Recent advances and clinical applications of deep learning in medical image analysis

Abstract

Deep learning has received extensive research interest in developing new medical image processing algorithms, and deep learning based models have been remarkably successful in a variety of medical imaging tasks to support disease detection and diagnosis. Despite the success, the further improvement of deep learning models in medical image analysis is majorly bottlenecked by the lack of large-sized and well-annotated datasets. In the past five years, many studies have focused on addressing this challenge. In this paper, we reviewed and summarized these recent studies to provide a comprehensive overview of applying deep learning methods in various medical image analysis tasks. Especially, we emphasize the latest progress and contributions of state-of-the-art unsupervised and semi-supervised deep learning in medical image analysis, which are summarized based on different application scenarios, including classification, segmentation, detection, and image registration. We also discuss major technical challenges and suggest possible solutions in the future research efforts.

Keywords: Attention; Classification; Deep learning; Detection; Medical images; Registration; Segmentation; Self-supervised learning; Semi-supervised learning; Unsupervised learning; Vision Transformer.

Conflict of interest statement

Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Copyright © 2022. Published by Elsevier B.V.

Figures

Figure 1.

The overall structure of this survey.

Figure 2.

A simple CNN for disease classification from MRI images (Anwar et al., 2018).

Figure 3.

(a) MoCo (He et al., 2020); (b) SimCLR (Chen et al., 2020a).

Figure 4.

Mean Teacher model application in medical image analysis (Li et al., 2020b). πi refers to the transformation operations, including rotation, flipping, and scaling. zi and z~i are network outputs.

Figure 5.

Units of different segmentation networks (a) forward convolutional unit (U-Net), (b) recurrent convolutional block (RCNN), (c) residual convolutional unit (residual U-Net), and (d) recurrent residual convolutional unit (R2U-Net) (Alom et al., 2018).

Figure 6.

(a) Transformer layer; (b) the architecture of TransUNet (Chen et al., 2021b)

Figure 7.

VoxelMorph (Balakrishnan et al., 2018).