Machine Learning Analysis Classifies Patients with Primary Angle-Closure Glaucoma Using Abnormal Brain White Matter Function

Abstract

Objective: Primary angle-closure glaucoma (PACG) is a globally prevalent, irreversible eye disease leading to blindness. Previous neuroimaging studies demonstrated that PACG patients were associated with gray matter function changes. However, whether the white matter(WM) function changes in PACG patients remains unknown. The purpose of the study is to investigate WM function changes in the PACG patients.

Methods: In total, 40 PACG patients and 40 well-matched HCs participated in our study and underwent resting-state functional magnetic resonance imaging (rs-fMRI) scans. We compared between-group differences between PACG patients and HC in the WM function using amplitude of low-frequency fluctuations (ALFF). In addition, the SVM method was applied to the construction of the PACG classification model.

Results: Compared with the HC group, ALFF was attenuated in right posterior thalamic radiation (include optic radiation), splenium of corpus callosum, and left tapetum in the PACG group, the results are statistically significant (GRF correction, voxel-level P < 0.001, cluster-level P < 0.05). Furthermore, the SVM classification had an accuracy of 80.0% and an area under the curve (AUC) of 0.86 for distinguishing patients with PACG from HC.

Conclusion: The findings of our study uncover abnormal WM functional alterations in PACG patients and mainly involves vision-related regions. These findings provide new insights into widespread brain damage in PACG from an alternative WM functional perspective.

Keywords: amplitude of low frequency fluctuations; primary angle-closure glaucoma; support vector machine; white matter.

Figure 1

Flow chart of preprocessing, metric calculation, and statistical analysis in this study. Data acquisition of fMRI and T1 in subjects (A) white matter function calculation(B) Differences in white matter function between the two groups and machine learning (C).

Figure 2

Results of the two components representing the white matter ALFF signal values by one-sample t-test in patients with (A) PACG and (B) HC.

Figure 3

Regions of WM showing ALFF differences between PACG and HC groups.

Figure 4

(A)Classification results using machine learning analysis based on ALFF values; (B)three dimensional confusion matrices from machine learning analysis; (C) function values of two groups with a scatter diagram; (D) the ROC curve of the SVM classifier with an AUC value of 0.86. (class 1: PACG group; class 2: HC group).