Predicting conversion of Alzheimer’s disease based on multi-modal fusion of neuroimaging and genetic data

Abstract

Abstract Identifying progressive mild cognitive impairment (pMCI) and stable mild cognitive impairment (sMCI) play a significant role in the early diagnosis of Alzheimer’s disease (AD) and can be helpful in early treatment to reduce the risk of conversion to AD. We proposed a classification method of sMCIs and pMCIs based on multi-modality data fusion of single-nucleotide polymorphisms (SNP), ratio of gray matter volume (RGV) obtained by morphometric measures, and sMRI images to predict the progression of AD. We validated the effectiveness of the proposed method by applying it to the task of identifying the disease status on the Alzheimer’s Disease Neuroimaging Initiative dataset. The results showed that the classification performances of our method was better than other state-of-the-art methods, and the accuracy rate for the classification of pMCI and sMCI reached 94.37%. The accuracy of our method was better than that of existing classification methods based on multi-modality images, and the accuracy rate for the classification of pMCI and sMCI reached 94.37%. Our study demonstrated that compared with unimodal and bimodal data, the method based on trimodal data fusion can better distinguish sMCI and pMCI, obtaining higher prediction accuracy for AD conversion. In addition, as a morphological feature, ratio of gray matter volume played a key role in distinguish of sMCI and pMCI, which can be used for the early diagnosis of AD.

Keywords

• AD
• sMCI
• pMCI
• Multi-modality data
• Prediction