引用文本：丘清, 胡磊, 刘再毅. 基于医学影像的计算机辅助诊断在胃肠道恶性肿瘤诊疗决策中的研究进展[J/CD]. 消化肿瘤杂志（电子版）, 2025, 17(2):237-243.

作者：丘清，胡磊，刘再毅

单位：南方医科大学附属广东省人民医院（广东省医学科学院） 放射科，广东 广州 510000

Authors：Qiu Qing, Hu Lei, Liu Zaiyi

Unit：Department of Radiology, Guangdong Provincial People’s Hospital (Guangdong Academy of Medical Sciences), Southern Medical University, Guangzhou 510000, Guangdong, China

摘要：

胃肠道恶性肿瘤是全球公共卫生的重要威胁。传统医学影像存在海量数据处理效率低下、肿瘤异质性识别有限及医师主观差异等诸多局限性。近年来，计算机辅助诊断（computer-aided diagnosis,  CAD）技术展现出重要价值。CAD的关键技术包括影像组学、机器学习和深度学习。传统影像组学依赖于手工特征提取与机器学习模型，存在流程繁琐、模型简单、泛化能力不足等瓶颈。深度学习具有强大的图像自动分割和自动特征提取能力、优秀的表示能力及泛化能力，能够出色地处理大规模和高维度数据。本文重点综述了基于医学影像的深度学习技术在胃肠道恶性肿瘤筛查、诊断、治疗及预后等方面的最新研究进展。

关键词：胃肠道恶性肿瘤；医学影像；计算机辅助诊断；深度学习

Abstract：

Gastrointestinal malignancies represent a critical global public health challenge. Conventional medical imaging techniques face multiple limitations, including inefficient processing of massive data volumes, inadequate identification of tumor heterogeneity, and inter-observer variability among clinicians. In recent years, computer-aided diagnosis (CAD) technology has emerged as a pivotal solution, with its core methodologies encompassing radiomics, machine learning, and deep learning. Traditional radiomics approaches, reliant on manual feature extraction and conventional machine learning models, are constrained by labor-intensive workflows, oversimplified models, and limited generalization capabilities. In contrast, deep learning demonstrates superior performance through automated image segmentation, self-optimized feature extraction, enhanced representational capacity, and exceptional generalization ability, particularly in handling large-scale, high-dimensional datasets. This review focuses on recent advancements in deep learning-based medical imaging technologies for gastrointestinal malignancies, systematically analyzing their transformative potential in clinical decision-making and precision oncology.

Key words：Gastrointestinal malignancies; Medical imaging; Computer-aided diagnosis; Deep learning

注：网络优先发布