Abstract
Purpose :
Algorithms intended to identify and monitor pathology are frequently proposed using modalities that are not routinely performed on asymptomatic subjects. The goal of this study was to test the feasibility of acquiring a cohort of older controls for use in evaluating such algorithms.
Methods :
Subjects aged 50 or older were recruited. Central corneal thickness (CCT) measurements were obtained using Aladdin (Topcon Healthcare/VISIA, San Giovanni Valdarno, Italy), followed by the Maestro2 (Topcon Healthcare, Tokyo, Japan) to capture optical coherence tomography (OCT) and color fundus images for both eyes of each subject. Subjects then underwent binocular visual field testing using TEMPO/IMOvifa (Topcon Healthcare/CREWT Medical Systems, Tokyo, Japan). Finally, a clinician performed an eye exam, including refraction and Goldman Applanation Tonometry (GAT), and reviewed the Maestro2 results to confirm the absence of any pathologies. Clinician review of fundus images was performed to generate cup to disc ratios. To demonstrate the applicability of this data in evaluating various parameter combinations, 4 OCT only models published by Fukai et al (Trans Vis Sci Tech. 2022;11(8):8), the OHTS algorithm (Gordon et al, Ophthalmology. 2007;114(1):10-19) , and an algorithm that utilizes OCT and other clinical parameters to predict the likelihood of glaucoma (Screening Score) were generated for all subjects. Mean, standard deviation (SD), median, and ranges, were evaluated for all measured and calculated values.
Results :
Twenty-six subjects were recruited, including 10 males (38%) and 16 females (62%), with an average age of 61.3 (standard deviation 5.9) years. Two subjects were excluded, one for corneal edema and one for epiretinal membrane. Table 1 shows the distribution of tested parameters for the remaining 24 subjects, as well as the distribution of the different scores calculated from the measured data. IOP was typical of a normal population, ranging between 9 and 20 mmHg for both eyes. CCT ranged from 468 to 631 µm. Visual field MD ranged from -0.34 to 0.75 decibels (dB).
Conclusions :
Convenience sampling of normal subjects can provide multi-modal data for evaluating normal values in multi-modal algorithms. Additional data collection will be required to generate a statistically significant sample.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.