基于深度学习的医学图像识别研究进展

近年来,随着医学影像技术的快速发展,医学图像分析步入大数据时代,如何从海量的医学图像数据中挖掘出有用信息,对医学图像识别带来巨大的挑战。深度学习是机器学习的一个新领域,传统的机器学习方法不能有效地挖掘到医学图像中蕴含的丰富信息,而深度学习通过模拟人脑建立分层模型,具有强大的自动特征提取、复杂模型构建以及高效的特征表达能力,更重要的是深度学习方法能从像素级的原始数据中逐级提取从底层到高层的特征,这为解决医学图像识别面临的新问题提供了新思路。首先阐述深度学习方法,列举深度学习方法的三种常见的实现模型,并介绍深度学习的训练过程;随后总结了深度学习方法在疾病检测与分类和病变识别两方面的应用情况,以及深度学习应用在医学图像识别中的两个共性问题;最后对深度学习在医学图像识别中存在的问题进行分析及展望.

Research Progress of Medical Image Recognition Based on Deep Learning

In recent years, with the rapid development of medical imaging technology, medical image analysis has entered the era of big data. How to extract useful information from a large number of medical image data has become one great challenge to medical image recognition. Deep learning is a new field of machine learning, conventional machine learning method can’t effectively extract enough information contained in the medical image, while the deep learning has the power of establishing a hierarchical model, powerful automatic feature extraction, complex model building and efficient feature expression through the simulation of the human brain. More importantly, deep learning method can extract the features from the bottom to the top level from the original data of the pixel level, which provides a new way to solve the new problems faced by medical image recognition. Based on a large number of domestic and foreign literatures, this paper elaborated the three methods of depth learning, enumerated three common implementation models of deep learning methods, and introduced the training process of depth learning. We summarized the application of deep learning in two aspects of disease detection and classification and lesions recognition, and summarized the two common problems in the application of deep learning in medical image recognition. The analysis and prospects of deep learning in medical image recognition problems were proposed and discussed as well.