Multimodal predictive classification of Alzheimer's disease based on attention‐combined fusion network: Integrated neuroimaging modalities and medical examination data

• DOI: 10.1049/ipr2.12841
• Publication Date: 2023-07-04T06:33:31Z
• Authors: Hui Chen; Huiru Guo; Longqiang Xing; Da Chen; Ting Yuan; Yunpeng Zhang; Xuedian Zhang
• Keywords: Modality (human–computer interaction); Feature (linguistics); Modalities

Abstract Early diagnosis of Alzheimer's disease (AD) plays a key role in preventing and responding to this neurodegenerative disease. It has shown that, compared with single imaging modality‐based classification AD, synergy exploration among multimodal neuroimages is beneficial for the pathological identification. However, effectively exploiting information still big challenge due lack efficient fusion methods. Herein, network based on attention mechanism proposed, which magnetic resonance (MRI) positron emission computed tomography (PET) images are converted into feature vectors same dimension, while demographic clinical data preprocessed through embedding. This model can focus important points, fuse more effectively, thus provide accurate prediction different stages. The results show that achieves an accuracy 84.1% triple tasks normal cognition (NC) versus mild cognitive impairment (MCI) AD 93.9% stable MCI (sMCI) progressive (pMCI). In contrast existing methods, our yields state‐of‐the‐art diagnosis, powerful promising practice.