Automated Quantification of B-scan-linked Hypertransmission from SD-OCT Scans in Dry Age-Related Macular Degeneration

Abstract

Purpose : Predictive biomarkers for the development of geographic atrophy (GA) in dry age-related macular degeneration (AMD) are needed to better identify patients who may benefit from treatment or also to help stratify future clinical trials. Hypertransmission defects have been described as a potential important SD-OCT biomarker for GA progression risk that has predominantly been described utilizing en face quantification. The goal of this analysis to evaluate the potential of a B-scan based hypertransmission machine learning-enabled quantification tool to evaluate eyes with dry AMD and compare baseline hypertransmission between eyes that convert to GA and those that remain stable.

Methods : This retrospective image analysis study was an IRB-approved assessment of eyes with dry AMD. All eyes had intermediate AMD with without GA at baseline. Eyes were then split into 2 groups based on GA conversion within 2 years of baseline. All SD-OCT scans were analyzed with an automated hypertransmission quantification platform. The amount of baseline hypertransmission defects were then compared between eyes that converted to GA and eyes that did not.

Results : Overall, 969 patients with intermediate AMD without GA and 2 years of follow-up were included. Of these 969, 15% were in the converter group based on the GA status after 2 years. In comparing the amount of baseline hypertransmission, eyes that converted to GA has approximately 17% greater amount of hypertransmission than eyes that did not convert. In the converter group had a mean macular hypertransmission burden of 6.85% (~ 2.47 mm²) and those in the non-converter group had a mean hypertransmission burden of 5.87% (~ 2.11 mm², p < 0.0001).

Conclusions : Using an automated B-scan based hypertransmission ML tool, significantly greater hypertransmission burden was identified at baseline in eyes that ultimately develop GA compared to eyes that do not. Further research will focus on different techniques for assessing hypertransmission and comparative assessment to other biomarkers for progression risk (e.g., ellipsoid zone integrity).

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.