ICL Sizing Calculators and Formulas Compared: OCOS, Reinstein, KS, NK, and AI

Every ICL sizing calculator is trying to answer one question: which of the four available EVO ICL lengths will land this eye's vault in a safe range? Over the years, surgeons have built a series of ICL sizing formulas to do it — each adding inputs beyond simple corneal diameter. Here is how the main ones compare, and where AI ICL sizing fits.

The traditional formulas

• OCOS (STAAR) nomogram — the manufacturer's calculator, based primarily on white-to-white (WTW) and anterior chamber depth (ACD). Simple and universal, but a coarse predictor of vault.
• Reinstein formula — uses very-high-frequency digital ultrasound to measure the sulcus directly, aiming to improve on WTW-based sizing.
• KS formula (Kamiya–Shoji) and NK formula (Nakamura) — regression formulas derived from large series, incorporating sulcus-to-sulcus (STS) and angle-to-angle (ATA) measurements.
• Kane ICL formula and calculators such as ICL Guru — newer tools blending multiple biometric inputs to refine size selection.

They share a common ceiling: each is a single formula mapping a handful of measurements to a size. Because the ciliary sulcus is only partly predicted by any one input — WTW, STS, or ATA — a formula that fits the average eye can still miss the eye in front of you.

Where AI ICL sizing is different

ICL Fit is not another regression formula — it is an AI model that learns the multi-parameter patterns driving ICL vault prediction from a large, real-world dataset of EVO ICL outcomes, using a non-contact Pentacam / AS-OCT scan. Instead of forcing one equation onto every eye, it weighs many anatomic features together to recommend the best-fit lens size. The approach is validated in the peer-reviewed VAULT and VAULT-OCT studies (Journal of Cataract & Refractive Surgery), which showed image-based machine learning predicts vault more accurately than conventional white-to-white nomograms.

Which should you use?

The formulas remain useful reference points, and many surgeons cross-check several. But the direction of travel is clear: away from single WTW nomograms and contact ultrasound, toward non-contact, image-based, outcome-trained prediction — the case made in AI ICL sizing without UBM. And remember that the target itself is not one number: see ICL vault explained for why vault is a range, not a fixed micron value.

For a plain-language overview of sizing methods for patients, see ICLSurgery.com; for surgical technique, the free ICL Workshop.