Related Experiment Video
Updated: Jul 31, 2026
12:50
Lesion Explorer: A Video-guided, Standardized Protocol for Accurate and Reliable MRI-derived Volumetrics in Alzheimer's Disease and Normal Elderly
Published on: April 14, 2014
Combined transformer encoder-CNN architecture with texture features for MRI-based Alzheimer's disease detection
G Divya Jyothi1, C Madan Kumar2, Ravi Kumar Kottala3
1Department of Computer Science and Engineering SR University, Ananthasagar, Hasanparthy, Warangal 506371, India.
Psychiatry Research. Neuroimaging
|July 3, 2026
Summary
This study introduces a novel deep learning model for early Alzheimer's disease (AD) detection using MRI scans. The combined modified Transformer Encoder-Convolutional Neural Network (CNN) achieved 96% accuracy, improving early diagnosis of Alzheimer's disease.
Area of Science:
- Neuroimaging
- Artificial Intelligence in Medicine
- Biomedical Engineering
Background:
- Alzheimer's disease (AD) onset precedes clinical symptoms, necessitating early diagnostic methods.
- Current diagnostic tools often fail to detect subtle early-stage structural brain changes in AD.
- This gap delays critical medical interventions for Alzheimer's disease.
Purpose of the Study:
- To propose a novel hybrid deep learning architecture for early Alzheimer's disease detection using MRI.
- To enhance the accuracy and reliability of Alzheimer's disease diagnosis through advanced image processing and feature extraction.
- To address the limitations of current diagnostic techniques in identifying early-stage AD.
Main Methods:
- A combined modified Transformer Encoder-Convolutional Neural Network (CNN) architecture was developed for AD detection.
- Magnetic Resonance Imaging (MRI) data underwent preprocessing with a Modified Gaussian Filtering technique (modGFT) for noise reduction and edge preservation.
- Feature extraction included shape features, Improved Median Robust Extended Local Binary Pattern (ImpMRELBP), and Pyramid Histogram of Oriented Gradients (PHOG).
- Data augmentation techniques were employed to increase the training dataset size and diversity.
Main Results:
- The proposed hybrid modTransEncd-CNN model demonstrated superior performance compared to traditional methods.
- The model achieved a high accuracy of 96% in classifying Alzheimer's disease.
- Integration of batch normalization and an enhanced attention module improved model training efficiency and generalization.
Conclusions:
- The developed modTransEncd-CNN model offers a promising approach for accurate and early detection of Alzheimer's disease.
- The hybrid architecture effectively leverages deep learning for analyzing MRI data in AD diagnosis.
- This advancement has the potential to significantly improve timely intervention strategies for Alzheimer's disease patients.