ESCRS - FP18.10 - Investigation Of Short-Term Crystalline Lens Positional Shifts Following Icl Implantation Using Ss-Oct Integrated With Multi-Task Deep Learning

Investigation Of Short-Term Crystalline Lens Positional Shifts Following Icl Implantation Using Ss-Oct Integrated With Multi-Task Deep Learning

Published 2025 - 43rd Congress of the ESCRS

Reference: FP18.10 | Type: Free paper | DOI: 10.82333/0e5w-8v66

Authors: Xun lun Sheng* 1 , Meijiao Ma 1

1Refractive Department,Gansu Aier Ophthalmiology and Optometry Hospital,Lanzhou,China

Purpose

To investigate the tilt and decentration of the crystalline lens shortly after implantable collamer lens (ICL) implantation.

Setting

A retrospective observational study in Shanxi Eye Hospital Affiliated to Shanxi Medical University

Methods

All enrolled patients underwent anterior segment swept-source optical coherence tomography (SS-OCT) both before and within one week after ICL surgery.​ A multitask network automatically conducted SS-OCT image segmentation and provided quantitative assessments of the crystalline lens's tilt and decentration. For data following a normal distribution, paired t-tests were performed, while for data not meeting normality assumptions, the Wilcoxon signed-rank test was utilized.

Results

The final data analysis included 42 eyes. No statistically significant difference was observed in the degree of crystalline lens tilt pre- and post- ICL implantation (P =0.1793), with both tilting temporally and demonstrating a significant positive correlation (r = 0.785, P <0.001). The degree of post-surgical crystalline lens decentration was significantly reduced compared to pre-implantation (P =0.0073), with the decentration predominantly in the upward direction, and demonstrating a strong positive correlation (r = 0.637, P<0.001). Additionally, ICL tilt exhibited a strong positive correlation with both pre- and post-surgery lens tilt (r = 0.687, P<0.001; r = 0.613, P<0.001), as well as with ICL decentration (r = 0.512, P=0.001).

Conclusions

SS-OCT combined with multi-task deep learning enables quantitative assessment of the spatial position of both the crystalline lens and ICL.​ Short-term observations indicate that ICL implantation impacts the position of the crystalline lens to varying extents.