Vault Accuracy Using deep Learning Technology

An AI-driven tool to accurately predict postoperative vault of the EVO Visian ICL™ (Implanted Collamer® Lens).

Team

Taj Nasser, MD

Greg Parkhurst, MD

Matt Hirabayashi, MD

Gurpal Virdi, MD

Objectives

To create an accurate, repeatable, and continuously improving machine-learning based tool for the prediction of post-operative ICL™ Vault using various imaging modalities (e.g., Ultrasound Biomicroscopy and Anterior Segment OCT). This approach is novel and image-based, training the model on the unique anatomy of each eye. The models are trained with data from refractive surgery cases in the United States at a high volume surgery center.

Methodology

3,059 images from 437 eyes of 221 consecutive patients who underwent ICL implantation were included. A neural network was trained on pre-operative Very High Frequency (VHF) digital ultrasound images, patient demographics, and postoperative vault. A neural network was chosen and extensively adapted for the purposes of this project.

Results – UBM

12.1 MAE: 66.3 µm

12.6 MAE: 103 µm

13.2 MAE: 66.3 µm

VAULT Prediction Error Stratified by Size and Magnitude of Prediction Error
ICL SizePercent of Predictions within Error Range
 ≤ 250 µm≤ 400 µm≤ 500 µm
12.199.7%99.8%100%
12.697.4%99.0%99.0%
13.292.4%95.1%96.6%
Comparison to Current Literature
StudyErrorLocationSample Size
Rocamora et al.132.0 μm  (MAE)Argentina115 Eyes 59 Patients
Kim et al.104.7 μm (MAE)Korea892 Eyes 471 Patients
Kang et al.106.88/143.69 μm  (MAE)Korea2756 Eyes
Shen et al.159.03 μm (RMSE)China6297 Eyes 3536 Patients
Chen et al.129.89 μm (MAE) China1941 Eyes 1941 Patients
Russo et al.96.94 µm (MAE)Italy561 Eyes 300 Patients
VAULT12.1: 66.3 µm (MAE)
12.6: 103 μm (MAE)
13.2: 91.8 µm (MAE)
USA3059 Images
437 Eyes
221 Patients

Read the JCRS Publication for VAULT: UBM

Next Steps

We are actively training the model on anterior segment OCTs. View the results below, pending publication.

Results – OCT

VAULT Prediction Error Stratified by Size and Magnitude of Prediction Error
ICL SizePercent of Predictions within Error Range
 ≤ 250 µm≤ 400 µm≤ 500 µm
12.199.7%99.8%100%
12.697.4%99.0%99.0%
13.292.4%95.1%96.6%

Vision

An user friendly interface will allow surgeons to upload their images and patient data for the model to process and return a clean output with predicted vault by ICL size. Eventually, the model can be integrated into UBM or OCT machines.

User Interface

(input page)

(results page)

(printout)

Copyright © 2023 | VAULT LLC