The Use Of Artificial Intelligence In Estimating Lens Density And Subsequent Energy Parameters

Artificial intelligence (AI) is changing the way we approach medicine. Several AI algorithms to detect and grade cataracts have been developed, based on slit lamp and colour fundus photographs. Equally, there are now AI-based formulas to calculate required intra-ocular lens (IOL) power, prior to cataract surgery. With the advances made in AI, we can now use innovative software to analyse OCT lens images to estimate density and ultimately titrate both laser and ultrasound energy on a case by case basis thus limiting potential endothelial damage.

University Teaching Hospital

n this proof-of-concept study, MaZda texture analysis software programme (version 4.6) was used to analyse anonymised OCT images, obtained by a Victus femtosecond laser platform, prior to the femtosecond laser procedure. 34 images from patients about to undergo FLACS were included. The software produced unique histograms and datasets for each OCT image, based on a region of interest (ROI) focussed on the crystalline lens. Patient demographics were identified, and any correlation between histogram mean and age were calculated using Spearman’s rank correlation.

Using the MaZda software, 34 images were analysed. The mean age and ROI histogram mean were 73.8 years and 115.8, respectively. Spearman’s rank correlation coefficient for age in correlation to histogram means showed a statistically significant association (Rs=.344, p=.047).

Texture analysis correlates well with the age of patients, a surrogate marker of lens density. Since age and CDE are also strongly linked, it can be asserted that texture analysis can be used to modulate both laser and ultrasound energy on a bespoke patient by patient basis thus reducing patient exposure and potentially deleterious effects.