Radiomic study for the diagnosis of Alzheimer's disease based on the structure anomaly of hippocampus

Objective To discuss the radiomic features of hippocampus related to Alzheimer disease (AD) diagnosis, and to build and compare the right and left hippocampal classification models for AD diagnosis. Methods A prospective study was conducted on AD patients who visited the memory clinic of Zhejiang Provincial People's Hospital from September 2016 to June 2018. Fifty-nine AD subjects and forty-four normal controls (NC) were scanned with three-dimensional T1-weighted magnetization-prepared rapid gradient echo (MPRAGE) sequence by GE 3. 0 T MR. The right and left hippocampus were segmented for each subject using an efficient learning-based deformable model. The two groups of data passed through feature calculation and feature selection respectively. Then two classification models of the right and left hippocampus were built respectively based on the selected features using the Logistic regression model. Results There were 385 features obtained after calculation from each group of hippocampus. And there were five features selected of the left and right hippocampus respectively. The accuracy, area under the receiver operating characteristic curve values, specificity, sensitivity, positive predictive value and negative predictive value of the right hippocampus model evaluation indexes were 0.69, 0.74, 50%, 83%, 68% and 70%, respectively. And those of the left hippocampus model were 0.75, 0.83, 79%, 72%, 81% and 69%, respectively. Conclusions The results demonstrated the potential of hippocampal radiomic features as a biomarker for the diagnosis of AD. The radiomic models of right and left hippocampus are helpful for classifying AD from NC subjects. And the model of left hippocampus has a higher diagnostic value than the right. Key words: Hippocampus; Alzheimer's disease; Magnetic resonance imaging; Radiomics