Title: The advancement of ensemble deep learning architecture for the detection and classification of brain tumours with MRI images

Authors: Ashish Gupta; Deepak Gupta; Manisha Pathak; Sharmila K. Wagh

Addresses: NITM, Gwalior, Madhya Pradesh, 474001, India ' ITM, Gwalior, Madhya Pradesh, 474001, India ' MITS, Gwalior, Madhya Pradesh, 474001, India ' MESCE, Pune, Maharashtra, 474001, India

Abstract: This research work's objective is to automatically identify brain tumours using deep learning models and magnetic resonance imaging (MRI) images. The study employed ensemble deep learning architectures, including EfficientNet, VGG16, InceptionV3, and MobileNet, along with class activation maps (CAMs) indicators. The study was conducted in two stages. In the first stage, the presence of a tumour in the MR images and different tumour types were detected using a multi-class approach. The tumour types considered in the study were normal, glioma tumour, meningioma tumour, and pituitary tumour. The accuracy values obtained in this stage were reported as 97.15% with EfficentNet, 96.47% with VGG16, 88.03% with InceptionV3, and 91.64% with MobileNet architectures. In the second stage, CAMs were created for each tumour group. CAMs are visualisation techniques that highlight the regions of an image that contribute the most to a particular classification decision. CAMs can be used as additional.

Keywords: deep learning; EfficientNet; VGG16; InceptionV3; MobileNet; magnetic resonance imaging; MRI.

DOI: 10.1504/IJBET.2024.138619

International Journal of Biomedical Engineering and Technology, 2024 Vol.45 No.1, pp.27 - 44

Received: 19 May 2023
Accepted: 25 Aug 2023
Published online: 16 May 2024 *

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