Research on Diagnosis Model of Endometrial Lesions by Hysteroscopy Based on Deep Learning Algorithm Combined with Grad-CAM
- VernacularTitle:基于深度学习算法联合Grad-CAM的宫腔镜子宫内膜病变诊断模型研究
- Author:
Mingliang CAO
1
;
Mi YIN
;
Qingbin WANG
;
Hanfeng ZHU
;
Xing LI
;
Jun ZHANG
;
Lin MAO
;
Xuefeng MU
;
Min CAO
;
Yutao MA
;
Jian WANG
;
Yan ZHANG
Author Information
1. 武汉大学人民医院妇产科,湖北武汉 430060
- Keywords:
Hysteroscopy;
Endometrial cancer;
Convolutional neural network;
Gradient class activation map;
Deep learning
- From:
Journal of Practical Obstetrics and Gynecology
2024;40(5):409-413
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To explore the effectiveness of a hysteroscopic endometrial lesion diagnosis model de-veloped based on deep learning(DL)algorithm combined with gradient-weighted class activation mapping(Grad-CAM)visualization technology.Methods:303 hysteroscopy videos(4781 images)of 291 patients who un-derwent hysteroscopy examination in the Department of Gynecology,Renmin Hospital of Wuhan University from June 1,2021 to December 31,2022 were selected.The dataset was divided into a training set(3703 images)and a test set(1078 images)by weight sampling method.After the training set was used for model learning and train-ing,two model architectures,residual neural network(ResNet18)and efficient neural network(EfficientNet-B0),were selected to verify the model in the test set by five-class and two-class classification tasks,respectively.Tak-ing histopathology as the gold standard,the diagnostic efficacy was evaluated to select the optimal model,and the Grad-CAM layer was embedded in the optimal model to output hysteroscopy images of Grad-CAM.Results:①In the five-class classification tasks,the accuracy of EfficientNet-B0 model(93.23％)was higher than that of Res-Net18 model(84.23％);the area under the curve(AUC)of EfficientNet-B0 model in the diagnosis of five disea-ses,including atypical endometrial hyperplasia,endometrial polyps,endometrial cancer,endometrial atypical hy-perplasia,and submucous myoma,was slightly higher than that of ResNet18 model,and the AUC of both models was almost above 0.980.②In the binary classification task of accuracy and the evaluation of specificity,the two models were similar,both above 93.00％,and the sensitivity of EfficientNet-B0 model(91.14％)was significantly better than that of ResNet18 model(77.22％).③EfficientNet-B0 model combined with Grad-CAM algorithm could identify the abnormal areas in the image.After biopsy and pathological examination,it was confirmed that about 95％of the marked areas in the model's output heatmap were lesion areas.Conclusions:The hysteroscopy di-agnostic model developed by EfficientNet-B0 model combined with Grad-CAM has high diagnostic accuracy,sen-sitivity,and specificity,and has application value in the diagnosis of endometrial lesions.
