Deep learning in medical image registration: a survey
- PDF / 2,220,548 Bytes
- 18 Pages / 595.276 x 790.866 pts Page_size
- 8 Downloads / 347 Views
ORIGINAL PAPER
Deep learning in medical image registration: a survey Grant Haskins1 · Uwe Kruger1 · Pingkun Yan1 Received: 26 February 2019 / Revised: 20 December 2019 / Accepted: 14 January 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract The establishment of image correspondence through robust image registration is critical to many clinical tasks such as image fusion, organ atlas creation, and tumor growth monitoring and is a very challenging problem. Since the beginning of the recent deep learning renaissance, the medical imaging research community has developed deep learning-based approaches and achieved the state-of-the-art in many applications, including image registration. The rapid adoption of deep learning for image registration applications over the past few years necessitates a comprehensive summary and outlook, which is the main scope of this survey. This requires placing a focus on the different research areas as well as highlighting challenges that practitioners face. This survey, therefore, outlines the evolution of deep learning-based medical image registration in the context of both research challenges and relevant innovations in the past few years. Further, this survey highlights future research directions to show how this field may be possibly moved forward to the next level. Keywords Image registration · Deep learning · Medical imaging · Convolutional neural networks
1 Introduction Image registration is the process of transforming different image datasets into one coordinate system with matched imaging contents, which has significant applications in medicine. Registration may be necessary when analyzing a pair of images that were acquired from different viewpoints, at different times, or using different sensors/modalities [41,122]. Until recently, image registration was mostly performed manually by clinicians. However, many registration tasks can be quite challenging and the quality of manual alignments is highly dependent upon the expertise of the user, which can be clinically disadvantageous. To address the potential shortcomings of manual registration, automatic registration has been developed. Although other methods for automatic image registration have been extensively explored prior to (and during) the deep learning renaissance, deep learning has changed the landscape of image registration This work was partially supported by NIH/NIBIB under awards R21EB028001 and R01EB027898, and NIH/NCI under a Bench-to-Bedside award.
B 1
Pingkun Yan [email protected] Department of Biomedical Engineering, Rensselaer Polytechnic Institute, Troy, NY 12180, USA
research [4]. Ever since the success of AlexNet in the ImageNet challenge of 2012 [3], deep learning has allowed for state-of-the-art performance in many computer vision tasks including, but not limited to: object detection [84], feature extraction [37], segmentation [87], image classification [3], image denoising [112], and image reconstruction [115]. Initially, deep learning was successfully used to augment the performan
Data Loading...
