Abstract

Info

Title:: Review of applications of deep learning in hysteroscopic image analysis

Author(s):: SONG Yang1; WANG Ranran2; ZHOU Jinting2; 1. School of Medicine, Wuhan University of Science and Technology, Wuhan 430000, China 2. Department of Obstetrics and Gynecology, Xiangyang Central Hospital, Hubei University of Arts and Science, Xiangyang 441100, China

Abstract:: Hysteroscopy is an indispensable technique in the diagnosis and treatment of endometrial lesions. In recent years, with the rapid development of artificial intelligence, deep learning technology has emerged as a novel approach for the in-depth analysis of hysteroscopic images, owing to its outstanding feature extraction capabilities and efficient learning performance on large-scale data. Through keyword retrieval, literature screening, quality assessment, and topic induction, this study systematically summarizes the applications of deep learning techniques such as CNN, DNN, U-Net, YOLO, and Transformer in hysteroscopic image analysis, encompassing the detection and classification of endometrial cancer, endometrial polyps, and endometritis, the therapeutic management of uterine fibroids, and the prediction of fertility outcomes following surgery for uterine adhesions. Additionally, the existing limitations and a prospect of the development trends regarding deep learning-based hysteroscopic image analysis are provided, expecting to promote the further advancement in relevant research.