基于深度学习的超声影像组学在乳腺癌中的研究进展

Advances in Application of Ultrasound-based Deep Learning Radiomics in Breast Cancer

基于深度学习的影像组学(deep learning radiomics,DLR)通过不同构架从医学图像中提取深层特征,并将提取出的深层特征进一步分析,辅助临床决策。相比传统影像组学,DLR能够自动地提取深层特征,不依赖于医师人工标注,进一步提高其在肿瘤诊断及预测预后中的准确性和可靠性。超声检查是乳腺癌早期诊断的主要方式。全文分析近几年基于超声的DLR在乳腺肿物良恶性的鉴别诊断、乳腺癌分子分型的预测、腋窝淋巴结状态评估、新辅助化疗疗效评估中的研究现状。

Deep learning radiomics(DLR)extracts high-level features from medical images through different frameworks,and further analyzes the extracted high-level features to assist clinical decision-making.Compared with traditional radiomics,DLR can automatically extract high-level features without relying on manual annotation by physicians,further improving accuracy and reliability in tumor diagnosis and prognosis.Ultrasonography is the main way of early diagnosis of breast cancer.This article reviews the research progress of ultrasound-based DLR in the differential diagnosis of benign and malignant breast tumors,the prediction of breast cancer molecular typing,the assessment of axillary lymph node status,and the evaluation of neoadjuvant chemotherapy efficacy.