| Literature DB >> 36237718 |
Abstract
Medical image analyses have been widely used to differentiate normal and abnormal cases, detect lesions, segment organs, etc. Recently, owing to many breakthroughs in artificial intelligence techniques, medical image analyses based on deep learning have been actively studied. However, sufficient medical data are difficult to obtain, and data imbalance between classes hinder the improvement of deep learning performance. To resolve these issues, various studies have been performed, and data augmentation has been found to be a solution. In this review, we introduce data augmentation techniques, including image processing, such as rotation, shift, and intensity variation methods, generative adversarial network-based method, and image property mixing methods. Subsequently, we examine various deep learning studies based on data augmentation techniques. Finally, we discuss the necessity and future directions of data augmentation. CopyrightsEntities:
Year: 2020 PMID: 36237718 PMCID: PMC9431833 DOI: 10.3348/jksr.2020.0158
Source DB: PubMed Journal: Taehan Yongsang Uihakhoe Chi ISSN: 1738-2637
Fig. 1Various traditional augmentation examples of chest X-ray images using Albumentations. Numbers below images represent: [1] horizontal flip, [2] vertical flip, [3] elastic transform, [4] grid distort, [5] optical distort, [6] sharp, [7] motion blur, [8] gaussian blur, [9] rotation, [10] shift, [11] zoom, [12] crop, and [13] gamma correct.
Fig. 2Basic architecture of generative adverserial network with example images of chest X-ray.
Fig. 3Examples of chest X-ray images generated using progressive growing of generative adverserial networks.
Fig. 4Classification error rates of deep learning models trained with various data augmentation techniques on the CIFAR-10-C dataset. Adapted from Hendrycks et al. ArXiv Preprint 2019;arXiv: 1912.02781 (41).
CIFAR = Canadian Institute For Advanced Research
Fig. 5Examples of various chest X-ray augmented images using recent augmentation techniques.