Research on an Optimized Deep Learning-Based Classification Model for Ovarian Cyst Ultrasound Images

Research on an Optimized Deep Learning-Based Classification Model for Ovarian Cyst Ultrasound Images | 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 Research on an Optimized Deep Learning-Based Classification Model for Ovarian Cyst Ultrasound Images 维梅 李, 玉华 夏, 咏曌 李, 星 伍, 琳 师 This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8756530/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 15 Apr, 2026 Read the published version in Journal of Ovarian Research → Version 1 posted 9 You are reading this latest preprint version Abstract Background Ovarian cysts are a common pelvic disorder in women, and accurate differentiation between benign and malignant types is essential for guiding treatment decisions and prognostic evaluations. However, traditional ultrasound examinations heavily depend on the operator's experience, introducing subjectivity and diagnostic inconsistencies. In recent years, deep learning technologies have demonstrated strong potential in intelligent medical imaging diagnostics, offering innovative solutions for automated and precise classification of ovarian cysts. Results Compared to subjective evaluations by senior ultrasound physicians (accuracy: 76.5%) and the O-RADS classification system (accuracy: 87.8%), the DenseNet121 model exhibited significant advantages in overall diagnostic efficacy (P < 0.05). Conclusions Deep learning models based on ultrasound images can effectively address noise and feature complexity in such imaging, enabling high-precision classification of benign and malignant ovarian cysts. These models hold strong potential for clinical adoption, providing physicians with objective and reliable decision-making support. Ovarian cyst Deep learning O-RADS Feature extraction Figures Figure 1 Figure 2 Figure 3 Figure 4 Background In recent years, deep learning technologies have demonstrated strong potential in intelligent medical imaging diagnostics [ 1 ] , offering innovative solutions for automated and precise classification of ovarian cysts [ 2 ] .Ovarian cysts represent a prevalent gynecological condition in women [ 3 ] , and accurate differentiation between benign and malignant types is critical for developing individualized treatment strategies and improving patient prognosis [ 4 ] . However, clinical diagnosis remains challenging, as some patients are preoperatively suspected of malignancy but postoperatively confirmed as benign, leading to overtreatment and psychological burden. Ultrasound examination [ 5 ] ,due to its non-invasive, convenient, and cost-effective nature, serves as the preferred imaging modality for initial screening and follow-up of ovarian cysts. Nevertheless, diagnostic outcomes are heavily influenced by the operator's experience and subjective judgment, resulting in substantial inter-observer variability. To enhance diagnostic consistency and accuracy, the Ovarian-Adnexal Reporting and Data System (O-RADS) has been internationally proposed [ 6 ] , standardizing ultrasound features into risk stratification categories to provide clinical references [ 7 ] . However, O-RADS exhibits uncertainty in diagnosing category 4 ovarian masses [ 8 ] , with approximately half of those classified as malignant later confirmed as benign postoperatively, indicating a potential issue of overdiagnosis. Additionally, while serum tumor markers can serve as supplementary indicators [ 9 ] , their sensitivity and specificity are limited when used in isolation. In recent years, deep learning has achieved remarkable advancements in medical image analysis [ 10 ] , enabling automated learning and extraction of high-dimensional image features while minimizing biases from manual feature selection [ 11 ] . This offers novel approaches for intelligent diagnosis of complex images. Existing studies have demonstrated that convolutional neural networks (CNNs) and their variants achieve excellent classification performance in lesion identification from ultrasound images of the breast, thyroid, and prostate [ 12 ] .However, research on deep learning for differentiating benign and malignant ovarian cysts remains relatively limited [ 13 ] , with a lack of direct comparative studies against O-RADS and clinical expert assessments. Ovarian cysts are a common pelvic disorder in women [ 14 ] .This study, based on ultrasound images from pathologically confirmed ovarian cysts [ 15 ] , constructed multiple end-to-end deep learning models [ 16 ] , including DenseNet121, ResNet34, GoogLeNet [ 17 ] , and Vision Transformer (ViT). Results indicated that the DenseNet121 model outperformed the others, achieving area under the curve (AUC) values of 0.914 and 0.913 on the training and validation sets, respectively. It also demonstrated superior accuracy, sensitivity, and specificity on the validation set. Grad-CAM visualizations further validated the model's reliability in extracting features from lesion regions, showing high consistency with clinical lesion distributions. Furthermore, the diagnostic efficacy of this model was compared with subjective assessments by senior ultrasound physicians and O-RADS grading results [ 18 ] , aiming to explore the application value of deep learning in differentiating benign and malignant ovarian cysts and to provide clinicians with more objective and reliable decision-support tools [ 19 ] . Compared to subjective evaluations by senior ultrasound physicians (accuracy: 76.5%) and the O-RADS classification system [ 20 ] (accuracy: 87.8%), the DenseNet121 model exhibited significant advantages in overall diagnostic efficacy (P < 0.05). Methods 2.1 Study Population This retrospective study included 327 patients with ovarian cysts who underwent surgery and received pathological confirmation at our institution from January 2023 to January 2025. All patients had complete preoperative two-dimensional ultrasound images and clinical data, comprising 249 benign cases and 78 malignant cases. The dataset was randomly divided into a training set (196 cases) and a validation set (131 cases) in a 6:4 ratio. Patients who were pregnant, lactating, or had a history of severe cardiac disease were excluded. 2.2 Data Collection and Preprocessing All patients underwent pelvic ultrasound examinations using the GE Voluson E8 color Doppler ultrasound system, equipped with a transvaginal probe (RIC5-9-D) and an abdominal probe (RAB2-5-D). Images were acquired in transverse, sagittal, and magnified lesion views, documenting details such as lesion number, size, morphology, borders, wall thickness, papillary projections, septations, blood flow signals, resistance index, and ascites. Images were stored in DICOM format, anonymized by removing patient information, and imported into the deep learning framework [ 21 ] . Preprocessing involved resizing to 224×224 pixels, pixel value normalization, and data augmentation techniques including random rotation, translation, and flipping to enhance model generalization. 2.3 Model Construction and Training Four end-to-end deep learning models—DenseNet121, ResNet34, GoogLeNet [ 22 ] , and Vision Transformer (ViT)—were selected as candidate networks. All models were initialized with ImageNet pre-trained weights and fine-tuned via transfer learning on the training set. Binary cross-entropy was used as the loss function, with Adam as the optimizer, an initial learning rate of 1e-4, a batch size of 32, and training for 100 epochs. AUC was monitored on the validation set, with an early stopping strategy implemented to prevent overfitting. 2.4 Performance Evaluation Metrics Postoperative pathological diagnosis served as the gold standard. Receiver operating characteristic (ROC) curves were plotted, and the area under the curve (AUC) with 95% confidence intervals was calculated. Additional metrics, including accuracy, sensitivity, specificity, precision, recall, and F1 score, were computed to assess classification performance. Grad-CAM was applied to the optimal model for visualization analysis, examining the consistency between the model's attention regions and clinical lesion distributions. 2.5 Statistical Analysis Statistical analyses were performed using SPSS 26.0 and Python. Continuous data are presented as mean ± standard deviation (x ± s), and categorical data as percentages (%). AUC comparisons between models were conducted using the DeLong test, while classification metrics such as accuracy were compared via chi-square or McNemar tests [ 23 ] . A p-value < 0.05 was considered statistically significant. Results 3.1 Patient Baseline Characteristics This study included a total of 327 patients with pathologically confirmed ovarian cysts, comprising 249 benign cases (76.15%) and 78 malignant cases (23.85%). The cohort was randomly divided into a training set of 196 cases (149 benign, 47 malignant) and a validation set of 131 cases (100 benign, 31 malignant) in a 6:4 ratio. No statistically significant differences were observed between the two groups in terms of age, lesion size, or pathological type distribution (P > 0.05), ensuring comparability. 3.2 Comparison of Classification Performance Among Models The classification performance of the four deep learning models is summarized in Table 1 . The DenseNet121 model demonstrated the highest AUC and overall classification performance on both the training and validation sets, with an AUC of 0.914 (95% CI: 0.8733–0.9548) on the training set and 0.913 (95% CI: 0.8584–0.9677) on the validation set. Its validation set accuracy, sensitivity, and specificity reached 87.8%, 82.6%, and 88.9%, respectively, which were significantly superior to those of ResNet34, GoogLeNet, and ViT (P < 0.05). Table 1 Classification Performance of the Four Deep Learning Models ModelName Acc AUC 95% CI Sensitivity Specificity Precision Recall Cohort densenet121 0.837 0.914 0.8733–0.9548 0.907 0.81 0.645 0.907 Train densenet121 0.878 0.913 0.8584–0.9677 0.826 0.889 0.613 0.826 Test resnet34 0.837 0.903 0.8605–0.9450 0.833 0.838 0.662 0.833 Train resnet34 0.649 0.844 0.7537–0.9351 0.957 0.583 0.328 0.957 Test googlenet 0.393 0.491 0.4005–0.5824 0.833 0.225 0.29 0.833 Train googlenet 0.435 0.524 0.3949–0.6538 0.739 0.37 0.2 0.739 Test ViT 0.276 0.407 0.3185–0.4954 1 0 0.276 1 Train ViT 0.817 0.628 0.4786–0.7774 0.435 0.898 0.476 0.435 Test 3.3 ROC Curve Analysis Figures 1 and 2 illustrate the ROC curves for the GoogLeNet and ViT models on the training and validation sets, respectively, where their AUC values approached random classification levels, indicating poor discriminative ability for benign versus malignant ovarian cysts. In contrast, the ROC curves for DenseNet121 (Fig. 3 ) and ResNet34 (Fig. 4 ) shifted markedly toward the upper-left corner, with DenseNet121 exhibiting the largest area under the curve and the optimal diagnostic performance. 3.4 Grad-CAM Visualization Results Several representative cases from the validation set were selected for Grad-CAM visualization of the DenseNet121 model's attention regions. The results revealed that the model primarily focused on lesion wall thickness, papillary projections, and septations, which aligned closely with key pathological signs. This suggests strong consistency between the model's decision-making basis and clinical observations (see Fig. 3 ). Figure 3 . Grad-CAM Visualization Results for the DenseNet121 Model : Red areas indicate the model's attention regions, showing high overlap with lesion distributions. 3.5 Comparative Analysis with O-RADS and Expert Assessments The DenseNet121 model's results were compared with subjective assessments by senior ultrasound physicians and the O-RADS grading system [ 24 ] . The model's accuracy on the validation set (87.8%) was significantly higher than that of manual assessments (76.5%) and slightly superior to O-RADS grading (87.8%), with statistically significant differences (P < 0.05). The model also demonstrated greater balance in sensitivity and specificity, reducing clinical risks associated with overdiagnosis or missed diagnoses. 3.6 Summary Overall, the DenseNet121 model exhibited the highest diagnostic performance and generalization capability in classifying benign and malignant ovarian cysts, performing well on both the training and validation sets. It effectively focused on key imaging features relevant to pathological diagnosis, outperforming traditional O-RADS classification and manual assessments [ 25 ] . This model holds substantial value for clinical promotion and as an auxiliary decision-making tool. Discussion This study, based on ultrasound images from 327 pathologically confirmed ovarian cyst cases, systematically compared the performance of four deep learning models—DenseNet121, ResNet34, GoogLeNet [ 26 ] , and Vision Transformer (ViT)—in classifying benign and malignant ovarian cysts. These models were also benchmarked against O-RADS grading and subjective assessments by senior ultrasound physicians [ 27 ] . The results demonstrated that DenseNet121 outperformed the other models in key metrics such as AUC, sensitivity, specificity, and accuracy. Its diagnostic performance not only surpassed manual evaluations but also exceeded O-RADS grading, indicating that deep learning-based automated diagnostic models offer significant advantages in differentiating benign and malignant ovarian cysts [ 28 ] . First, the superior performance of DenseNet121 may be closely linked to its architectural features. DenseNet facilitates feature reuse and efficient gradient propagation through dense cross-layer connections, enabling the capture of fine-grained features in ultrasound images. This is particularly well-suited for handling medical images with complex textures and subtle boundaries. In contrast, while ResNet34 incorporates residual connections, its shallower network depth limits feature representation capabilities, resulting in reduced accuracy on the validation set. The poorer performance of GoogLeNet and ViT in this study could be attributed to the relatively limited ultrasound image dataset, ViT's strong dependence on large-scale data, and GoogLeNet's outdated network design. These findings underscore the importance of selecting architectures optimized for small-sample learning in medical imaging scenarios. Compared to O-RADS grading, the DenseNet121 model achieved a better balance between sensitivity and specificity. Although O-RADS provides a standardized grading system that effectively reduces diagnostic variability among operators [ 29 ] , it exhibits ambiguity in risk assessment for category 4 lesions, often leading to overtreatment in clinical practice. The results of this study show that deep learning models, by automatically extracting imaging features, can more precisely differentiate between 4a and 4b subcategory lesions, thereby minimizing misdiagnoses. Relative to subjective assessments by experienced ultrasound physicians, the model not only delivered higher accuracy but also enabled rapid, stable, and reproducible diagnoses, mitigating the impact of inter-individual experience variations. [ 11 ] Furthermore, the Grad-CAM visualization results provided additional evidence of the DenseNet121 model's interpretability. The model's attention regions were primarily focused on lesion wall thickness, papillary projections, and areas with abundant blood flow—key indicators in the clinical diagnosis of malignant ovarian tumors. These findings suggest that deep learning models can go beyond "black-box" classifications to offer reasonable visual explanations, enhancing clinicians' trust in the model and facilitating its integration into real-world clinical workflows. Although the results of this study hold certain clinical reference value, several limitations remain. First, as a single-center retrospective study with a relatively limited sample size—particularly with a low proportion of malignant cases—it may affect the model's generalizability. Second, the study relied solely on two-dimensional static ultrasound images as input, excluding three-dimensional ultrasound or dynamic video data, which could overlook spatial or temporal features. Additionally, the comparisons were limited to experts from a single institution and O-RADS grading results; future studies could incorporate multi-center and multi-physician involvement to strengthen external validity. Future research could explore the following directions: (1) expanding sample sizes and incorporating multi-center datasets to further enhance model robustness and generalization; (2) investigating multimodal deep learning models that integrate diverse data sources, such as ultrasound video sequences, serum tumor markers [ 30 ] , and patient clinical information, to achieve superior diagnostic efficacy; (3) examining the model's performance across different ovarian tumor subtypes and stages, while assessing its potential in early ovarian cancer screening; and (4) combining explainable artificial intelligence techniques to improve model transparency and clinical acceptability, thereby promoting its deployment in clinical decision-support systems. Conclusions In summary, the DenseNet121 deep learning model developed in this study exhibited excellent performance in classifying benign and malignant ovarian cysts, significantly improving diagnostic accuracy and consistency while reducing subjective errors. It holds strong potential as a clinical auxiliary diagnostic tool. With further advancements in multi-center validation and multimodal integration, deep learning is poised to become a vital complement to ovarian tumor risk stratification and treatment decision-making, supporting the evolution of precision medicine. Declarations Acknowledgements We are grateful to Lichang Zhong and Kaihua Fan for their valuable contributions to subject recruitment and eligibility assessment, patient liaison and longitudinal follow-up, and the management of study documentation. Authors’ contributions W.M.L. and Y.H.X. designed the project. W.M.L. ,Y.H.X., Y.Z.L. ,X.W. and L.S. performed experiments. W.M.L wrote the manuscript. W.M.L., Y.H.X. ,Y.Z.L., X.W., and L.S. collected the samples. W.M.L., Y.H.X. , Y.Z.L., X.W., and L.S. analyzed the data. W.M.L., Y.H.X. , Y.Z.L. reviewed and edited the manuscript. W.M.L.acquired funding Funding This work was supported by the Shanghai Key Clinical Research Center Fund[grant number 2023ZZ02006] Data availability The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Ethics approval and consent to participate This study was approved by the Institutional Review Board (IRB) of the Sixth People's Hospital of Shanghai, China (Approval No.:2025-KY-105(K) Date of Review:2025-03-27) . The study was conducted in accordance with the Good Clinical Practice (GCP) guidelines and the Ethical Review Guidelines for Drug Clinical Trials issued by the National Medical Products Administration (NMPA) of the People’s Republic of China, as well as the principles of ICH-GCP. It also adheres to the requirements of the Declaration of Helsinki and complies with relevant laws and regulations in China.Written informed consent was obtained from all individual participants included in the study. Consent for publication Not applicable. Competing interests The authors declare that they have no competing interests. Disclosure of Interests The authors declare no conflict of interests. Author details 1 Department of Ultrasound in Medicine, Shanghai Sixth People's Hospital Affiliated to Shanghai 6 Jiao Tong University School of Medicine, Shanghai, 201306, China References BARRERA F J, BROWN E D L, ROJO A, et al. Application of machine learning and artificial intelligence in the diagnosis and classification of polycystic ovarian syndrome: a systematic review [J]. Front Endocrinol (Lausanne), 2023, 14: 1106625. ANDREOTTI R F, TIMMERMAN D, STRACHOWSKI L M, et al. O-RADS US Risk Stratification and Management System: A Consensus Guideline from the ACR Ovarian-Adnexal Reporting and Data System Committee [J]. Radiology, 2020, 294(1): 168-85. XIE W T, WANG Y Q, XIANG Z S, et al. Efficacy of IOTA simple rules, O-RADS, and CA125 to distinguish benign and malignant adnexal masses [J]. J Ovarian Res, 2022, 15(1): 15. SHANMUGAVADIVEL K, M S M, T R M, et al. Optimized polycystic ovarian disease prognosis and classification using AI based computational approaches on multi-modality data [J]. BMC Med Inform Decis Mak, 2024, 24(1): 281. SUHA S A, ISLAM M N. An extended machine learning technique for polycystic ovary syndrome detection using ovary ultrasound image [J]. Sci Rep, 2022, 12(1): 17123. STRACHOWSKI L M, JHA P, PHILLIPS C H, et al. O-RADS US v2022: An Update from the American College of Radiology's Ovarian-Adnexal Reporting and Data System US Committee [J]. Radiology, 2023, 308(3): e230685. PHILLIPS C H, PATEL-LIPPMANN K, HUANG J, et al. Ovarian-Adnexal Reporting and Data System Ultrasound v2022: From Origin to Everyday Use [J]. Radiol Clin North Am, 2025, 63(1): 29-44. SHEN L, SADOWSKI E A, GUPTA A, et al. The Ovarian-Adnexal Reporting and Data System (O-RADS) US Score Effect on Surgical Resection Rate [J]. Radiology, 2024, 313(1): e240044. PANG C, MA Y, SHI W, et al. Prognostic significance of serum tumor markers in various pathologic subtypes of gastric cancer [J]. J Gastrointest Surg, 2024, 28(5): 694-702. BUDDENKOTTE T, RUNDO L, WOITEK R, et al. Deep learning-based segmentation of multisite disease in ovarian cancer [J]. Eur Radiol Exp, 2023, 7(1): 77. REKA S, PRABA T S, PRASANNA M, et al. Automated high precision PCOS detection through a segment anything model on super resolution ultrasound ovary images [J]. Sci Rep, 2025, 15(1): 16832. AGYEKUM E A, WANG Y G, ISSAKA E, et al. Predicting the efficacy of microwave ablation of benign thyroid nodules from ultrasound images using deep convolutional neural networks [J]. BMC Med Inform Decis Mak, 2025, 25(1): 161. WANG R, CAI Y, LEE I K, et al. Evaluation of a convolutional neural network for ovarian tumor differentiation based on magnetic resonance imaging [J]. Eur Radiol, 2021, 31(7): 4960-71. DOKRAS A. Polycystic ovary syndrome in 2025-insights and innovations [J]. Fertil Steril, 2025, 124(5 Pt 2): 907-9. RUAN Z, YU Z, QIN Q, et al. Diagnostic value of ultrasound elastography in polycystic ovary syndrome: a systematic review and meta-analysis [J]. Gynecol Endocrinol, 2024, 40(1): 2352139. DUTTA P, SATHI K A, HOSSAIN M A, et al. Conv-ViT: A Convolution and Vision Transformer-Based Hybrid Feature Extraction Method for Retinal Disease Detection [J]. J Imaging, 2023, 9(7). SHA M. Segmentation of ovarian cyst in ultrasound images using AdaResU-net with optimization algorithm and deep learning model [J]. Sci Rep, 2024, 14(1): 18868. YI Y Y, LI C, ZHU W J, et al. Diagnostic performance of contrast-enhanced ultrasound (CEUS) combined with Ovarian-Adnexal Reporting and Data System (O-RADS) ultrasound risk stratification for adnexal masses: a systematic review and meta-analysis [J]. Clin Radiol, 2024, 79(9): e1167-e75. JEYASHANKER P, GEORGEWILLIAM SUNDARAM A G V, SADAGOPAN P, et al. Advanced holographic convolutional dense networks and Tangent runner optimization for enhanced polycystic ovarian disease classification [J]. Sci Rep, 2025, 15(1): 15719. GUO Y, STRACHOWSKI L M. Beyond the AJR: O-RADS Ultrasound Risk Stratification Performs Within Expected Range With Excellent Diagnostic Performance in Nonselected Female Patients in the United States [J]. AJR Am J Roentgenol, 2023, 220(3): 450. ZHAO B, WEN L, HUANG Y, et al. A Deep Learning-Based Automatic Recognition Model for Polycystic Ovary Ultrasound Images [J]. Balkan Med J, 2025, 42(5): 419-28. FANG D, HUANG Y, LI S, et al. A semi-supervised convolutional neural network for diagnosis of pancreatic ductal adenocarcinoma based on EUS-FNA cytological images [J]. BMC Cancer, 2025, 25(1): 495. SHARIF H, ARSALAN S, REHMAN N, et al. Comparative effects of empagliflozin and metformin on metabolic dysfunction in polycystic ovary syndrome, a double blinded randomized control feasibility trial [J]. BMC Womens Health, 2025, 25(1): 565. VARA J, MANZOUR N, CHACóN E, et al. Ovarian Adnexal Reporting Data System (O-RADS) for Classifying Adnexal Masses: A Systematic Review and Meta-Analysis [J]. Cancers (Basel), 2022, 14(13). SOLIS CANO D G, CERVANTES FLORES H A, DE LOS SANTOS FARRERA O, et al. Sensitivity and Specificity of Ultrasonography Using Ovarian-Adnexal Reporting and Data System Classification Versus Pathology Findings for Ovarian Cancer [J]. Cureus, 2021, 13(9): e17646. SUNDARI M S, SAILAJA N V, SWAPNA D, et al. Transfer learning-enhanced CNN model for integrative ultrasound and biomarker-based diagnosis of polycystic ovarian disease [J]. Sci Rep, 2025, 15(1): 34519. LI H, LI G, GAO Y, et al. Contrast-enhanced ultrasound and Ovarian-Adnexal Reporting and Data System ultrasound classification for risk assessment of ovarian and adnexal lesions: a systematic review and meta-analysis [J]. Quant Imaging Med Surg, 2026, 16(1): 44. PANJWANI B, YADAV J, MOHAN V, et al. Optimized Machine Learning for the Early Detection of Polycystic Ovary Syndrome in Women [J]. Sensors (Basel), 2025, 25(4). KATLARIWALA P, WILSON M P, PI Y, et al. Reliability of ultrasound ovarian-adnexal reporting and data system amongst less experienced readers before and after training [J]. World J Radiol, 2022, 14(9): 319-28. WANG Y, MA X, LUO J, et al. Expression of Serum PSA, Nesfatin-1, and AMH in Patients with Polycystic Ovary Syndrome [J]. Cell Mol Biol (Noisy-le-grand), 2022, 67(5): 57-63. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 15 Apr, 2026 Read the published version in Journal of Ovarian Research → Version 1 posted Editorial decision: Revision requested 18 Feb, 2026 Reviews received at journal 17 Feb, 2026 Reviews received at journal 12 Feb, 2026 Reviewers agreed at journal 11 Feb, 2026 Reviewers agreed at journal 11 Feb, 2026 Reviewers invited by journal 11 Feb, 2026 Editor assigned by journal 09 Feb, 2026 Submission checks completed at journal 09 Feb, 2026 First submitted to journal 01 Feb, 2026 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-8756530","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":590272316,"identity":"e0146e8c-d839-405f-ba43-ae0371c92235","order_by":0,"name":"维梅 李","email":"","orcid":"","institution":"Shanghai Sixth People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"维梅","middleName":"","lastName":"李","suffix":""},{"id":590272317,"identity":"9ff371b5-3c7a-4b9c-894e-b5716f29c929","order_by":1,"name":"玉华 夏","email":"","orcid":"","institution":"Shuyuan Community Health Service Center","correspondingAuthor":false,"prefix":"","firstName":"玉华","middleName":"","lastName":"夏","suffix":""},{"id":590272318,"identity":"a3c375d8-083b-477b-93cd-a87a40d15e80","order_by":2,"name":"咏曌 李","email":"","orcid":"","institution":"Mudanjiang Medical University","correspondingAuthor":false,"prefix":"","firstName":"咏曌","middleName":"","lastName":"李","suffix":""},{"id":590272320,"identity":"4d29e6a8-0d6f-49cd-a364-263d8734ce56","order_by":3,"name":"星 伍","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA1klEQVRIiWNgGAWjYDACZijNx8B84MCHH6RoYWNgSzw4s4cU29gYeIwPc7ARodLgOO/hFww1d/LY2Hs+HGbgYZDnFzuAX4tkM1+aBcOxZ8VsPGc3HC6wYDCcOTsBvxZ+Zh4zAwa2w4ltErkbDs/gYUgwuE1ACxtYyz+gFvk3Dw7zsBGhBWiL8QPGNpAtPAzEaZFs5jFjYOwD+SXNABjIEoT9YnD+jPEHhm938vjZDz/+8OGHjTy/NAEtIO9I/2E4AFMmQVA5CDB/YEBoGQWjYBSMglGACQD9fEGPWXYF2wAAAABJRU5ErkJggg==","orcid":"","institution":"Shanghai Sixth People's Hospital","correspondingAuthor":true,"prefix":"","firstName":"星","middleName":"","lastName":"伍","suffix":""},{"id":590272321,"identity":"3c643a46-67b9-4e30-883f-30afe702668a","order_by":4,"name":"琳 师","email":"","orcid":"","institution":"Shanghai Sixth People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"琳","middleName":"","lastName":"师","suffix":""}],"badges":[],"createdAt":"2026-02-01 13:38:22","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8756530/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8756530/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s13048-026-02090-1","type":"published","date":"2026-04-15T15:59:17+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":102851890,"identity":"b2f1451f-e5aa-48c4-8025-3c2f44417cb6","added_by":"auto","created_at":"2026-02-17 14:33:34","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":77352,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eROC Curves of the GoogLeNet Model on the Training and Validation Sets\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-8756530/v1/a585ce601ae669f1d865d73c.png"},{"id":102851892,"identity":"ec1dfb6b-5381-421e-8c83-27fcf2d386de","added_by":"auto","created_at":"2026-02-17 14:33:34","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":80172,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eROC Curves of the ViT Model on the Training and Validation Sets\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-8756530/v1/a519dbe28a266a49f525fac1.png"},{"id":102851889,"identity":"b20ccd12-ca31-4ccd-8cde-e3c92be87165","added_by":"auto","created_at":"2026-02-17 14:33:34","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":73177,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eROC Curves of the DenseNet121 Model on the Training and Validation Sets\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-8756530/v1/dfa0cc0deeef5457ff165907.png"},{"id":102851893,"identity":"13d3a9a0-7fa0-4b1d-bc46-46182cbe80ee","added_by":"auto","created_at":"2026-02-17 14:33:34","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":77527,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eROC Curves of the ResNet34 Model on the Training and Validation Sets\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-8756530/v1/693b1dd951706db040d900ea.png"},{"id":107351554,"identity":"f25c81aa-7236-488c-8bb9-0f6f904fdc4a","added_by":"auto","created_at":"2026-04-20 16:11:24","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":639867,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8756530/v1/274be925-806a-4bb2-8e88-c98ef4579cd2.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Research on an Optimized Deep Learning-Based Classification Model for Ovarian Cyst Ultrasound Images","fulltext":[{"header":"Background","content":"\u003cp\u003eIn recent years, deep learning technologies have demonstrated strong potential in intelligent medical imaging diagnostics\u003csup\u003e[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]\u003c/sup\u003e, offering innovative solutions for automated and precise classification of ovarian cysts\u003csup\u003e[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]\u003c/sup\u003e.Ovarian cysts represent a prevalent gynecological condition in women\u003csup\u003e[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]\u003c/sup\u003e, and accurate differentiation between benign and malignant types is critical for developing individualized treatment strategies and improving patient prognosis\u003csup\u003e[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]\u003c/sup\u003e. However, clinical diagnosis remains challenging, as some patients are preoperatively suspected of malignancy but postoperatively confirmed as benign, leading to overtreatment and psychological burden. Ultrasound examination\u003csup\u003e[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]\u003c/sup\u003e ,due to its non-invasive, convenient, and cost-effective nature, serves as the preferred imaging modality for initial screening and follow-up of ovarian cysts. Nevertheless, diagnostic outcomes are heavily influenced by the operator's experience and subjective judgment, resulting in substantial inter-observer variability.\u003c/p\u003e \u003cp\u003eTo enhance diagnostic consistency and accuracy, the Ovarian-Adnexal Reporting and Data System (O-RADS) has been internationally proposed\u003csup\u003e[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]\u003c/sup\u003e, standardizing ultrasound features into risk stratification categories to provide clinical references\u003csup\u003e[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]\u003c/sup\u003e. However, O-RADS exhibits uncertainty in diagnosing category 4 ovarian masses\u003csup\u003e[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]\u003c/sup\u003e, with approximately half of those classified as malignant later confirmed as benign postoperatively, indicating a potential issue of overdiagnosis. Additionally, while serum tumor markers can serve as supplementary indicators\u003csup\u003e[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]\u003c/sup\u003e, their sensitivity and specificity are limited when used in isolation.\u003c/p\u003e \u003cp\u003eIn recent years, deep learning has achieved remarkable advancements in medical image analysis\u003csup\u003e[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]\u003c/sup\u003e, enabling automated learning and extraction of high-dimensional image features while minimizing biases from manual feature selection\u003csup\u003e[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]\u003c/sup\u003e. This offers novel approaches for intelligent diagnosis of complex images. Existing studies have demonstrated that convolutional neural networks (CNNs) and their variants achieve excellent classification performance in lesion identification from ultrasound images of the breast, thyroid, and prostate \u003csup\u003e[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]\u003c/sup\u003e.However, research on deep learning for differentiating benign and malignant ovarian cysts remains relatively limited\u003csup\u003e[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]\u003c/sup\u003e, with a lack of direct comparative studies against O-RADS and clinical expert assessments.\u003c/p\u003e \u003cp\u003eOvarian cysts are a common pelvic disorder in women\u003csup\u003e[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]\u003c/sup\u003e.This study, based on ultrasound images from pathologically confirmed ovarian cysts\u003csup\u003e[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]\u003c/sup\u003e, constructed multiple end-to-end deep learning models\u003csup\u003e[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]\u003c/sup\u003e, including DenseNet121, ResNet34, GoogLeNet \u003csup\u003e[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]\u003c/sup\u003e, and Vision Transformer (ViT). Results indicated that the DenseNet121 model outperformed the others, achieving area under the curve (AUC) values of 0.914 and 0.913 on the training and validation sets, respectively. It also demonstrated superior accuracy, sensitivity, and specificity on the validation set. Grad-CAM visualizations further validated the model's reliability in extracting features from lesion regions, showing high consistency with clinical lesion distributions. Furthermore, the diagnostic efficacy of this model was compared with subjective assessments by senior ultrasound physicians and O-RADS grading results\u003csup\u003e[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]\u003c/sup\u003e, aiming to explore the application value of deep learning in differentiating benign and malignant ovarian cysts and to provide clinicians with more objective and reliable decision-support tools\u003csup\u003e[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eCompared to subjective evaluations by senior ultrasound physicians (accuracy: 76.5%) and the O-RADS classification system\u003csup\u003e[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]\u003c/sup\u003e (accuracy: 87.8%), the DenseNet121 model exhibited significant advantages in overall diagnostic efficacy (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05).\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Study Population\u003c/h2\u003e \u003cp\u003eThis retrospective study included 327 patients with ovarian cysts who underwent surgery and received pathological confirmation at our institution from January 2023 to January 2025. All patients had complete preoperative two-dimensional ultrasound images and clinical data, comprising 249 benign cases and 78 malignant cases. The dataset was randomly divided into a training set (196 cases) and a validation set (131 cases) in a 6:4 ratio. Patients who were pregnant, lactating, or had a history of severe cardiac disease were excluded.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Data Collection and Preprocessing\u003c/h2\u003e \u003cp\u003eAll patients underwent pelvic ultrasound examinations using the GE Voluson E8 color Doppler ultrasound system, equipped with a transvaginal probe (RIC5-9-D) and an abdominal probe (RAB2-5-D). Images were acquired in transverse, sagittal, and magnified lesion views, documenting details such as lesion number, size, morphology, borders, wall thickness, papillary projections, septations, blood flow signals, resistance index, and ascites. Images were stored in DICOM format, anonymized by removing patient information, and imported into the deep learning framework\u003csup\u003e[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]\u003c/sup\u003e. Preprocessing involved resizing to 224\u0026times;224 pixels, pixel value normalization, and data augmentation techniques including random rotation, translation, and flipping to enhance model generalization.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Model Construction and Training\u003c/h2\u003e \u003cp\u003eFour end-to-end deep learning models\u0026mdash;DenseNet121, ResNet34, GoogLeNet\u003csup\u003e[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]\u003c/sup\u003e, and Vision Transformer (ViT)\u0026mdash;were selected as candidate networks. All models were initialized with ImageNet pre-trained weights and fine-tuned via transfer learning on the training set. Binary cross-entropy was used as the loss function, with Adam as the optimizer, an initial learning rate of 1e-4, a batch size of 32, and training for 100 epochs. AUC was monitored on the validation set, with an early stopping strategy implemented to prevent overfitting.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4 Performance Evaluation Metrics\u003c/h2\u003e \u003cp\u003ePostoperative pathological diagnosis served as the gold standard. Receiver operating characteristic (ROC) curves were plotted, and the area under the curve (AUC) with 95% confidence intervals was calculated. Additional metrics, including accuracy, sensitivity, specificity, precision, recall, and F1 score, were computed to assess classification performance. Grad-CAM was applied to the optimal model for visualization analysis, examining the consistency between the model's attention regions and clinical lesion distributions.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.5 Statistical Analysis\u003c/h2\u003e \u003cp\u003eStatistical analyses were performed using SPSS 26.0 and Python. Continuous data are presented as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;standard deviation (x\u0026thinsp;\u0026plusmn;\u0026thinsp;s), and categorical data as percentages (%). AUC comparisons between models were conducted using the DeLong test, while classification metrics such as accuracy were compared via chi-square or McNemar tests\u003csup\u003e[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]\u003c/sup\u003e. A p-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 was considered statistically significant.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Patient Baseline Characteristics\u003c/h2\u003e \u003cp\u003eThis study included a total of 327 patients with pathologically confirmed ovarian cysts, comprising 249 benign cases (76.15%) and 78 malignant cases (23.85%). The cohort was randomly divided into a training set of 196 cases (149 benign, 47 malignant) and a validation set of 131 cases (100 benign, 31 malignant) in a 6:4 ratio. No statistically significant differences were observed between the two groups in terms of age, lesion size, or pathological type distribution (P\u0026thinsp;\u0026gt;\u0026thinsp;0.05), ensuring comparability.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e3.2 Comparison of Classification Performance Among Models\u003c/h2\u003e \u003cp\u003eThe classification performance of the four deep learning models is summarized in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. The DenseNet121 model demonstrated the highest AUC and overall classification performance on both the training and validation sets, with an AUC of 0.914 (95% CI: 0.8733\u0026ndash;0.9548) on the training set and 0.913 (95% CI: 0.8584\u0026ndash;0.9677) on the validation set. Its validation set accuracy, sensitivity, and specificity reached 87.8%, 82.6%, and 88.9%, respectively, which were significantly superior to those of ResNet34, GoogLeNet, and ViT (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eClassification Performance of the Four Deep Learning Models\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"9\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModelName\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAcc\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAUC\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e95% CI\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eSpecificity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003ePrecision\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eRecall\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eCohort\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003edensenet121\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.837\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.914\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.8733\u0026ndash;0.9548\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.907\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.81\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.645\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.907\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003eTrain\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003edensenet121\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.878\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.913\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.8584\u0026ndash;0.9677\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.826\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.889\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.613\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.826\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003eTest\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eresnet34\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.837\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.903\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.8605\u0026ndash;0.9450\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.833\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.838\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.662\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.833\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003eTrain\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eresnet34\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.649\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.844\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.7537\u0026ndash;0.9351\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.957\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.583\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.328\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.957\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003eTest\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003egooglenet\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.393\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.491\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.4005\u0026ndash;0.5824\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.833\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.225\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.29\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.833\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003eTrain\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003egooglenet\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.435\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.524\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.3949\u0026ndash;0.6538\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.739\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.37\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.739\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003eTest\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eViT\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.276\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.407\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.3185\u0026ndash;0.4954\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.276\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003eTrain\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eViT\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.817\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.628\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.4786\u0026ndash;0.7774\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.435\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.898\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.476\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.435\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003eTest\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=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e3.3 ROC Curve Analysis\u003c/h2\u003e \u003cp\u003eFigures \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e illustrate the ROC curves for the GoogLeNet and ViT models on the training and validation sets, respectively, where their AUC values approached random classification levels, indicating poor discriminative ability for benign versus malignant ovarian cysts. In contrast, the ROC curves for DenseNet121 (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e) and ResNet34 (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e) shifted markedly toward the upper-left corner, with DenseNet121 exhibiting the largest area under the curve and the optimal diagnostic performance.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e3.4 Grad-CAM Visualization Results\u003c/h2\u003e \u003cp\u003eSeveral representative cases from the validation set were selected for Grad-CAM visualization of the DenseNet121 model's attention regions. The results revealed that the model primarily focused on lesion wall thickness, papillary projections, and septations, which aligned closely with key pathological signs. This suggests strong consistency between the model's decision-making basis and clinical observations (see Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Figure\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. \u003cb\u003eGrad-CAM Visualization Results for the DenseNet121 Model\u003c/b\u003e: Red areas indicate the model's attention regions, showing high overlap with lesion distributions.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e3.5 Comparative Analysis with O-RADS and Expert Assessments\u003c/h2\u003e \u003cp\u003eThe DenseNet121 model's results were compared with subjective assessments by senior ultrasound physicians and the O-RADS grading system\u003csup\u003e[\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]\u003c/sup\u003e. The model's accuracy on the validation set (87.8%) was significantly higher than that of manual assessments (76.5%) and slightly superior to O-RADS grading (87.8%), with statistically significant differences (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05). The model also demonstrated greater balance in sensitivity and specificity, reducing clinical risks associated with overdiagnosis or missed diagnoses.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003e3.6 Summary\u003c/h2\u003e \u003cp\u003eOverall, the DenseNet121 model exhibited the highest diagnostic performance and generalization capability in classifying benign and malignant ovarian cysts, performing well on both the training and validation sets. It effectively focused on key imaging features relevant to pathological diagnosis, outperforming traditional O-RADS classification and manual assessments\u003csup\u003e[\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]\u003c/sup\u003e. This model holds substantial value for clinical promotion and as an auxiliary decision-making tool.\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study, based on ultrasound images from 327 pathologically confirmed ovarian cyst cases, systematically compared the performance of four deep learning models\u0026mdash;DenseNet121, ResNet34, GoogLeNet\u003csup\u003e[\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]\u003c/sup\u003e, and Vision Transformer (ViT)\u0026mdash;in classifying benign and malignant ovarian cysts. These models were also benchmarked against O-RADS grading and subjective assessments by senior ultrasound physicians\u003csup\u003e[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]\u003c/sup\u003e. The results demonstrated that DenseNet121 outperformed the other models in key metrics such as AUC, sensitivity, specificity, and accuracy. Its diagnostic performance not only surpassed manual evaluations but also exceeded O-RADS grading, indicating that deep learning-based automated diagnostic models offer significant advantages in differentiating benign and malignant ovarian cysts\u003csup\u003e[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eFirst, the superior performance of DenseNet121 may be closely linked to its architectural features. DenseNet facilitates feature reuse and efficient gradient propagation through dense cross-layer connections, enabling the capture of fine-grained features in ultrasound images. This is particularly well-suited for handling medical images with complex textures and subtle boundaries. In contrast, while ResNet34 incorporates residual connections, its shallower network depth limits feature representation capabilities, resulting in reduced accuracy on the validation set. The poorer performance of GoogLeNet and ViT in this study could be attributed to the relatively limited ultrasound image dataset, ViT's strong dependence on large-scale data, and GoogLeNet's outdated network design. These findings underscore the importance of selecting architectures optimized for small-sample learning in medical imaging scenarios.\u003c/p\u003e \u003cp\u003eCompared to O-RADS grading, the DenseNet121 model achieved a better balance between sensitivity and specificity. Although O-RADS provides a standardized grading system that effectively reduces diagnostic variability among operators\u003csup\u003e[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]\u003c/sup\u003e, it exhibits ambiguity in risk assessment for category 4 lesions, often leading to overtreatment in clinical practice. The results of this study show that deep learning models, by automatically extracting imaging features, can more precisely differentiate between 4a and 4b subcategory lesions, thereby minimizing misdiagnoses. Relative to subjective assessments by experienced ultrasound physicians, the model not only delivered higher accuracy but also enabled rapid, stable, and reproducible diagnoses, mitigating the impact of inter-individual experience variations.\u003csup\u003e[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]\u003c/sup\u003e\u003c/p\u003e \u003cp\u003e Furthermore, the Grad-CAM visualization results provided additional evidence of the DenseNet121 model's interpretability. The model's attention regions were primarily focused on lesion wall thickness, papillary projections, and areas with abundant blood flow\u0026mdash;key indicators in the clinical diagnosis of malignant ovarian tumors. These findings suggest that deep learning models can go beyond \"black-box\" classifications to offer reasonable visual explanations, enhancing clinicians' trust in the model and facilitating its integration into real-world clinical workflows.\u003c/p\u003e \u003cp\u003eAlthough the results of this study hold certain clinical reference value, several limitations remain. First, as a single-center retrospective study with a relatively limited sample size\u0026mdash;particularly with a low proportion of malignant cases\u0026mdash;it may affect the model's generalizability. Second, the study relied solely on two-dimensional static ultrasound images as input, excluding three-dimensional ultrasound or dynamic video data, which could overlook spatial or temporal features. Additionally, the comparisons were limited to experts from a single institution and O-RADS grading results; future studies could incorporate multi-center and multi-physician involvement to strengthen external validity.\u003c/p\u003e \u003cp\u003eFuture research could explore the following directions: (1) expanding sample sizes and incorporating multi-center datasets to further enhance model robustness and generalization; (2) investigating multimodal deep learning models that integrate diverse data sources, such as ultrasound video sequences, serum tumor markers\u003csup\u003e[\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]\u003c/sup\u003e, and patient clinical information, to achieve superior diagnostic efficacy; (3) examining the model's performance across different ovarian tumor subtypes and stages, while assessing its potential in early ovarian cancer screening; and (4) combining explainable artificial intelligence techniques to improve model transparency and clinical acceptability, thereby promoting its deployment in clinical decision-support systems.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eIn summary, the DenseNet121 deep learning model developed in this study exhibited excellent performance in classifying benign and malignant ovarian cysts, significantly improving diagnostic accuracy and consistency while reducing subjective errors. It holds strong potential as a clinical auxiliary diagnostic tool. With further advancements in multi-center validation and multimodal integration, deep learning is poised to become a vital complement to ovarian tumor risk stratification and treatment decision-making, supporting the evolution of precision medicine.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe are grateful to Lichang Zhong and Kaihua Fan for their valuable contributions to subject recruitment and eligibility assessment, patient liaison and longitudinal follow-up, and the management of study documentation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eW.M.L. and Y.H.X. designed the project.\u003c/p\u003e\n\u003cp\u003eW.M.L. ,Y.H.X., Y.Z.L. ,X.W. \u0026nbsp; and L.S. performed experiments.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eW.M.L wrote the manuscript.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eW.M.L., Y.H.X. ,Y.Z.L., X.W., and L.S. collected the samples.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eW.M.L., Y.H.X. , Y.Z.L., X.W., and L.S. analyzed the data.\u003c/p\u003e\n\u003cp\u003eW.M.L., Y.H.X. , Y.Z.L. reviewed and edited the manuscript.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eW.M.L.acquired funding\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by the Shanghai Key Clinical Research Center Fund[grant number 2023ZZ02006]\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was approved by the Institutional Review Board (IRB) of the Sixth People\u0026apos;s Hospital of Shanghai, China (Approval No.:2025-KY-105(K) Date of Review:2025-03-27) . The study was conducted in accordance with the Good Clinical Practice (GCP) guidelines and the Ethical Review Guidelines for Drug Clinical Trials issued by the National Medical Products Administration (NMPA) of the People\u0026rsquo;s Republic of China, as well as the principles of ICH-GCP. It also adheres to the requirements of the Declaration of Helsinki and complies with relevant laws and regulations in China.Written informed consent was obtained from all individual participants included in the study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDisclosure of Interests\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no conflict of interests.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor details\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003csup\u003e1\u003c/sup\u003eDepartment of Ultrasound in Medicine, Shanghai Sixth People\u0026apos;s Hospital Affiliated to Shanghai 6 Jiao Tong University School of Medicine, Shanghai, 201306, China\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eBARRERA F J, BROWN E D L, ROJO A, et al. Application of machine learning and artificial intelligence in the diagnosis and classification of polycystic ovarian syndrome: a systematic review [J]. Front Endocrinol (Lausanne), 2023, 14: 1106625.\u003c/li\u003e\n \u003cli\u003eANDREOTTI R F, TIMMERMAN D, STRACHOWSKI L M, et al. O-RADS US Risk Stratification and Management System: A Consensus Guideline from the ACR Ovarian-Adnexal Reporting and Data System Committee [J]. Radiology, 2020, 294(1): 168-85.\u003c/li\u003e\n \u003cli\u003eXIE W T, WANG Y Q, XIANG Z S, et al. Efficacy of IOTA simple rules, O-RADS, and CA125 to distinguish benign and malignant adnexal masses [J]. J Ovarian Res, 2022, 15(1): 15.\u003c/li\u003e\n \u003cli\u003eSHANMUGAVADIVEL K, M S M, T R M, et al. Optimized polycystic ovarian disease prognosis and classification using AI based computational approaches on multi-modality data [J]. BMC Med Inform Decis Mak, 2024, 24(1): 281.\u003c/li\u003e\n \u003cli\u003eSUHA S A, ISLAM M N. An extended machine learning technique for polycystic ovary syndrome detection using ovary ultrasound image [J]. Sci Rep, 2022, 12(1): 17123.\u003c/li\u003e\n \u003cli\u003eSTRACHOWSKI L M, JHA P, PHILLIPS C H, et al. O-RADS US v2022: An Update from the American College of Radiology\u0026apos;s Ovarian-Adnexal Reporting and Data System US Committee [J]. Radiology, 2023, 308(3): e230685.\u003c/li\u003e\n \u003cli\u003ePHILLIPS C H, PATEL-LIPPMANN K, HUANG J, et al. Ovarian-Adnexal Reporting and Data System Ultrasound v2022: From Origin to Everyday Use [J]. Radiol Clin North Am, 2025, 63(1): 29-44.\u003c/li\u003e\n \u003cli\u003eSHEN L, SADOWSKI E A, GUPTA A, et al. The Ovarian-Adnexal Reporting and Data System (O-RADS) US Score Effect on Surgical Resection Rate [J]. Radiology, 2024, 313(1): e240044.\u003c/li\u003e\n \u003cli\u003ePANG C, MA Y, SHI W, et al. Prognostic significance of serum tumor markers in various pathologic subtypes of gastric cancer [J]. J Gastrointest Surg, 2024, 28(5): 694-702.\u003c/li\u003e\n \u003cli\u003eBUDDENKOTTE T, RUNDO L, WOITEK R, et al. Deep learning-based segmentation of multisite disease in ovarian cancer [J]. Eur Radiol Exp, 2023, 7(1): 77.\u003c/li\u003e\n \u003cli\u003eREKA S, PRABA T S, PRASANNA M, et al. Automated high precision PCOS detection through a segment anything model on super resolution ultrasound ovary images [J]. Sci Rep, 2025, 15(1): 16832.\u003c/li\u003e\n \u003cli\u003eAGYEKUM E A, WANG Y G, ISSAKA E, et al. Predicting the efficacy of microwave ablation of benign thyroid nodules from ultrasound images using deep convolutional neural networks [J]. BMC Med Inform Decis Mak, 2025, 25(1): 161.\u003c/li\u003e\n \u003cli\u003eWANG R, CAI Y, LEE I K, et al. Evaluation of a convolutional neural network for ovarian tumor differentiation based on magnetic resonance imaging [J]. Eur Radiol, 2021, 31(7): 4960-71.\u003c/li\u003e\n \u003cli\u003eDOKRAS A. Polycystic ovary syndrome in 2025-insights and innovations [J]. Fertil Steril, 2025, 124(5 Pt 2): 907-9.\u003c/li\u003e\n \u003cli\u003eRUAN Z, YU Z, QIN Q, et al. Diagnostic value of ultrasound elastography in polycystic ovary syndrome: a systematic review and meta-analysis [J]. Gynecol Endocrinol, 2024, 40(1): 2352139.\u003c/li\u003e\n \u003cli\u003eDUTTA P, SATHI K A, HOSSAIN M A, et al. Conv-ViT: A Convolution and Vision Transformer-Based Hybrid Feature Extraction Method for Retinal Disease Detection [J]. J Imaging, 2023, 9(7).\u003c/li\u003e\n \u003cli\u003eSHA M. Segmentation of ovarian cyst in ultrasound images using AdaResU-net with optimization algorithm and deep learning model [J]. Sci Rep, 2024, 14(1): 18868.\u003c/li\u003e\n \u003cli\u003eYI Y Y, LI C, ZHU W J, et al. Diagnostic performance of contrast-enhanced ultrasound (CEUS) combined with Ovarian-Adnexal Reporting and Data System (O-RADS) ultrasound risk stratification for adnexal masses: a systematic review and meta-analysis [J]. Clin Radiol, 2024, 79(9): e1167-e75.\u003c/li\u003e\n \u003cli\u003eJEYASHANKER P, GEORGEWILLIAM SUNDARAM A G V, SADAGOPAN P, et al. Advanced holographic convolutional dense networks and Tangent runner optimization for enhanced polycystic ovarian disease classification [J]. Sci Rep, 2025, 15(1): 15719.\u003c/li\u003e\n \u003cli\u003eGUO Y, STRACHOWSKI L M. Beyond the AJR: O-RADS Ultrasound Risk Stratification Performs Within Expected Range With Excellent Diagnostic Performance in Nonselected Female Patients in the United States [J]. AJR Am J Roentgenol, 2023, 220(3): 450.\u003c/li\u003e\n \u003cli\u003eZHAO B, WEN L, HUANG Y, et al. A Deep Learning-Based Automatic Recognition Model for Polycystic Ovary Ultrasound Images [J]. Balkan Med J, 2025, 42(5): 419-28.\u003c/li\u003e\n \u003cli\u003eFANG D, HUANG Y, LI S, et al. A semi-supervised convolutional neural network for diagnosis of pancreatic ductal adenocarcinoma based on EUS-FNA cytological images [J]. BMC Cancer, 2025, 25(1): 495.\u003c/li\u003e\n \u003cli\u003eSHARIF H, ARSALAN S, REHMAN N, et al. Comparative effects of empagliflozin and metformin on metabolic dysfunction in polycystic ovary syndrome, a double blinded randomized control feasibility trial [J]. BMC Womens Health, 2025, 25(1): 565.\u003c/li\u003e\n \u003cli\u003eVARA J, MANZOUR N, CHAC\u0026oacute;N E, et al. Ovarian Adnexal Reporting Data System (O-RADS) for Classifying Adnexal Masses: A Systematic Review and Meta-Analysis [J]. Cancers (Basel), 2022, 14(13).\u003c/li\u003e\n \u003cli\u003eSOLIS CANO D G, CERVANTES FLORES H A, DE LOS SANTOS FARRERA O, et al. Sensitivity and Specificity of Ultrasonography Using Ovarian-Adnexal Reporting and Data System Classification Versus Pathology Findings for Ovarian Cancer [J]. Cureus, 2021, 13(9): e17646.\u003c/li\u003e\n \u003cli\u003eSUNDARI M S, SAILAJA N V, SWAPNA D, et al. Transfer learning-enhanced CNN model for integrative ultrasound and biomarker-based diagnosis of polycystic ovarian disease [J]. Sci Rep, 2025, 15(1): 34519.\u003c/li\u003e\n \u003cli\u003eLI H, LI G, GAO Y, et al. Contrast-enhanced ultrasound and Ovarian-Adnexal Reporting and Data System ultrasound classification for risk assessment of ovarian and adnexal lesions: a systematic review and meta-analysis [J]. Quant Imaging Med Surg, 2026, 16(1): 44.\u003c/li\u003e\n \u003cli\u003ePANJWANI B, YADAV J, MOHAN V, et al. Optimized Machine Learning for the Early Detection of Polycystic Ovary Syndrome in Women [J]. Sensors (Basel), 2025, 25(4).\u003c/li\u003e\n \u003cli\u003eKATLARIWALA P, WILSON M P, PI Y, et al. Reliability of ultrasound ovarian-adnexal reporting and data system amongst less experienced readers before and after training [J]. World J Radiol, 2022, 14(9): 319-28.\u003c/li\u003e\n \u003cli\u003eWANG Y, MA X, LUO J, et al. Expression of Serum PSA, Nesfatin-1, and AMH in Patients with Polycystic Ovary Syndrome [J]. Cell Mol Biol (Noisy-le-grand), 2022, 67(5): 57-63.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"journal-of-ovarian-research","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"jovr","sideBox":"Learn more about [Journal of Ovarian Research](http://ovarianresearch.biomedcentral.com)","snPcode":"13048","submissionUrl":"https://submission.nature.com/new-submission/13048/3","title":"Journal of Ovarian Research","twitterHandle":"@BioMedCentral","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Ovarian cyst, Deep learning, O-RADS, Feature extraction","lastPublishedDoi":"10.21203/rs.3.rs-8756530/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8756530/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eOvarian cysts are a common pelvic disorder in women, and accurate differentiation between benign and malignant types is essential for guiding treatment decisions and prognostic evaluations. However, traditional ultrasound examinations heavily depend on the operator's experience, introducing subjectivity and diagnostic inconsistencies. In recent years, deep learning technologies have demonstrated strong potential in intelligent medical imaging diagnostics, offering innovative solutions for automated and precise classification of ovarian cysts.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eCompared to subjective evaluations by senior ultrasound physicians (accuracy: 76.5%) and the O-RADS classification system (accuracy: 87.8%), the DenseNet121 model exhibited significant advantages in overall diagnostic efficacy (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05).\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eDeep learning models based on ultrasound images can effectively address noise and feature complexity in such imaging, enabling high-precision classification of benign and malignant ovarian cysts. These models hold strong potential for clinical adoption, providing physicians with objective and reliable decision-making support.\u003c/p\u003e","manuscriptTitle":"Research on an Optimized Deep Learning-Based Classification Model for Ovarian Cyst Ultrasound Images","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-17 14:33:29","doi":"10.21203/rs.3.rs-8756530/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-02-18T06:40:04+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-02-17T06:33:06+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-02-12T12:33:06+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"84189079328254564290088147879078795312","date":"2026-02-12T00:52:21+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"102920125384277321587986799012242544453","date":"2026-02-11T23:47:07+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-02-11T23:31:06+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-02-09T12:09:14+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-02-09T12:03:26+00:00","index":"","fulltext":""},{"type":"submitted","content":"Journal of Ovarian Research","date":"2026-02-01T13:31:25+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"journal-of-ovarian-research","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"jovr","sideBox":"Learn more about [Journal of Ovarian Research](http://ovarianresearch.biomedcentral.com)","snPcode":"13048","submissionUrl":"https://submission.nature.com/new-submission/13048/3","title":"Journal of Ovarian Research","twitterHandle":"@BioMedCentral","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"6945936d-88ed-45ee-96e9-f462a3ce8b9c","owner":[],"postedDate":"February 17th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2026-04-20T16:07:51+00:00","versionOfRecord":{"articleIdentity":"rs-8756530","link":"https://doi.org/10.1186/s13048-026-02090-1","journal":{"identity":"journal-of-ovarian-research","isVorOnly":false,"title":"Journal of Ovarian Research"},"publishedOn":"2026-04-15 15:59:17","publishedOnDateReadable":"April 15th, 2026"},"versionCreatedAt":"2026-02-17 14:33:29","video":"","vorDoi":"10.1186/s13048-026-02090-1","vorDoiUrl":"https://doi.org/10.1186/s13048-026-02090-1","workflowStages":[]},"version":"v1","identity":"rs-8756530","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8756530","identity":"rs-8756530","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}