消化道内窥镜图像异常的人工智能诊断方法研究进展

Advances in Artificial Intelligence Diagnosis Methods for Abnormal Images of Digestive Tract Endoscopy

随着深度学习的出现,图像处理不再局限于人工提取特征,转而对图像进行端到端的预测,实现了人工智能在图像处理领域的又一历史性飞越。作为人工智能医疗领域的热点应用,内镜图像异常检测能够准确快速地筛选整个消化道的异常,为医生提供诊断帮助。该文围绕消化道图像最为常见的息肉、出血、溃疡等异常,对其智能诊断方法展开研究,并探讨机器学习在消化内镜异常检测的应用现状,最后展望了未来消化道内窥镜病灶智能诊断的研究方向。

With the advent of deep learning,image processing is no longer limited to manually extracting features.End-to-end prediction of images has achieved another historic leap in artificial intelligence in the field of image processing.As a central application in the field of artificial intelligence medical treatment,endoscopic image abnormality detection can accurately and quickly screen abnormalities in the entire digestive tract,and provide doctors with diagnostic assistance.This article focuses on the most common abnormalities in the digestive tract images,including polyps,hemorrhage and ulcers.The application status of machine learning in the detection of digestive endoscopy abnormalities is discussed.Finally,the future research directions of intelligent diagnosis of digestive endoscopy lesions are prospected.