Comparing Three AI Models for Measuring Eye Movements
Author Information
Author(s): Muacevic Alexander, Adler John R, Chuntranapaporn Suthicha, Choontanom Raveewan, Srimanan Worapot
Primary Institution: Phramongkutklao Hospital
Hypothesis
This study primarily aimed to compare the accuracy of three convolutional neural network (CNN) models in measuring the four positions of ocular duction.
Conclusion
The three CNN models did not significantly differ in terms of efficacy in detecting various duction positions, but MobileNet showed potential advantages.
Supporting Evidence
- MobileNet had the shortest training time compared to ResNet and EfficientNet.
- All models achieved over 70% accuracy in detecting ocular duction.
- MobileNet demonstrated narrower confidence intervals for average prediction accuracy.
Takeaway
The study tested three computer models to see which one could best measure how our eyes move. They found that all three did a good job, but one model was faster and easier to use.
Methodology
The study included 526 subjects over 18, capturing ocular images in various gaze positions and measuring duction using three CNN models.
Potential Biases
The study did not include participants with ocular conditions, which may limit generalizability.
Limitations
The data used for evaluation were not collected from the same subjects, limiting the robustness of statistical comparisons.
Participant Demographics
Adult participants aged 18 and over visiting the ophthalmology outpatient department.
Statistical Information
P-Value
0.936
Confidence Interval
0.5-1 for MobileNet; 0.5-1.5 for ResNet and EfficientNet
Statistical Significance
p=0.936
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website