AI-Assisted Retinal Vascular Measurement and its Association with White Matter Hyperintensities

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

This article is not in any list yet, why not save it to one of your lists.
Log in to save this article

Abstract

Objective To investigate the correlation between retinal vascular changes and the severity of white matter hyperintensities (WMH) in patients with presumed vascular origin of WMH using spectral-domain optical coherence tomography (SD-OCT), optical coherence tomography angiography (OCTA), and artificial intelligence (AI) technology. Methods A total of 126 WMH patients who were able to complete cranial MRI and fundus examination were included and divided into mild WMH group and moderate-to-severe WMH group. Retinal arteriovenous vascular imaging was obtained using SD-OCT, and various parameters of the vascular cross-section in the retinal images were automatically extracted using a self-developed ultra-fine semantic segmentation method and software [1] . Retinal capillary density images in the macular area were obtained using OCTA, and retinal blood flow density parameters were calculated using ImagJ software. Intergroup comparisons were made using t-tests, and the correlation between retinal vascular parameters and WMH severity was explored using multivariate logistic regression. A predictive model for WMH was constructed based on the multivariate logistic regression equation. Results There were statistically significant differences in age, systolic blood pressure (SBP), and diastolic blood pressure (DBP) between the two groups of subjects (P < 0.001). Compared with the mild WMH group, the moderate-to-severe WMH group had a wider retinal venular outer diameter (RVOD) in the temporal inferior retina (P = 0.014), and lower superficial capillary plexus-vascular perfusion density (SCP-VPD) (P = 0.036) and deep capillary plexus-vascular perfusion density (DCP-VPD) (P < 0.001). ROC curve analysis showed that the area under the curve (AUC) for predicting moderate-to-severe WMH based on age, DBP, SCP-VPD, and DCP-VPD were 0.678, 0.828, 0.608, and 0.815, respectively, and the AUC for the combined four indicators was 0.940. Conclusion Age and DBP are risk factors for WMH severity. Changes in retinal blood flow density and RVOD in the temporal inferior retina of the right eye are associated with WMH severity.

Article activity feed