Title: Classification of magnetic resonance images of brain using concatenated deep neural network

Authors: Abhishek Das; Mihir Narayan Mohanty

Addresses: Department of Electronics and Communication Engineering, ITER, Siksha 'O' Anusandhan (Deemed to be University), Bhubaneswar, Odisha, India ' Department of Electronics and Communication Engineering, ITER, Siksha 'O' Anusandhan (Deemed to be University), Bhubaneswar, Odisha, India

Abstract: Deep learning is playing a vital role in medical image analysis due to its ability of auto feature extraction. Various works have been developed for brain image classification with models with high complexity but less performance. In our work, we have explored the deep learning techniques in the ensemble and stacking approach with less complexity and improved performance. Convolutional neural network, recurrent neural network, and long-short-term memory are used as base classifiers for feature extraction and first stage classification. The predictions of the base classifiers are fed to the multilayer perceptron model for second stage training and classification. The performance of the proposed model is verified with a brain magnetic resonance image dataset online available at Kaggle. 97% classification accuracy is achieved in the proposed method on the brain MRI dataset which is representing a competitive result concerning the state-of-the-art methods to the best of our knowledge.

Keywords: brain image classification; convolutional neural network; recurrent neural network; long short-term memory; LSTM; multilayer perceptron; MLP; ensemble learning.

DOI: 10.1504/IJMIC.2022.10052107

International Journal of Modelling, Identification and Control, 2022 Vol.41 No.1/2, pp.4 - 11

Received: 20 Mar 2021
Accepted: 07 Aug 2021

Published online: 22 Nov 2022 *

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