Gurpal Virdi, MD

Publications

Research

Peer-reviewed research on EVO ICL sizing, vault prediction, and outcomes: the science behind the AI sizing tools I build.

My research centers on the hardest technical problem in modern refractive surgery: predicting Implantable Collamer Lens (ICL) vault and choosing the right EVO / EVO+ ICL size for an individual eye. Using deep learning and anterior-segment OCT (AS-OCT), my collaborators and I built VAULT and VAULT-OCT, image-based artificial-intelligence models that estimate postoperative vault from a preoperative scan, with the goal of reducing vault-related ICL exchanges and improving refractive outcomes for patients with high myopia and astigmatism. The work spans sizing accuracy, safety in shallow anterior chambers, and how vault behaves in the living eye.

The VAULT Method

The VAULT Method is how I approach ICL sizing: predict the best-fitting lens for the individual eye from anterior-segment imaging, rather than choosing a size from white-to-white corneal diameter and a single formula (the OCOS, Reinstein, KS, NK, or Parkhurst nomograms). It is grounded in the peer-reviewed VAULT and VAULT-OCT models below and a dataset of 756 real EVO ICL surgical cases, and it is applied in ICL Fit. The guiding principle: postoperative vault is a physiologic range, not a fixed micron target, so the goal is best fit for each eye, not one universal number.

Peer-reviewed articles

Featured data: vault is dynamic

From Dynamic Changes in ICL Vault and Anterior Chamber Angle Under Varying Lighting Conditions (Clinical Ophthalmology, 2026), where we imaged 100 post-ICL eyes under scotopic, mesopic, and photopic lighting. The finding underpins how I approach sizing: vault is a physiologic range, not a fixed micron target, so the goal is best fit for the individual eye.

Box plot of ICL vault in microns across scotopic, mesopic, and photopic lighting, showing mean vault falling from about 496 to 394 microns as lighting brightens.
Vault by lighting. Mean vault dropped from 495.6 µm (dark) to 393.8 µm (bright): a ~100 µm swing in the same eye (p < 0.0001).
Box plot of anterior chamber angle in degrees across scotopic, mesopic, and photopic lighting, showing the angle widening in brighter light.
Angle by lighting. The anterior chamber angle widened from 27.2° to 29.7° under bright light; vault and angle track with pupil size.
Scatter plot showing eyes with higher mean vault also have greater vault variability across lighting.
Higher vaults vary more. Variability rose with mean vault (ρ = 0.395, p < 0.0001): the highest vaults are also the least stable.
Scatter plot showing only a weak relationship between ICL power and vault variability.
Power barely matters. ICL power had only a weak effect (ρ = −0.224); vault is driven by lens size, haptic position, and sulcus anatomy.

Conference presentations

Educational articles

More from this work: the research is applied in ICL Fit (AI ICL sizing) and IOL Reference; explained for patients at ICLSurgery.com/research; and listed in full on the CV.

Common questions

What is VAULT-OCT?

VAULT-OCT is a deep-learning model that predicts a patient's postoperative ICL vault from a single preoperative anterior-segment OCT (AS-OCT) image. It was developed with collaborators at Parkhurst NuVision and published in the Journal of Cataract & Refractive Surgery, and is designed to make EVO ICL sizing more accurate and reduce vault-related exchanges.

Why does ICL vault matter?

Vault is the gap between the ICL and the eye's natural lens. Too little vault can risk cataract; too much can crowd the angle and raise eye pressure. Because vault depends on lens size and anterior-segment anatomy, accurate sizing is the central technical challenge of EVO ICL surgery.

How does AI improve ICL sizing?

Traditional sizing uses white-to-white and anterior-chamber-depth formulas. AI models such as VAULT and VAULT-OCT learn directly from anterior-segment images and real surgical outcomes, capturing sulcus and anatomy the formulas miss; the approach behind the ICL Fit sizing tool.

Can people with a shallow anterior chamber get an EVO ICL?

Often, yes. My research on eyes with anterior chamber depth under 3.0 mm found EVO ICL can be safe and effective for myopia in carefully sized, appropriately selected eyes, though candidacy always requires a full exam.

Citations are provided for transparency and education. A publication is not medical advice.