The Ontario Neurodegenerative Research Initiative retinal imaging dataset

Theme: Neuro-ophthalmology
What: Neuro-ophthalmology
Part of: Neuro-ophthalmology II / Neuro-ophtalmolgie II
When: 6/1/2024, 02:00 PM - 03:30 PM
Where: Room | Salle 714 B

Abstract

Purpose: To compare peripapillary retinal nerve fibre layer (pRNFL) and macular retinal (mRT) thicknesses between disease groups in the Ontario Neurodegenerative Disease Initiative (ONDRI). ONDRI is a longitudinal, multi-site, observational cohort study which includes participants with five diseases: (1) Alzheimer’s Disease/Mild Cognitive Impairment (AD/MCI) (2) Amyotrophic lateral sclerosis (ALS); (3) Frontotemporal dementia spectrum disorders (FTD) (4) Parkinson’s Disease (PD), (5) cerebrovascular disease (CVD) with a history of stroke. Amyloid negative cognitively normal controls (CNC) were included from the Brain Eye Amyloid Memory study.  Participants underwent clinical, neuropsychology, speech, eye tracking, gait/balance, MRI, genomics and SD OCT assessments.

Study Design: observational cohort

Methods:  Participants enrolled in ONDRI with SDOCT retinal images were included.  Exclusion criteria were history of glaucoma, wet AMD, retinal surgery and uncontrolled diabetes (high A1C).   Fundus photographs, 3 SDOCT images of the posterior pole and 3 images of the optic nerve head were acquired in both eyes using the Heidelberg Spectralis (software version 6.0.13.0, Heidelberg Engineering GmbH, Heidelberg, Germany). Expert observers inspected  fundus photographs to exclude glaucoma or suspect glaucoma, other optic neuropathies and maculopathies. SDOCT images were then inspected by trained and expert observers to exclude poor quality scans (Q score <17), maculopathies that could affect retinal thickness and suspect/confirmed optic neuropathies.  One posterior pole image and one pRNFL image underwent automated retinal segmentation (Heidelberg HEYEX version 6.3.4.0 software). Trained observers inspected the boundary lines of the cross-sectional B scans from the posterior pole image for segmentation errors in the ILM and BM for mRT, and the ILM and RNFL boundary lines for the pRNFL thickness, and manually corrected segmentation errors. Thickness values were batch exported. Mean/SD thickness of all 9 sectors within the ETDRS grid for mRT, and the global and 6 sectors for pRNFL were calculated. A linear mixed model tested for disease group as a predictor of pRNFL and mRT.

Results: There were 390 subjects (769 eyes) in the pRNFL dataset and 387 (755 eyes) in the mRT dataset (n= AD/MCI 86/86, PD 116/115, CVD 91/92, FTD38/36 ALS n=16/16, CNCs n=43/42 respectively). Mean age ranged from 61.3 years (ALS) to 71.3 years (AD/MCI). The FTD and PD groups were predictive of thicker pRNFL (p <0.001 and p=0.047 respectively), and FTD was predictive of thicker mRT (p=0.039). 

Conclusion: This dataset provides significant potential to explore associations with brain imaging, cognition, and other test platforms within disease and between disease groups.  Longitudinal and sublayer analyses can be explored to determine the utility for SDOCT to be used to identify potential biomarkers for neurodegenerative disease.

 

Presenter(s)

Presenting Author: Wendy Hatch

Additional Author(s):

Saffire Krance, Sunnybrook Health Sciences Centre

Christopher Hudson, University of Waterloo

Faryan Tayyari, Kensington Eye Institute

Edward Margolin, University of Toronto

Efrem Mandelcorn, University of Toronto

Jonathan Micieli, University of Toronto

M. Amin Banihashemi, University

Carmen Balian, Kensington Eye Institute

Richard Cheng, Kensingtonhealth.org

Ilse Belgraver, Baycrest Health Sciences

---