Related Experiment Video
Updated: Sep 10, 2025
Multimodal Volumetric Retinal Imaging by Oblique Scanning Laser Ophthalmoscopy oSLO and Optical Coherence Tomography OCT
Published on: August 4, 2018
Validation of a new implantable collamer lens sizing algorithm based on SS-OCT images
Pierre Zéboulon1, Nicole Mechleb, Maria Rizk
1From the Department of Anterior and Refractive Surgery, Rothschild Foundation Hospital, Paris, France (Zéboulon, Mechleb, Rizk, Flamant, Gatinel, Saad); Institute of Ophthalmology, Halle, Germany (Duncker); Dr. Neuhann's Eye Clinic, Munich, Germany.
A new deep learning model accurately sizes Implantable Collamer Lenses (ICL) using OCT images, outperforming traditional methods and improving patient outcomes. This AI tool aids surgeons in making better ICL sizing decisions.
Area of Science:
- Ophthalmology
- Medical Imaging
- Artificial Intelligence
Background:
- Accurate sizing of Implantable Collamer Lenses (ICL) is crucial for optimal refractive outcomes and patient satisfaction.
- Current ICL sizing methods may have limitations, potentially leading to suboptimal vault and refractive results.
Purpose of the Study:
- To evaluate a novel deep learning (DL) model for predicting ICL size using raw Swept-Source Optical Coherence Tomography (OCT) images.
- To compare the performance of the DL model against the standard STAAR nomogram for ICL sizing.
Main Methods:
- A retrospective external validation study was conducted across two European clinics.
- Preoperative OCT images from 848 eyes (429 patients) implanted with EVO ICL V4 were analyzed.
- The DL model predicted ICL vault, confidence levels, and probability of achieving a 250-750 μm vault (P250-750), compared against actual postoperative vaults.
Main Results:
- The DL model demonstrated a lower Mean Absolute Error (MAE) (146 ± 113 μm) compared to the STAAR nomogram (186 ± 149 μm; p < 0.001).
- The model achieved high accuracy, predicting vaults within ±250 μm (81.7%) and ±300 μm (90.7%).
- For cases outside the desired vault range, the model recommended more appropriate sizes in 70.6% of instances, and aligned with final sizes in 81.8% of lens exchange cases.
Conclusions:
- The DL-based ICL sizing model accurately predicts postoperative vault using raw OCT images.
- The model provides valuable metrics like P250-750, aiding clinical decision-making and potentially reducing sizing errors.
- This AI-driven approach shows promise for improving patient outcomes in ICL surgery.
