Related Experiment Video
Updated: Jun 16, 2025
09:47
Author Spotlight: Advancing Alzheimer's Research – Exploring Early Detection and Multi-Omics Approaches
Published on: December 15, 2023
998
A Convolutional Mixer-Based Deep Learning Network for Alzheimer's Disease Classification from Structural Magnetic
M Krithika Alias Anbu Devi1, K Suganthi1
1School of Electronics Engineering, Vellore Institute of Technology, Chennai 600127, India.
Diagnostics (Basel, Switzerland)
|June 13, 2025
Summary
This study introduces a novel deep learning model for Alzheimer's disease (AD) classification using structural MRI scans. The model achieves high accuracy in identifying AD stages, offering a promising tool for early diagnosis and intervention.
Area of Science:
- Neuroimaging and Artificial Intelligence
- Medical Image Analysis
- Machine Learning for Healthcare
Background:
- Alzheimer's disease (AD) is a progressive neurodegenerative disorder impacting cognitive function, necessitating accurate diagnostic tools for timely intervention.
- Current diagnostic methods require improvement for precise staging of AD progression, especially given the challenges of class imbalance in medical imaging datasets.
Purpose of the Study:
- To develop and validate a novel, efficient deep learning architecture for classifying Alzheimer's disease stages using structural MRI (sMRI).
- To address class imbalance in medical imaging datasets using a hybrid sampling technique.
- To enhance model interpretability through an explainable AI method.
Main Methods:
- A novel AD classification architecture combining depthwise separable and traditional convolutional layers for efficient feature extraction from sMRI scans.
- Implementation of a hybrid sampling approach (SMOTE + ENN) to mitigate class imbalance issues.
- Utilization of an Activation Space Occlusion Sensitivity Map (ASOP) for explainable AI, highlighting critical image regions influencing classification.
Main Results:
- The proposed model achieved superior performance compared to established transfer learning architectures (VGG19, DenseNet201, EfficientNetV2S, MobileNet, ResNet152, InceptionV3, Xception).
- Exceptional classification metrics were obtained: 98.87% accuracy, 98.86% F1 score, 98.80% precision, and 98.69% recall for AD stage classification.
- The hybrid sampling and ASOP methods effectively addressed class imbalance and provided visual explanations for model decisions.
Conclusions:
- The developed deep learning model demonstrates high efficacy and efficiency for classifying Alzheimer's disease stages from sMRI.
- The integration of advanced convolutional techniques, hybrid sampling, and explainable AI offers a robust solution for neurodegenerative disease diagnosis.
- This approach holds significant potential for improving early detection and personalized treatment planning in Alzheimer's disease management.
