ARVO 2024: Feasibility and clinical efficacy of an OCT B-scan of interest tool

At this year's ARVO meeting in Seattle, Washington, the Eye Care Network spoke with Niranchana Manivannan, PhD. She spoke about the poster presentation, "Feasibility and clinical efficacy of an OCT B-scan of interest tool."

Video Transcript:

Note: The below transcript has been lightly edited for clarity.

Niranchana Manivannan, PhD:

[The] poster is titled, feasibility and clinical efficacy of the OCT B-scan of interest tool. What it really does is there is this cool deep learning-based AI tool, which looks at any of the OCT macular volumes obtained from Cirrus, and then it can highlight scans or petal scans which contains pathology of interest. To develop this, we worked with retina specialists and graded over 76,000 B-scans. This poster specifically is looking at the clinical efficacy of the tool. To do that, we took 2 different data sets, and it has data from all 4000, 5000, and 6000 models from Cirrus, and those data are graded by experts, and the clinical performance is evaluated. I'm very happy to say that the clinical performance the efficacy is highly effective. We have specificity and accuracy over 90% and sensitivity is about 87.8%. And then one of the important thing[s] is in a typically healthy cohort, which is mean in the healthy population. The negative prediction values over 99%, and that was really shows us that this clinical tool can be really useful in the clinic to look for the B-scans in OCT volumes, which contains a pathologies [that] are of interest to the doctors.