NLM DIR Seminar Schedule
UPCOMING SEMINARS
-
July 3, 2025 Matthew Diller
Using Ontologies to Make Knowledge Computable -
July 15, 2025 Noam Rotenberg
Cell phenotypes in the biomedical literature: a systematic analysis and the NLM CellLink text mining corpus
RECENT SEMINARS
-
July 3, 2025 Matthew Diller
Using Ontologies to Make Knowledge Computable -
July 1, 2025 Yoshitaka Inoue
Graph-Aware Interpretable Drug Response Prediction and LLM-Driven Multi-Agent Drug-Target Interaction Prediction -
June 10, 2025 Aleksandra Foerster
Interactions at pre-bonding distances and bond formation for open p-shell atoms: a step toward biomolecular interaction modeling using electrostatics -
June 3, 2025 MG Hirsch
Interactions among subclones and immunity controls melanoma progression -
May 29, 2025 Harutyun Sahakyan
In silico evolution of globular protein folds from random sequences
Scheduled Seminars on Jan. 18, 2024
Contact NLMDIRSeminarScheduling@mail.nih.gov with questions about this seminar.
Abstract:
Age-related macular degeneration (AMD) is an eye disease that causes central vision loss in the elderly. AMD is a progressive disease with no symptoms at the early stages to severe symptoms like waviness of straight lines at the late stages. Reticular pseudodrusen (RPD) are subretinal drusenoid deposits (SDD) located above the retinal pigment epithelium layer. RPD are important because their highest rates of occurrence are predictors of progression to the end stages of AMD. Therefore, the detection of RPD can help manage AMD. RPD features are particularly distinguishable using volumetric spectral domain optical coherence tomography (SD-OCT). In this work, we used the age-related eye diseases study 2 (AREDS2) Ancillary OCT Study dataset which was obtained at the National Eye Institute (NEI) and other centers. SD-OCT scans can provide detailed information about the changes in retinal layers associated with RPD. The dataset included 1304 SD-OCT scans. We could transfer RPD labels from Fundus Auto Fluorescence (FAF) images taken at the same visit of each participant for 826 SD-OCT scans. We divided the labeled scans at the participant level into training (70%), validation (10%), and test (20%) sets. We developed 3D deep convolutional neural networks to process the whole SD-OCT scan volume. Our results on the test set showed an area under the receiver characteristic operating curve (AUC) of 0.88. We could improve the AUC to 0.91 by using semi-supervised learning with the unlabeled SD-OCT scans.