Real-time intraoperative identification of malignant ovarian tumors using deep learning of clinical gross specimen images

Background:Ovarian cancer is one of the most common gynecologic malignancies and has a high mortality. A proper operation strategy is key to improving patient prognosis, the timely intraoperative identification of malignant ovarian tumors is challenging for clinical physicians. Therefore, this study aims to develop a deep learning model that can be used intraoperatively to differentiate malignant from benign ovarian tumors based on gross specimen images.Methods: This retrospective cohort study was conducted in an academic tertiary hospital. The study collected 662 cut-in-half gross images of specimens from 243 patients as a training dataset to develop convolutional neural network (CNN)-based models for differentiating malignant and benign ovarian tumors. Different deep learning algorithms including classic CNN, VGG16, Xception, InceptionV3, MobileNet, NASNet, EfficientNetB0 and ResNet50, were used for training and validation. Another 56 patients with 129 cut-in-half images were used as the independent test dataset. A smartphone application with the integration of the optimized model was developed to perform intraoperative identification of malignant ovarian tumors.Findings: Among the eight different CNN models, the MobileNet model had a better discriminating ability in classification (accuracy: 69.77%, recall: 61.84%, precision: 82.46%, AUC: 0.7149) and the performance was more precise than that of junior physicians (accuracy: 71.83%, recall: 80.48%, precision: 74.33%).InterpretationThis study demonstrated the feasibility of differentiating between ovarian malignancies and benign tumors using clinical gross images. The real-time intraoperative assessment of ovarian tumors using the mobile application has the potential to accelerate clinical decision making before obtaining pathology reports during the operation.Funding: This study was funded in part by the National Science and Technology Council, Taiwan, Taipei Veterans General Hospital, and Szu-Yuan research foundation of internal medicine, supported by the Mt. Jade Young Scholarship Award from the Ministry of Education, Taiwan, as well as Brain Research Center, National Yang Ming Chiao Tung University, and the Ministry of Education (Aim for the Top University Plan), Taipei, Taiwan.Funding: This study was funded in part by the National Science and Technology Council, Taiwan, Taipei Veterans General Hospital, and Szu-Yuan research foundation of internal medicine, supported by the Mt. Jade Young Scholarship Award from the Ministry of Education, Taiwan, as well as Brain Research Center, National Yang Ming Chiao Tung University, and the Ministry of Education (Aim for the Top University Plan), Taipei, Taiwan.Declaration of Interest: All authors declare no conflicts of interest.Ethical Approval: This study was approved by the Institutional Review Board (IRB) of Taipei Veterans General Hospital (Registration number: 2021-05- 005AC#1, 2022-06-008BC).