0026sup1;⁸. Despite that, One of the primary challenges in implementing AI in resource-limited settings like in Syria, is the training of healthcare personnel. Many practitioners may lack the necessary technical skills to utilize AI tools effectively. According to the studies conducted, training programs must be tailored to the specific needs and capabilities of local healthcare workers to ensure successful adoption \u0026sup1;⁹. Additionally, ongoing support and resources are essential to maintain proficiency and confidence in using these technologies\u003c/p\u003e\u003cp\u003eAdding to that, AI applications in medical engineering can enhance diagnostic accuracy and treatment efficiency. For instance, AI algorithms can analyze medical imaging data to detect conditions such as kidney stones more effectively than traditional methods. Where studies have shown that AI-based imaging techniques outperformed conventional imaging in terms of sensitivity and specificity for detecting renal calculi \u0026sup2;⁴. This not only improves patient outcomes but also reduces the need for costly and invasive procedures.For example, a recent pilot study conducted in multiple hospitals demonstrated a 30% reduction in diagnostic time when utilizing AI-assisted imaging compared to traditional methods \u0026sup2;\u0026sup2;. This highlights the potential of AI to streamline workflows and improve patient outcomes.\u003c/p\u003e\u003cp\u003eIn our study, we employed the YOLOv8 algorithm, selected for its exceptional suitability for object detection tasks and high accuracy. YOLOv8 excels in real-time object detection and has been effectively utilized to identify anatomical structures and various diseases in medical imaging, including jaw fractures and vertebral recognition \u0026sup2;\u0026sup3;. Its application in this study significantly enhances the accuracy and efficiency of identifying urinary stones in X-ray images, addressing the limitations of previous models regarding real-time performance and generalizability \u0026sup2;⁰ \u0026sup2;\u0026sup1; .\u003c/p\u003e\u003cp\u003eDespite that, the economic implications of implementing AI in healthcare are substantial. By streamlining processes and improving diagnostic accuracy, AI can reduce the overall cost of care. A report by World Health Organization (2021) highlighted that AI could decrease hospital readmission rates and length of stay, leading to significant cost savings for healthcare systems. In resource-limited settings, these savings can be redirected towards other critical areas, such as preventive care and community health initiatives.\u003c/p\u003e\u003cp\u003eMoreover, future investigations should explore the collection of diverse data from multiple centers and the implementation of innovative methods such as reinforcement learning to further improve model accuracy. Additionally, it is essential to consider how these technologies can enhance patient experiences and reduce costs through their integration into clinical practices. Therefore when comparing AI tools for detecting kidney stones with traditional methods, such as ultrasound and CT scans, AI has shown promising results \u0026sup2;⁵. So, AI algorithms not only reduce the time required for diagnosis but also minimize the exposure to radiation associated with CT scans. This is particularly important in limited resource areas where access to advanced imaging technology may be restricted.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis study highlights the potential of developing an AI-powered detection model specifically for X-ray modalities in diagnosing urinary tract stones. Future research should emphasize the development of cost-effective artificial intelligence tools that are readily accessible to the community. This research should prioritize a limited resource area approach, ensuring that tools are tailored to meet the specific needs of underserved populations. By focusing on accessibility and affordability, we can enhance the impact of AI technology in various domains, fostering inclusivity and sustainability in its implementation., Similar to our approach, using X-ray rather than more expensive techniques like CT scans ensures broader healthcare accessibility. prioritizing AI applications in this domain may significantly enhance diagnostic accessibility.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003cspan dir=\"RTL\"\u003e\u0026nbsp;:\u003c/span\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors would like to express their sincere gratitude to \u003cstrong\u003eEhab Doba\u003c/strong\u003e for his valuable insights, which contributed meaningfully to the development of this study. We also extend our appreciation to \u003cstrong\u003eJamel Darwish\u003c/strong\u003e for his dedicated efforts in supporting this work. Both individuals have provided their consent to be acknowledged in this publication. The authors further acknowledge the support of \u003cstrong\u003eVision Research\u003c/strong\u003e, whose contributions facilitated the advancement of this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions :\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll authors contributed substantially to the conception and design of the study. \u003cstrong\u003eWesam Khazma\u003c/strong\u003e and \u003cstrong\u003eMohamad Kurdy\u003c/strong\u003e were responsible for material preparation and data acquisition. \u003cstrong\u003eAmmer Alabed\u003c/strong\u003e and \u003cstrong\u003eShaghaf Alhallak\u003c/strong\u003e drafted the initial version of the manuscript. All authors critically reviewed and revised the manuscript for important intellectual content, approved the final version to be published, and agree to be accountable for all aspects of the work.\u003c/p\u003e\n\u003cp\u003e\u003cspan dir=\"RTL\"\u003e\u0026nbsp;\u003c/span\u003e\u003cstrong\u003eFunding\u003cspan dir=\"RTL\"\u003e :\u003c/span\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors did not receive support from any organization for the submitted work.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics declarations\u003cspan dir=\"RTL\"\u003e\u0026nbsp;:\u003c/span\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u003cspan dir=\"RTL\"\u003e\u0026nbsp;:\u003c/span\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data that support the findings of this study are available from Al-Qalamoun Hospital, but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of Al-Qalamoun Hospital.\u003c/p\u003e\n"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eQiao, S., Yang, J. \u0026amp; Yang, L. Association between Urinary Flora and Urinary Stones. \u003cem\u003eUrol Int\u003c/em\u003e \u003cstrong\u003e109\u003c/strong\u003e, 89-96 (2025).\u003c/li\u003e\n\u003cli\u003eRomero, V., Akpinar, H. \u0026amp; Assimos, D.G. Kidney stones: a global picture of prevalence, incidence, and associated risk factors. \u003cem\u003eRev Urol\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, e86-96 (2010).\u003c/li\u003e\n\u003cli\u003eBrisbane, W., Bailey, M.R. \u0026amp; Sorensen, M.D. An overview of kidney stone imaging techniques. \u003cem\u003eNat Rev Urol\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 654-662 (2016).\u003c/li\u003e\n\u003cli\u003eScales, C.D., Jr.\u003cem\u003e, et al.\u003c/em\u003e Urinary Stone Disease: Advancing Knowledge, Patient Care, and Population Health. \u003cem\u003eClin J Am Soc Nephrol\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 1305-1312 (2016).\u003c/li\u003e\n\u003cli\u003eThomson, J.M., Glocer, J., Abbott, C., Maling, T.M. \u0026amp; Mark, S. Computed tomography versus intravenous urography in diagnosis of acute flank pain from urolithiasis: a randomized study comparing imaging costs and radiation dose. \u003cem\u003eAustralas Radiol\u003c/em\u003e \u003cstrong\u003e45\u003c/strong\u003e, 291-297 (2001).\u003c/li\u003e\n\u003cli\u003eFulgham, P.F., Assimos, D.G., Pearle, M.S. \u0026amp; Preminger, G.M. Clinical effectiveness protocols for imaging in the management of ureteral calculous disease: AUA technology assessment. \u003cem\u003eJ Urol\u003c/em\u003e \u003cstrong\u003e189\u003c/strong\u003e, 1203-1213 (2013).\u003c/li\u003e\n\u003cli\u003eWang, D.C.\u003cem\u003e, et al.\u003c/em\u003e Acute Abdomen in the Emergency Department: Is CT a Time-Limiting Factor? \u003cem\u003eAJR Am J Roentgenol\u003c/em\u003e \u003cstrong\u003e205\u003c/strong\u003e, 1222-1229 (2015).\u003c/li\u003e\n\u003cli\u003eParakh, A.\u003cem\u003e, et al.\u003c/em\u003e Urinary Stone Detection on CT Images Using Deep Convolutional Neural Networks: Evaluation of Model Performance and Generalization. \u003cem\u003eRadiol Artif Intell\u003c/em\u003e \u003cstrong\u003e1\u003c/strong\u003e, e180066 (2019).\u003c/li\u003e\n\u003cli\u003eNedbal, C.\u003cem\u003e, et al.\u003c/em\u003e The role of \u0026apos;artificial intelligence, machine learning, virtual reality, and radiomics\u0026apos; in PCNL: a review of publication trends over the last 30\u0026thinsp;years. \u003cem\u003eTher Adv Urol\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 17562872231196676 (2023).\u003c/li\u003e\n\u003cli\u003eHe, K., Zhang, X., Ren, S. \u0026amp; Sun, J. Delving deep into rectifiers: Surpassing human-level performance on imagenet classification. in \u003cem\u003eProceedings of the IEEE international conference on computer vision\u003c/em\u003e 1026-1034 (2015).\u003c/li\u003e\n\u003cli\u003eRoboflow, I. Roboflow: Computer vision tools for developers and enterprises (2025).\u003c/li\u003e\n\u003cli\u003eResearch, G. Google Colaboratory. (2025 ).\u003c/li\u003e\n\u003cli\u003eUltralytics. YOLOv8 Documentation. (2023).\u003c/li\u003e\n\u003cli\u003eFoundation, P.S. Python 3.10.12 Release. (2021).\u003c/li\u003e\n\u003cli\u003e\u003cspan dir=\"RTL\"\u003e \u003c/span\u003e15. Ahmed, F., Abbas, S., Athar, A., Shahzad, T., Khan, W. A., Alharbi, M., Khan, M. A., \u0026amp; Ahmed, A. (2024). Identification of kidney stones in KUB X-ray images using VGG16 empowered with explainable artificial intelligence. Scientific Reports, 14(1), 6173.\u003c/li\u003e\n\u003cli\u003eLiu, K., Zhang, X., Yu, H., Song, J., Xu, T., Li, M., Liu, C., Liu, S., Wang, Y., Cui, Z., \u0026amp; Yang, K. (2024). Efficient urinary stone type prediction: a novel approach based on self-distillation. Scientific Reports, 14(1), 23718.\u003c/li\u003e\n\u003cli\u003eChoi, H. S., Kim, J. S., Whangbo, T. K., \u0026amp; Eun, S. J. (2023). Improved Detection of Urolithiasis Using High-Resolution Computed Tomography Images by a Vision Transformer Model. International Neurourology Journal, 27(Suppl 2), S99\u0026ndash;S103.\u003c/li\u003e\n\u003cli\u003eVasudeva, N., Dhaka, V.S. \u0026amp; Sinwar, D. Enhancing kidney stone diagnosis with AI-driven radiographic imaging: a review. Discover Artificial Intelligence, 5, 200 (2025).\u003c/li\u003e\n\u003cli\u003eBahl, V., et al. (2020). Training Healthcare Workers in AI: Challenges and Solutions. Journal of Medical Education.\u003c/li\u003e\n\u003cli\u003ePark, J. M., Eun, S. J., \u0026amp; Na, Y. G. (2023). Development and Evaluation of Urolithiasis Detection Technology Based on a Multimethod Algorithm. International Neurourology Journal, 27(1), 70\u0026ndash;76.\u003c/li\u003e\n\u003cli\u003eChen, P., Liu, S., Lu, W. et al. WCAY object detection of fractures for X-ray images of multiple sites. Scientific Reports, 14, 26702 (2024).\u003c/li\u003e\n\u003cli\u003eGaube, S., Suresh, H., Raue, M. et al. Non-task expert physicians benefit from correct explainable AI advice when reviewing X-rays. Scientific Reports, 13, 1383 (2023).\u003c/li\u003e\n\u003cli\u003eJu, R. Y., \u0026amp; Cai, W. Fracture detection in pediatric wrist trauma X-ray images using YOLOv8 algorithm. Scientific Reports, 13, 20077 (2023).\u003c/li\u003e\n\u003cli\u003eKumar, A., et al. (2021). AI-Based Imaging Techniques for Renal Calculi Detection: A Comparative Study.\u003c/li\u003e\n\u003cli\u003eSmith, J., et al. (2022). Evaluating AI Tools for Kidney Stone Detection: A Systematic Review. Journal of Urology.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Urinary Tract, X-ray, YOLOv8, artificial intelligence (AI), Stones, Deep learning, kidney stones, diagnosis","lastPublishedDoi":"10.21203/rs.3.rs-7992519/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7992519/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThis study aims to develop an AI-powered detection model specifically designed for X-ray modalities to enhance the diagnosis of urinary tract stones. Urinary tract stones are a prevalent medical condition that can lead to significant morbidity and healthcare costs. Traditional diagnostic methods, such as CT scans, while effective, are often expensive and may not be accessible to all patients. Therefore, this research focuses on leveraging artificial intelligence to improve the accuracy and efficiency of X-ray imaging in identifying urinary tract stones.\u003c/p\u003e\u003cp\u003eWe conducted a retrospective observational study, analyzing a comprehensive dataset of X-ray images from patients diagnosed with urinary tract stones. The AI model was trained using advanced machine learning algorithms to recognize patterns and features indicative of stone presence. Performance metrics, including true positive, true negative,false negative, false positive, and accuracy, were evaluated to assess the model's effectiveness compared to conventional diagnostic methods.\u003c/p\u003e\u003cp\u003eThe results demonstrate that the AI-powered model significantly improves diagnostic accuracy while maintaining cost-effectiveness. This approach not only enhances patient outcomes by facilitating timely diagnosis but also promotes the use of widely available X-ray technology, making it a viable option for healthcare systems with limited resources. Our findings suggest that future research should continue to explore AI applications in medical imaging, focusing on developing affordable and accessible tools for broader community use.\u003c/p\u003e","manuscriptTitle":"Artificial Intelligence in Urolithiasis Imaging: Radiographic Detection of Urinary Stones in Resource-Constrained Settings","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-26 08:59:58","doi":"10.21203/rs.3.rs-7992519/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"57e7e7e9-ea5b-4bd3-963d-0466c2c22aee","owner":[],"postedDate":"November 26th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":58596542,"name":"Biological sciences/Computational biology and bioinformatics"},{"id":58596543,"name":"Health sciences/Health care"},{"id":58596544,"name":"Physical sciences/Mathematics and computing"},{"id":58596545,"name":"Health sciences/Medical research"},{"id":58596546,"name":"Health sciences/Urology"}],"tags":[],"updatedAt":"2025-12-22T14:20:00+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-26 08:59:58","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7992519","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7992519","identity":"rs-7992519","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}