基于机器学习的乳腺癌影像基因组学研究综述
Survey on Machine Learning-based Radiogenomic Study of Breast Cancer
-
摘要: 影像基因组学是一种将基因组数据与成像特征关联的高通量研究方法, 现已广泛应用到乳腺癌分子亚型的鉴别、癌症风险的评估等方面。基于机器学习和大数据技术的影像基因组学在乳腺癌的个性化诊断和治疗等方面都显示出巨大的潜力, 因此, 对机器学习技术在乳腺癌影像基因组学中的研究现状和应用前景进行了总结。首先, 介绍乳腺癌的基因特征和乳腺癌影像数据获取方法, 分析机器学习技术在乳腺癌良性/恶性预测方面的应用; 然后, 对比应用于乳腺癌影像分割问题的深度学习方法, 并分析乳腺癌影像基因组学模型; 最后, 指出当前研究的局限性以及乳腺癌影像基因组学进一步的研究方向。Abstract: Radiogenomics is a high-throughput research method that correlates genomic data with imaging features, and is now applied widely to the identification of molecular subtypes of breast cancer and the assessment of cancer risk. Radiogenomics, based on machine learning and big data technologies, has shown tremendous potential in personalized diagnosis and treatment of breast cancer. To summarize the current research status and future prospects of machine learning technology in breast cancer radiogenomics, the genetic characteristics of breast cancer and the methods for obtaining breast cancer imaging data were first introduced, and the application of machine learning technology in predicting the benign/malignant nature of breast cancer was analyzed. Subsequently, deep learning methods applied to breast cancer image segmentation problems were compared and breast cancer radiogenomics models were analyzed. Finally, the current limitations of research and further research directions in breast cancer radiogenomics were pointed out.