Title: A comprehensive analysis of GAN-based techniques for medical image synthesis

Authors: S. Sandhya; M. Senthil Kumar; B. Chidhambararajan

Addresses: Department of IT, SRM Valliammai Engineering College, Kattankulathur, India ' Department of Cyber Security, SRM Valliammai Engineering College, Kattankulathur, India ' Department of ECE, SRM Valliammai Engineering College, Kattankulathur, India

Abstract: Identifying the internal structure from the multi-modal medical images has gotten prominent consideration in recent years. The task of such identification is done with the help of an unsupervised deep learning technique known as generative adversarial network which is utilised in the field of image processing like registration of images, augmentation, generation of medical images, reconstruction of images and finally image translation. With these vast applications of generative adversarial network, it is used in medical imaging for the purpose of generating the medical images and cross modality medical image synthesis. Cross-modality synthesis, which creates CT-like pictures from MR images, is thought to be helpful for a number of reasons, one of which is to cut down on the additional acquisition time and expense. Creating new training samples with the appearance confined by the anatomical components defined in the accessible modality is another rationale. This paper aims at providing the various available architectures under generative adversarial network for the synthesis of medical images across cross modalities.

Keywords: deep learning; medical imaging; generative adversarial network; GAN; unsupervised deep learning; cross modality.

DOI: 10.1504/IJCRC.2024.138265

International Journal of Creative Computing, 2024 Vol.2 No.2, pp.172 - 177

Received: 19 Apr 2023
Accepted: 02 Aug 2023
Published online: 30 Apr 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article