Authors
T G Ramabharathi, Kamalraj Subramaniam
Published in
Psychiatry research. Neuroimaging. Volume 362. Pages 112269. Jun 08, 2026. Epub Jun 08, 2026.
Abstract
This article presents a unique deep learning technique to identify AD using data from magnetic resonance imaging (MRI). However, deep learning models' lack of interpretability prevents them from being used in clinical settings, where explainability is crucial for winning over medical personnel. In order to diagnose AD, this work proposes a self-attending bidirectional gated recurrent unit (SA-Bi-GRU) method based on explainable AI (XAI) that makes use of a deep learning model. Before the diagnosis process, an integrated Ternary pattern and Fourier-Bessel series expansion based empirical wavelet transform (TP-FBSE-EWT) method is used to extract features. Then, a hybrid binary teaching learning and Horse herd optimization (H-BTL-HHO) algorithm is presented to minimize the dimensions and screen properties of brain regions associated with AD. Additionally, by using the Gradient-weighted Class Activation Mapping (Grad-CAM) technique, the proposed structure seeks to improve the interpretability of deep learning models, giving clinicians important insights into disease diagnosis and an understanding of the decision-making process. The process is implemented using the MATLAB tool. The simulation findings reveal that the proposed CAD system for clinical score prediction outperforms prevailing systems by boosting accuracy, sensitivity, and specificity to 99.97%, 99.34%, and 98.89% for multi-class problems, respectively.
PMID:
42364284
Bibliographic data and abstract were imported from PubMed on 28 Jun 2026.
Read full publication at:
Please sign in
to see all details.
Advertisement
Stats
- Recommendations n/a n/a positive of 0 vote(s)
- Views 7
- Comments 0