ICL Sizing Research & Publications — The Science Behind ICL Fit

Peer-reviewed publications

Traditional ICL sizing relies on white-to-white (WTW) measurement and single-formula nomograms — the OCOS (STAAR) nomogram, the Reinstein formula, the KS formula (Kamiya–Shoji), the NK formula (Nakamura), and the Parkhurst nomogram — which only approximate the sulcus geometry that actually governs vault. The peer-reviewed VAULT and VAULT-OCT studies (Journal of Cataract & Refractive Surgery) introduced deep-learning models that learn from preoperative anterior-segment imaging (Pentacam / AS-OCT) paired with measured post-operative vault, reducing vault-related ICL exchanges and improving predictability. Crucially, vault is not a fixed target number — it varies by eye and shifts with lighting — so the aim is the best-fit lens size for each eye, not a single micron value. ICL Fit is the AI-based ICL sizing agent that brings this research to Pentacam-based sizing for everyday clinical use.

Hirabayashi M, Cantu E, Abramson A, Virdi G, Nasser T, Parkhurst G. Dynamic Changes in Implantable Collamer Lens Vault and Anterior Chamber Angle Under Varying Lighting Conditions. Clinical Ophthalmology. April 2026. DOI · PubMed
Hirabayashi M, Virdi G, Nasser T, Abramson A, Parkhurst G. VAULT-OCT: Vault Accuracy Using Deep Learning Technology for Predicting ICL Postoperative Vault with Anterior Segment Optical Coherence Tomography. Journal of Cataract & Refractive Surgery. July 2025. DOI · PubMed
Hirabayashi M, Virdi G, Nasser T, Abramson A, Parkhurst G. Accuracy of Reported Sizes of the EVO/EVO+ Visian Implantable Collamer Lenses. Cureus. August 2025. DOI · PubMed
Hirabayashi M, Virdi G, Nasser T, Libfraind L, Zapata E, Abramson A, Parkhurst G. ICL Exchanges or Explants Due to Sizing in a United States High Volume Center. Clinical Ophthalmology. July 2025. DOI · PubMed
Hirabayashi M, Virdi G, Abramson A, Nasser T, Vanrachack A, Parkhurst G. Clinical Outcomes of Implantable Collamer Lenses for the Treatment of Myopia in Eyes with Anterior Chamber Depth (<3.0 mm) at a Single Center in the United States. Clinical Ophthalmology. 2025. DOI · PubMed
Nasser T, Hirabayashi M, Virdi G, Abramson A, Parkhurst G. VAULT: Vault Accuracy Using Deep Learning Technology — A Novel Image-Based AI Model for Predicting Implantable Collamer Lens Postoperative Vault. Journal of Cataract & Refractive Surgery. April 2024. DOI · PubMed
Hirabayashi M, Cooley A, Shakarchi F, Vanrachack A, Ducker L, Shelby C, Coleman W, Parkhurst G, LoBue S. Phakic Intraocular Lenses as a Potential Treatment for Pigment Dispersion Syndrome. Cureus. February 2026. DOI · PubMed

Featured study: vault is a moving target, not a number

In Dynamic Changes in ICL Vault and Anterior Chamber Angle Under Varying Lighting Conditions (Clinical Ophthalmology, 2026), the team imaged 100 eyes after EVO / EVO Toric ICL implantation under scotopic, mesopic, and photopic lighting on the MS-39. The data show that vault is a dynamic, physiologic range — not a fixed value to chase — which is the principle behind how ICL Fit sizes for best fit rather than one ideal micron number. Below are the study's own figures, explained.

Box-and-whisker plot of ICL vault (µm) across scotopic, mesopic, and photopic lighting, showing vault falling as lighting brightens. — **Figure 1 — ICL vault by lighting condition (100 eyes).** Mean vault fell from 495.6 µm in the dark (scotopic) to 486.7 µm (mesopic) to just 393.8 µm in bright light (photopic) — a roughly 100 µm swing in the same eye, all highly significant (p < 0.0001). The single number your surgeon records depends on the room lighting when the scan was taken.

Box plot of anterior chamber angle (degrees) across scotopic, mesopic, and photopic lighting, showing the angle widening under brighter light. — **Figure 2 — Anterior chamber angle by lighting condition.** As the pupil constricts in bright light the anterior chamber angle widens — from 27.2° (scotopic) to 29.7° (photopic), p < 0.0001. Vault and angle move together with pupil size, which is why a lens sized to one lighting snapshot behaves differently across a real day.

Scatter plot of vault variability (range across lighting) versus mean vault, with an upward regression line. — **Figure 3 — Higher-vault eyes vary more.** Vault variability rose with mean vault (ρ = 0.395, p < 0.0001): the higher an eye's average vault, the more it swings between dark and bright conditions. Lower-vault eyes stay comparatively stable. This is direct evidence that a high 'target' number is also the least stable one.

Scatter plot of vault variability versus ICL dioptric power, with a shallow downward regression line. — **Figure 4 — Lens power barely matters.** ICL power had only a weak negative relationship with variability (ρ = −0.224, p = 0.025). This reinforces the core sizing principle: vault is driven by lens length, haptic position, and sulcus anatomy — not by dioptric power.

Together the figures make the clinical case: because vault shifts ~100 µm with lighting and the highest vaults are the least stable, an ICL is best judged by fit for the individual eye across conditions rather than a single target number. Read the full study on PubMed.

Frequently asked questions

What is the research behind ICL Fit?

ICL Fit is grounded in peer-reviewed research on ICL sizing and vault prediction — most notably the VAULT and VAULT-OCT studies published in the Journal of Cataract & Refractive Surgery, which introduced deep-learning models that predict post-operative ICL vault from anterior-segment imaging. The same team has published on ICL exchange rates, shallow anterior chambers, and how vault changes with lighting.

Is AI ICL sizing peer-reviewed?

Yes. The VAULT (2024) and VAULT-OCT (2025) papers in the Journal of Cataract & Refractive Surgery are peer-reviewed, PubMed-indexed studies showing that image-based machine learning can predict ICL vault more accurately than conventional white-to-white nomograms.

Who developed the VAULT and VAULT-OCT models?

They were developed by Matt Hirabayashi, MD, Gurpal Virdi, MD, Taj Nasser, MD, and Gregory Parkhurst, MD — refractive surgeons and researchers at Parkhurst NuVision, one of the highest-volume EVO ICL teams in the United States. Dr. Parkhurst was the highest-volume implanter in the pivotal US FDA EVO ICL clinical trial.

Is vault a fixed target number?

No. Vault is not a single target value — it varies from eye to eye and changes with lighting as the pupil dilates and constricts (shown in the team's Dynamic Changes in ICL Vault study). The goal of sizing is the best-fit lens size for an individual eye, not one 'ideal' micron number.

Selected presentations

Virdi G. ICL Sizing Using Multiple AI-Based Modalities. Presented at American-European Congress of Ophthalmic Surgery (AECOS) Summer Symposium. 2026.
Hirabayashi M, Virdi G, Nasser T, Abramson A, Parkhurst G. VAULT-OCT: Vault Accuracy Using Deep Learning Technology for Predicting ICL Postoperative Vault with AS-OCT. Presented at American Society of Cataract and Refractive Surgery (ASCRS). 2024.
Virdi G, Hirabayashi M, Nasser T, Parkhurst G. The Future of AI ICL Sizing is OCT, Not UBM. Presented at STAAR Surgeon Summit. 2024.
Virdi G, Hirabayashi M, Nasser T, Parkhurst G. VAULT-OCT: Deep Learning to Predict ICL Vault from OCT. Presented at YMDC Live. 2025.

The team

ICL Fit is developed by a team including Matt Hirabayashi, MD (co-founder of ICL Fit), Gurpal Virdi, MD, Taj Nasser, MD, and Gregory Parkhurst, MD — refractive surgeons and researchers who developed the VAULT and VAULT-OCT deep-learning vault-prediction models, published in the Journal of Cataract & Refractive Surgery, Clinical Ophthalmology, and Cureus. Several are also co-founders of Refractive Foundations, whose free EVO ICL Workshop covers sizing, power selection, and off-label use.

Learn more

Read the ICL sizing FAQ, the ICL Education overview, the ICL Fit blog, or the plain-language guide for patients.