Management of Vogt-Koyangi-Harada disease with deep learning-based volume image enhancement and quantification of optical coherence tomography

Abstract

Purpose : Investigate the relationship between quantitative biomarkers obtained with optical coherence tomography (OCT) and a deep learning analysis and disease recurrence and management.

Methods : A single-medical center retrospective analysis was designed. This study obtained scans of swept-source OCT at a medical center. Thirty-three eyes of 17 patients (7 females, 10 males) with Vogt-Koyanagi-Harada disease (VKH) or sympathetic ophthalmitis (SO) were imaged consecutively between October 2012 and January 2021. Choroidal vessel structure was segmented and visualized in 3D, after which quantitative vessel volume maps are generated. Region-based choroidal vessel volume (CV), choroidal stroma volume (SV), and vessel index (VI) were analyzed for disease severity.

Results : OCT-based CV volume maps disclose regional CV changes in patients with VKH or SO. Two metrics in recurrence VKH, (i) choroidal volume at 1-, 3- and 6-month (p=0.01, 0.04 and 0.03), (ii) CV volume at one months (p=0.03), were higher than well treated VKH.

Conclusions : The deep-learning analysis of OCT images described here provides a 3D visualization of how the choroid may reflect disease severity in VKH patients. Moreover, CV or SV before treatment may become the biomarker for recurrence of VKH.

This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.