AI ICL Sizing Without UBM: How ICL Fit Predicts EVO ICL Vault from a Scan

In EVO ICL (Implantable Collamer Lens) surgery, sizing is the variable that decides the outcome. The right lens length lands the ICL vault in a safe, comfortable range; the wrong one crowds the angle or sits too low against the crystalline lens, driving exchanges. ICL Fit is an AI ICL sizing agent built to solve exactly this problem — predicting postoperative vault and the best-fit lens size for an individual eye from a single non-contact scan, with no UBM (ultrasound biomicroscopy) required.

Why ICL sizing is the bottleneck

Conventional sizing leans on white-to-white (WTW) nomograms — a single linear formula that maps corneal diameter to a lens size. But WTW is only a rough proxy for the sulcus anatomy that actually governs vault, so the same measurement can produce very different vaults in different eyes. Surgeons chasing more precision have historically added UBM to image the ciliary sulcus, accepting a contact ultrasound exam, extra chair time, and operator-dependent measurements — often without a proportional gain in accuracy.

What ICL Fit does

ICL Fit turns anterior-segment imaging into a vault prediction a surgeon can act on before surgery. Instead of one formula, it learns the multi-parameter patterns that drive vault from a large, real-world dataset of EVO ICL surgical outcomes, and returns a best-fit size for the eye in front of you. It is grounded in the founding team's peer-reviewed, PubMed-indexed research in the Journal of Cataract & Refractive Surgery and Clinical Ophthalmology:

• VAULT (2024, JCRS) — a novel image-based AI model showing machine learning predicts ICL postoperative vault more accurately than conventional white-to-white nomograms.
• VAULT-OCT (2025, JCRS) — extends the approach to anterior-segment OCT, using deep learning to predict vault directly from the scan.
• Dynamic Changes in ICL Vault (2026) — demonstrates vault is not a fixed number: it shifts as the pupil dilates and constricts under different lighting.

OCT and Pentacam, not UBM

The imaging shift is the whole point. Anterior-segment OCT and Pentacam-based tomography are non-contact: a quick, comfortable, reproducible scan with no probe on the anesthetized eye. That removes the operator variability of sulcus-to-sulcus ultrasound and, paired with AI, predicts the outcome that matters — vault — rather than a surrogate dimension. Better accuracy and a better patient experience come from the same change in technology, not a trade-off between them.

Best-fit-per-eye, not a target vault number

A persistent misconception is that ICL sizing means hitting one "ideal" vault in microns. It doesn't. Vault varies between eyes and even changes with lighting within the same eye, as the team's dynamic-vault study showed. The goal is the best-fit lens size for the individual eye — the size most likely to stay in a safe range across real-world conditions. That is precisely the question an outcome-trained AI model is built to answer, which is why image-based AI ICL sizing is displacing both UBM and one-size-fits-all nomograms.

A complete EVO ICL pathway

Accurate sizing pairs with confident technique. The free ICL Workshop — an 18-video series from the Parkhurst NuVision team — covers ICL power selection, loading and insertion, toric ICL rotation, preventing IOP spikes, and challenging cases. It sits within Refractive Foundations, the educational initiative of the World Council of Refractive Surgery Fellowships, and is explained for patients at ICLSurgery.com. Together with ICL Fit for sizing, they form an evidence-based route to a confident, high-volume EVO ICL practice — no UBM required.

Keep reading: ICL sizing calculators and formulas compared (OCOS, Reinstein, KS, NK, and AI) and ICL vault explained — what counts as a good vault, and why it changes with lighting.