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

Figures & data

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).

Table 1 Demographics and Visual Measurements Between Two Groups

Condition	PACG Group	HC Group	T-Value	P-value
Gender (male/female)	23/17	19/21	N/A	0.370
Age (years)	57.31 ± 11.60	55.36 ± 13.52	0.692	0.491
BCVA-OD	0.38 ± 0.52	1.13 ± 0.16	−8.718	< 0.001*
BCVA-OS	0.43 ± 0.37	1.07 ± 0.25	−9.064	< 0.001*
IOP-L(mmHg)	21.95±4.07	15.26±1.25	−9.937	< 0.001*
IOP-R(mmHg)	22.58±5.71	15.75±2.18	−7.067	< 0.001*
MD-L(dB)	−11.76±9.25	N/A	N/A	N/A
MD-R(dB)	−12.77±10.56	N/A	N/A	N/A

Notes: Chi-square test for gender; Independent t-test was used for other normally distributed continuous data. Data are presented as mean ± standard deviation. *p < 0.001.

Abbreviations: PACG, primary angle-closure glaucoma; HC, healthy control; BCVA, best-corrected visual acuity; OD, oculus dexter; OS, oculus sinister; IOP, Intraocular pressure; MD, Mean deviation of visual field; L, left; R, right; N/A, not applicable.

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.

Table 2 Regions of WM Showing Abnormal ALFF in Patients with PACG Compared with HCs

Metrics	Tract (JHU-Atlas)	Voxels	MNI Coordinates			T value
			x	y	z
ALFF	Posterior thalamic radiation (include optic radiation) R	94	−33	−42	18	−3.42476
	Splenium of corpus callosum	65	−6	−36	9	−3.42013
	Tapetum_L	27	21	−42	21	−3.52558

Note: The statistical threshold was set at voxel with P <0:001 and cluster with P <0:05 for GRF correction.

Abbreviations: WM, white matter; ALFF, amplitude of low-frequency fluctuation; PACG, primary angle-closure glaucoma; HC, healthy control; MNI, Montreal Neurological Institute; x, y, and z are the locations of the peak voxels in standard MNI coordinates; R, right; L, left.

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).