HDFT-MViT: A Progressive Core-Enhanced Mix Framework for Alzheimer's Disease Classification using MRI images - Takeaways - MDSpire

HDFT-MViT: A Progressive Core-Enhanced Mix Framework for Alzheimer's Disease Classification using MRI images

  • By

  • Zhang, Dongyan

  • Zhang, Jincan

  • Liu, Bo

  • Liu, Min

  • Chen, Wenna

  • Du, Ganqin

  • May 31, 2026

  • 0 min

Share

  • 1

    HDFT-MViT is a lightweight hybrid architecture designed for classifying Alzheimer's disease using MRI imaging.

  • 2

    The model combines MobileNetV2 for local feature extraction and a lightweight Transformer for long-range dependency modeling.

  • 3

    HDFT-MViT employs a hierarchical dynamic filter and a channel attention mechanism to enhance feature discriminability.

  • 4

    It achieved state-of-the-art classification accuracies of 98.85% and 98.07% on the ADNI-1 and ADNI-2 MRI datasets, respectively.

  • 5

    The framework balances local detail perception and global semantic understanding while maintaining only 3.46 million parameters.

Original Source(s)

HDFT-MViT: A Progressive Core-Enhanced Mix Framework for Alzheimer's Disease Classification using MRI images

  • by Zhang, Dongyan , Zhang, Jincan , Liu, Bo , Liu, Min , Chen, Wenna , Du, Ganqin

  • May 31, 2026

Related Content

Stage-stratified benefits of AI-radiomics PET in early Alzheimer’s disease: a systematic review and meta-analysis

Expression of GPR34 in microglia remains stable in human Alzheimer’s disease

Frontal Release Signs and Future Decline in Research Participants With Intact Cognition