Title: mUNET: glioma segmentation with delineation of tumour sub-regions using optimised architecture of UNET

Authors: Sonal Gore; Jayant Jagtap

Addresses: Symbiosis International University, Pune and Department of Computer Engineering, Pimpri Chinchwad College of Engineering, Pune, India ' Department of Electronics and Telecommunication Engineering, Symbiosis Institute of Technology, Symbiosis International University, Pune, India

Abstract: Glioma tumour is aggressive due to infiltrative nature, that exhibits unpredictable shape, size, morphology. Therefore, its manual segmentation from MRI remains challenging task. Study presents automated segmentation using T2-FLAIR data of BRATS challenge by proposing modified UNET (mUNET) with optimised architecture to segment tumour in enhancing, necrosis and oedema sub-regions. Work was evaluated using six test cases by proposing slight modifications in UNET. Experimentation has gained comparable results with slightly better accuracy of 99.39% and loss of 0.0054, as compared to original UNET. Moreover, test cases were capable to complete segmentation at speed of 0.713 milliseconds per image.

Keywords: brain tumour; glioma; MRI; FLAIR; segmentation; deep learning; convolutional neural network; modified UNET; mUNET: BRATS data; tumour subregions.

DOI: 10.1504/IJMEI.2024.139891

International Journal of Medical Engineering and Informatics, 2024 Vol.16 No.4, pp.312 - 323

Received: 25 Sep 2021
Accepted: 26 Feb 2022
Published online: 09 Jul 2024 *

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