Attention-based Efficient Classification for 3D MRI Image of Alzheimer's Disease
Yihao Lin, Ximeng Li, Yan Zhang, Jinshan Tang
TL;DR
This paper tackles early Alzheimer’s disease diagnosis from 3D MRI by reducing computational cost without sacrificing accuracy. It introduces a CNN framework with a pre-trained ResNet backbone, a post-fusion scheme that converts 3D volumes into a discriminative 2D representation, and a Convolutional Block Attention Module to emphasize salient regions. Through RankSVM-inspired depth fusion and CBAM-based attention, the method achieves strong diagnostic performance while lowering training complexity compared to 3D CNN baselines. Evaluations on the ADNI dataset demonstrate that the proposed post-fusion approach with attention attains competitive metrics, suggesting meaningful improvements in efficient 3D MRI analysis for clinical screening. The work also outlines future enhancements, including preprocessing refinements and the integration of genomic or other non-imaging data to further boost diagnostic capability.
Abstract
Early diagnosis of Alzheimer Diagnostics (AD) is a challenging task due to its subtle and complex clinical symptoms. Deep learning-assisted medical diagnosis using image recognition techniques has become an important research topic in this field. The features have to accurately capture main variations of anatomical brain structures. However, time-consuming is expensive for feature extraction by deep learning training. This study proposes a novel Alzheimer's disease detection model based on Convolutional Neural Networks. The model utilizes a pre-trained ResNet network as the backbone, incorporating post-fusion algorithm for 3D medical images and attention mechanisms. The experimental results indicate that the employed 2D fusion algorithm effectively improves the model's training expense. And the introduced attention mechanism accurately weights important regions in images, further enhancing the model's diagnostic accuracy.