Classification Prediction of Alzheimer's Disease and Vascular Dementia Using Physiological Data and ECD SPECT Images

Abstract

Alzheimer's disease (AD) and vascular dementia (VaD) are the two most common forms of dementia. However, their neuropsychological and pathological features often overlap, making it difficult to distinguish between AD and VaD. In addition to clinical consultation and laboratory examinations, clinical dementia diagnosis in Taiwan will also include Tc-99m-ECD SPECT imaging examination. Through machine learning and deep learning technology, we explored the feasibility of using the above clinical practice data to distinguish AD and VaD. We used the physiological data (33 features) and Tc-99m-ECD SPECT images of 112 AD patients and 85 VaD patients in the Taiwanese Nuclear Medicine Brain Image Database to train the classification model. The results, after filtering by the number of SVM RFE 5-fold features, show that the average accuracy of physiological data in distinguishing AD/VaD is 81.22% and the AUC is 0.836; the average accuracy of training images using the Inception V3 model is 85% and the AUC is 0.95. Finally, Grad-CAM heatmap was used to visualize the areas of concern of the model and compared with the SPM analysis method to further understand the differences. This research method can quickly use machine learning and deep learning models to automatically extract image features based on a small amount of general clinical data to objectively distinguish AD and VaD.

Keywords: Alzheimer’s disease; ECD SPECT images; classification prediction; vascular dementia.

Figure 1

The flowchart of the entire data training process in this study.

Figure 2

The boxplots compare six features before and after normalization. (A,B) are distributions of Age; (C,D) are distributions of WBC; (E,F) are distributions of HDL; (G,H) are distributions of creatinine; (I,J) are distributions of free T4; (K,L) are distributions of folic acid. Outliers are indicated by a “+” symbol.

Figure 3

ROC curves of SVM model trained on physiological data for AD/VaD. (A) using complete data features (33); (B) using RFE-selected features (19).

Figure 4

ROC curves of InceptionV3 model trained on ECD image for AD/VaD.

Figure 5

The average Grad-CAM heatmap for AD and VaD groups. (A) Average Grad-CAM map of AD testing set. (B) Average Grad-CAM map of VaD testing set.

Figure 6

The histogram of pixel number in each brain region calculated from ‘lit-up’ areas from the Grad-CAM heatmap for AD and VaD groups. L, Left; R, Right; Sup, superior; Mid, middle; Inf, inferior; Med, medial; Ant, anterior; Post, posterior; Orb, orbital; Oper, operculum; Supp, supplementary; Tri, triangularis. The top ten are marked with square boxes, blue for the AD group and red for the VaD group. The square boxes with white background indicates greater than two-thirds of the maximum value; light gray represents greater than one-third of the maximum value; and dark gray represents less than one-third of the maximum value.