Validation of an Artificial Intelligence-Based Anterior Chamber Depth Estimation Using a Smartphone-Compatible Slit...

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Accurate anterior chamber depth (ACD) measurement is crucial for identifying glaucoma risk.
• Traditional methods for ACD measurement are time-consuming and technically demanding.
• An AI algorithm was previously developed to estimate ACD from slit lamp images.

Purpose of the Study:

• To evaluate the performance of a previously developed AI algorithm for ACD estimation.
• To assess the AI model's accuracy when used with the smartphone-compatible Smart Eye Camera (SEC).
• To compare AI-driven ACD estimates from the SEC against gold-standard anterior segment OCT (AS-OCT) measurements.

Main Methods:

• A retrospective analysis of images from 556 phakic eyes of 329 Asian patients was conducted.
• Anterior chamber depth (ACD) was measured using both the SEC with embedded AI and AS-OCT.
• Performance metrics included mean absolute error (MAE), mean squared error (MSE), Pearson correlation, and intraclass correlation coefficient (ICC).

Main Results:

• The AI algorithm integrated into the SEC achieved a mean absolute error (MAE) of 0.119 ± 0.0949 mm.
• A strong correlation (R = 0.922) was found between AI-estimated ACD and AS-OCT measurements.
• The intraclass correlation coefficient (ICC) of 0.903 indicated excellent agreement between the AI and AS-OCT methods.

Conclusions:

• The AI model within the SEC platform accurately estimates anterior chamber depth (ACD).
• The SEC offers a portable and user-friendly tool for ACD screening.
• This technology shows promise for widespread, accessible glaucoma risk assessment in various clinical settings.