# Inspecting the Retina: Oculomotor Patterns and Accuracy in Fundus Image Interpretation by Novice Versus Experienced Eye Care Practitioners

## TL;DR

Experienced eye care professionals interpret fundus images more accurately and with more focused eye movements than novices, suggesting that training could improve accuracy in less experienced clinicians.

## Contribution

This study reveals how expertise affects eye-tracking patterns and diagnostic accuracy in fundus image interpretation.

## Key findings

- Experienced practitioners had significantly higher diagnostic accuracy (83%) compared to novices (70%).
- Experienced clinicians showed more systematic visual search patterns, while novices had unorganized scan paths.
- Diagnostic accuracy in novices correlated with fixation count, and laterality accuracy correlated with total dwelling time.

## Abstract

Visual search behavior, influenced by expertise, prior knowledge, training, and visual fatigue, is crucial in ophthalmic diagnostics. This study investigates differences in eye-tracking strategies between novice and experienced eye care practitioners during fundus image interpretation. Forty-seven participants, including 37 novices (first- to fourth-year optometry students) and 10 experienced optometrists (≥2 years of experience), viewed 20 fundus images (10 normal, 10 abnormal) while their eye movements were recorded using an Eyelink1000 Plus gaze tracker (2000 Hz). Diagnostic and laterality accuracy were assessed, and statistical analyses were conducted using Sigma Plot 12.0. Results showed that experienced practitioners had significantly higher diagnostic accuracy (83 ± 6.3%) than novices (70 ± 12.9%, p < 0.005). Significant differences in oculomotor behavior were observed, including median latency (p < 0.001), while no significant differences were found in median peak velocity (p = 0.11) or laterality accuracy (p = 0.97). Diagnostic accuracy correlated with fixation count in novices (r = 0.54, p < 0.001), while laterality accuracy correlated with total dwelling time (r = −0.62, p < 0.005). The experienced practitioners demonstrated systematic and focused visual search patterns, whereas the novices exhibited unorganized scan paths. Enhancing training with visual feedback could improve fundus image analysis accuracy in novice clinicians.