A Novel 3D Corneal Diameter Profile And Artificial Intelligence Driven Estimation Of Vault Space After Implantation Of Implantable Collamer Lens (Icl)

This study aims to improve the prediction of vault space following  implantable collamer lens (ICL) surgery based on 3D corneal diameter (CD) quantification and artificial intelligence (AI) analyses

Tertiary Eye Care Centre, India 

We retrospectively analyzed 50 eyes post-ICL surgery, using scheimpflug corneal tomography scans for 3D CD profiles. Parameters like minimum, maximum, horizontal, vertical, delta (max - min), meridian of minimum, and meridian of maximum CD were extracted. A skewed Gaussian curve was fitted to delta CD values, providing amplitude, center, sigma, and gamma parameters. The post-operative vault reported in patients was quantified using ultrasound biomicroscopy (UBM). UBM also provided a predicted vault space based on a virtual implantation of the selected ICL on the UBM image. The AI model combined the UBM predicted vault space with 3D corneal diameter and compared the final prediction with post-operative vault space reported in patients.

The AI identified maximum CD, maximum angle, mean nasal angle and UBM predicted post-operative vault space as significant determinants of the final obtained vault space in patients (R²=0.74, p<0.001). The mean error between the AI predicated vault space and measured in patient post-operatively was 0±170 micron. Intra-class correlation between measured post-op vault and UBM predicted vault space was 0.85. This improved to 0.92 between measured post-op vault and AI predicted vault space.

Our study suggests a significant improvement in predicted vault space when combing AI and 3D corneal diameter with the UBM predicted vault space. Further, analyses with larger sample size are required to confirm the results.