Home optical coherence tomography (OCT) offers an opportunity for personalizing wet age-related macular degeneration (AMD) monitoring. Nodal Home OCT is a patient-operated OCT system, and Notal OCT Analyzer is an artificial intelligence (AI) algorithm that detects and quantifies retinal fluid in exudative AMD. Researchers analyzed the Notal OCT Analyzer’s ability to identify and quantify retinal fluid in output of a patient self-operated home OCT device and found that AI performed comparably to human readers. The results of the study were published as part of the American Society of Retina Specialists 2020 Virtual Annual Meeting.
Data were collected from five clinics where patients with AMD self-imaged with Notal Home OCT. From each scan (88 B-scans), an average of 10 B-scans were selected for manual segmentation. Each B-scan was labeled pixel-wise into four compartments: vitreous/outer layers, retina, subretinal fluid (SRF), and intraretinal fluid (IRF). Eyes were randomized 8:1 to a learning or validation set.
For the training and validation dataset generation for segmentation, there were 355 eyes from 239 patients with AMD (mean age, 78 years; range, 54-92 years). A total of 3,428 B-scans were manually segmented, 75% of which had fluid, including 2,936 B-scans of 311 eyes in the learning set and 492 B-scans of 44 eyes in the validation set. For the training and validation dataset generation for classification, there were 553 eyes from 312 patients with AMD (mean age, 78 years; range, 45-97 years).
“Nodal OCT Analyzer performed well compared with the human grader for B-scan segmented fluid area,” the researchers noted. The Pearson correlation of fluid area was 0.98 for SRF (P<0.00001) and 0.90 for IRF (P<0.00001). The SRF pixel-wise recall was 0.72 and precision was 0.86. The IRF recall and precision were 0.80 and 0.77, respectively. The sensitivity for detecting the presence of SRF in B-scans was 0.99 and the specificity was 0.98. The sensitivity and specificity for detecting IRF were 0.99 and 0.97, respectively.
Daily home imaging generates a high volume of images and requires AI-based physician assistance to identify, quantify, and track disease activity biomarkers, according to the researchers.
“This home-based OCT, which can analyze fluid status, its dynamics, and visualize the locations of fluid, has the potential to accurately monitor AMD disease activity in a patient self-operated home-use environment,” the researchers concluded, “… which will be a paradigm shift in the management of [these] patients.”
Kim JE, Lally DR, Elman MJ, et al. Performance of a Novel Deep Learning Algorithm for Automatic Retinal Fluid Quantification in Home OCT Images. Presented during the ASRS 2020 Virtual Annual Meeting, July 24-26, 2020.