eduzhai > Applied Sciences > Engineering >

Tensorizing GAN with High-Order Pooling for Alzheimers Disease Assessment

  • king
  • (0) Download
  • 20210507
  • Save

... pages left unread,continue reading

Document pages: 15 pages

Abstract: It is of great significance to apply deep learning for the early diagnosis ofAlzheimer s Disease (AD). In this work, a novel tensorizing GAN with high-orderpooling is proposed to assess Mild Cognitive Impairment (MCI) and AD. Bytensorizing a three-player cooperative game based framework, the proposed modelcan benefit from the structural information of the brain. By incorporating thehigh-order pooling scheme into the classifier, the proposed model can make fulluse of the second-order statistics of the holistic Magnetic Resonance Imaging(MRI) images. To the best of our knowledge, the proposed Tensor-train,High-pooling and Semi-supervised learning based GAN (THS-GAN) is the first workto deal with classification on MRI images for AD diagnosis. Extensiveexperimental results on Alzheimer s Disease Neuroimaging Initiative (ADNI)dataset are reported to demonstrate that the proposed THS-GAN achieves superiorperformance compared with existing methods, and to show that both tensor-trainand high-order pooling can enhance classification performance. Thevisualization of generated samples also shows that the proposed model cangenerate plausible samples for semi-supervised learning purpose.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×