RESUMO
AIM: To evaluate the application of an intelligent diagnostic model for pterygium. METHODS: For intelligent diagnosis of pterygium, the attention mechanisms-SENet, ECANet, CBAM, and Self-Attention-were fused with the lightweight MobileNetV2 model structure to construct a tri-classification model. The study used 1220 images of three types of anterior ocular segments of the pterygium provided by the Eye Hospital of Nanjing Medical University. Conventional classification models-VGG16, ResNet50, MobileNetV2, and EfficientNetB7-were trained on the same dataset for comparison. To evaluate model performance in terms of accuracy, Kappa value, test time, sensitivity, specificity, the area under curve (AUC), and visual heat map, 470 test images of the anterior segment of the pterygium were used. RESULTS: The accuracy of the MobileNetV2+Self-Attention model with 281 MB in model size was 92.77%, and the Kappa value of the model was 88.92%. The testing time using the model was 9ms/image in the server and 138ms/image in the local computer. The sensitivity, specificity, and AUC for the diagnosis of pterygium using normal anterior segment images were 99.47%, 100%, and 100%, respectively; using anterior segment images in the observation period were 88.30%, 95.32%, and 96.70%, respectively; and using the anterior segment images in the surgery period were 88.18%, 94.44%, and 97.30%, respectively. CONCLUSION: The developed model is lightweight and can be used not only for detection but also for assessing the severity of pterygium.
RESUMO
RATIONALE AND OBJECTIVES: To evaluate the performance of three-dimensional (3D) amide proton transfer-weighted (APTw) MRI in the differentiation between benign and malignant ovarian masses based on single-slice and all-slice analysis of cystic regions. MATERIALS AND METHODS: Patients were consecutively recruited and underwent conventional pelvic MRI and APTw MRI. Two radiologists independently assessed ovarian masses blinded to the histopathological results. Three APTw SI values were generated from the cystic regions of the masses: (1) APTw SI of a single representative slice (RS); (2) average (AVE) of APTw SIs of all slices of the mass; (3) area-weighted (AW) average of APTw SIs of all slices of the mass. O-RADS MRI score of each mass was reported. Independent sample t-test and receiver operating characteristic (ROC) curve analysis were performed for comparison. Inter- and intra-observer reliability were assessed by the intraclass correlation coefficient (ICC) and quadratic kappa coefficient. RESULTS: 46 ovarian masses were included for final analysis. The three APTw SI values were higher in cystic regions of malignant ovarian masses compared with benign lesions (p<0.0001). ROC curve analysis showed no significant difference in diagnostic performance among three APTw SI values and the O-RADS MRI score (AUC: RS-APTw SI, 0.930; AVE-APTw SI, 0.927; AW-APTw SI, 0.935; O-RADS score, 0.937). CONCLUSIONS: APTw MRI may be used as a noninvasive tool for the differentiation of benign and malignant ovarian masses based on the analysis of the cystic regions.