Related Experiment Video
Updated: Mar 29, 2026
05:36
Subjective Refraction Test Using a Smartphone for Vision Screening
Published on: October 18, 2024
2.0K
Prediction of Cataract Severity Using Slit Lamp Images from a Portable Smartphone Device: A Pilot Study.
David Z Chen1,2, Changshuo Liu3, Junran Wu3
1Department of Ophthalmology, National University Hospital, Singapore 119024, Singapore.
Sensors (Basel, Switzerland)
|March 28, 2026
Summary
This study shows that a smartphone-based slit lamp can predict cataract severity using deep learning, even without pupil dilation. This offers a potential new method for objective community screening of cataracts.
Area of Science:
- Ophthalmology
- Artificial Intelligence
- Medical Imaging
Background:
- Cataract diagnosis traditionally requires in-person dilated eye exams by ophthalmologists.
- Objective community screening for cataracts using portable, non-dilating devices is currently lacking.
Purpose of the Study:
- To investigate the feasibility of predicting cataract severity using deep learning on images from a smartphone-based slit lamp prototype.
- To assess the model's performance with and without pupil dilation.
Main Methods:
- A prospective cross-sectional pilot study captured slit lamp images using a smartphone prototype.
- A Swin transformer model was trained to classify cataract severity based on images.
- The Pentacam nuclear staging score (PNS) from dilated pupils served as the ground truth.
Main Results:
- The deep learning model achieved 81.25% accuracy for undilated images and 74.38% for dilated images.
- Heat maps successfully identified relevant anatomical areas in certain images.
- The average age of participants was 65.3 years, with an average PNS score of 1.57.
Conclusions:
- A portable smartphone slit lamp device shows promise for estimating cataract density without dilation.
- This technology could enable objective, community-based cataract screening.
- Further validation in larger, diverse populations is warranted.
