Background Artificial intelligence (AI) in medical imaging diagnostics has huge potential, but human judgement is still indispensable. We propose an AI-aided teaching method that leverages generative AI to train students on many images while preserving patient privacy. Methods A web-based course was designed using 600 synthetic ultra-widefield (UWF) retinal images to teach students to detect disease in these images. The images were generated by stable diffusion, a large generative foundation model, which we fine-tuned with 6285 real UWF images from six categories: five retinal diseases (age-related macular degeneration, glaucoma, diabetic retinopathy, retinal detachment and retinal vein occlusion) and normal. 161 trainee orthoptists took the course. They were evaluated with two tests: one consisting of UWF images and another of standard field (SF) images, which the students had not encountered in the course. Both tests contained 120 real patient images, 20 per category. The students took both tests once before and after training, with a cool-off period in between. Results On average, students completed the course in 53 min, significantly improving their diagnostic accuracy. For UWF images, student accuracy increased from 43.6% to 74.1% (p<0.0001 by paired t-test), nearly matching the previously published state-of-the-art AI model’s accuracy of 73.3%. For SF images, student accuracy rose from 42.7% to 68.7% (p<0.0001), surpassing the state-of-the-art AI model’s 40%. Conclusion Synthetic images can be used effectively in medical education. We also found that humans are more robust to novel situations than AI models, thus showcasing human judgement’s essential role in medical diagnosis.

Smooth pursuit eye movements and saccadic eye movements are vital for precise vision. Therefore, tests for eye movement are important for assessing nervous or muscular diseases. However, objective measurements are not frequently performed due to the need for a polygraph system, electrodes, amplifier, and personal computer for data analysis. To address this, we developed an all-in-one eye-movement-measuring device that simultaneously presents visual stimuli, records eye positions, and examines its feasibility for evaluating eye movements. This device generates stimulus that induces eye movements and records those movements continuously. The horizontal or vertical eye movements of 16 participants were measured at various visual target speeds of 20–100 deg/s. The maximum cross-correlation coefficient (rho max) between the eye and visual target positions was used as an index of eye movement accuracy. A repeated-measures multi-way analysis of variance was performed, with the main effect being that rho max significantly decreased as the visual target speed increased. The average (±standard deviation) rho max values across all velocities were 0.995 ± 0.008 and 0.967 ± 0.062 in the horizontal and vertical directions, respectively, and were significantly higher for horizontal eye movements than for vertical eye movements. Moreover, rho max and saccadic frequency were significantly correlated for the slowest and fastest visual target motions. These suggest that our device enables accurate measurements of eye movements. We believe our new measurement device can be applied clinically for easily and objectively evaluating eye movements.

Athletes, particularly ball game players, benefit from focusing on fast-moving visual targets. In this study, we hypothesized that athletes, specifically professional basketball players, exhibit more accurate saccadic eye movements compared to non-athletes. To test this hypothesis, we assessed the eye movements of eight female professional basketball players from the Niigata Albirex BB (basketball player group) and compared them to eight sex-matched healthy volunteers from a previous study. Eye movements were measured using the Eye Score device. The visual target moved continuously for 20 s at a speed of 100 degrees/s in regular triangular wave-like movements horizontally. The subjects tracked the visual targets with maximum accuracy. The total amplitudes of evoked eye movements during tracking were 37.96 ± 0.82 for the basketball player group and 27.74 ± 2.55 deg (mean ± SEM) for the non-athlete group, indicating a significant difference. These findings suggest that basketball players may achieve accurate saccades by employing predictive saccades. The saccades of the basketball players were concentrated at the fifth frame (0.17 s after the onset of target motion), exhibiting a more peaked distribution than the normal curve. The results imply that athletes may execute predictive saccades when tracking targets with predictable movements, even outside of actual competition.