Deep Learning Approaches for Early Diagnosis of Neurological Brain Disorders

Ahobila Sashank Sarma, Shaik Allabakash, S. Thenmalar

2025

Abstract

Neurological disorders such as Alzheimer's disease, Parkinson's disease, epilepsy, and stroke present significant challenges in early diagnosis and management. Deep learning has shown great potential in analyzing multi-modal neurological data, including medical imaging and genetic information. However, traditional deep learning models struggle with effective multi-modal data integration. Existing graph learning techniques address inter-subject relationships but face difficulties in optimally fusing imaging, genetic, and clinical data. To overcome these challenges, we propose an advanced deep learning-based Convolutional Neural Network (CNN) framework that enhances the Graph-Based Fusion (GBF) approach by incorporating convolutional and transformer-based models for multi-modal feature extraction and classification. Our imaging-genetic fusion module employs attention mechanisms to derive meaningful representations, while a multi-graph fusion module integrates imaging, genetic, and clinical features for improved diagnostic accuracy. Extensive validation using the ADNI and PPMI datasets demonstrates that the proposed deep learning- enhanced GBF model achieves an accuracy of 88%, outperforming traditional GBF techniques. This integration of deep learning with graph-based fusion provides a more precise and early detection framework for neurological disorders. The proposed deep learning-enhanced GBF model leverages attention mechanisms and multi-graph fusion to integrate imaging, genetic, and clinical data, achieving 88% accuracy on ADNI and PPMI datasets. This approach enhances diagnostic precision and facilitates early detection of neurological disorders.

Paper Citation

in Harvard Style

Sarma A., Allabakash S. and Thenmalar S. (2025). Deep Learning Approaches for Early Diagnosis of Neurological Brain Disorders. In Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25; ISBN 978-989-758-777-1, SciTePress, pages 431-442. DOI: 10.5220/0013899600004919

in Bibtex Style

@conference{icrdicct`2525,
author={Ahobila Sarma and Shaik Allabakash and S. Thenmalar},
title={Deep Learning Approaches for Early Diagnosis of Neurological Brain Disorders},
booktitle={Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25},
year={2025},
pages={431-442},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013899600004919},
isbn={978-989-758-777-1},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25
TI - Deep Learning Approaches for Early Diagnosis of Neurological Brain Disorders
SN - 978-989-758-777-1
AU - Sarma A.
AU - Allabakash S.
AU - Thenmalar S.
PY - 2025
SP - 431
EP - 442
DO - 10.5220/0013899600004919
PB - SciTePress