Incurable Alzheimer's disease (AD) poses significant challenges for elderly individuals and their families, making early diagnosis crucial.
The authors introduce a new model called Local and Global Graph ConvNeXt, which combines convolutional neural networks and Transformers to better extract both local and global features from structural magnetic resonance imaging (sMRI).
Their model demonstrates impressive performance, achieving 95.81% accuracy while utilizing fewer parameters and floating point operations per second (FLOPS) compared to existing diagnostic models.