Three-Dimensional OCT for Prediction of True Lens Position During Femtosecond Refractive Laser–Assisted Cataract Surgery

Tuesday, April 21, 2015: 10:16 AM
Room 1B (San Diego Convention Center)
Joseph J.K. Ma, MD, FRCSC
Alice Zhu

Purpose
To predict postoperative intraocular true lens position (TLP) from the use of an algorithm utilizing 3 dimensional OCT data (a3D-OCT) obtained intraoperatively during femtosecond ReLACS procedures to potentially improve postoperative refraction.

Methods
A retrospective review of 120 consecutive eyes that had undergone ReLACS was performed to improve refractive outcomes. An algorithm utilizing 3-D biometric OCT data was derived, and this was compared to predictions based on multivariate ocular biometry measured preoperatively and TLP measured postoperatively. TLP was obtained from masked, manually validated Scheimpflug imaging commencing at >1 month postop. Theoretical anterior chamber depth estimates of lens position (t1-ELP and t2-ELP) were calculated based on preoperative biometry and this was compared both TLP and a3D-OCT predictions. Correlation coefficients, t-tests and Bland Altman plots were performed.

Results
Average postoperative follow-up was 87.8 d (SD:32). The average TLP was 4.99mm (SD: 0.30). The average a3D-OCT prediction was 5.02mm (SD 0.32) with a correlation coefficient of r=0.68 (P<0.01). In comparison, the average multivariate estimate of TLP with based on preoperative axial length, pre-op ACD, lens thickness, keratometry and refraction was 5.71 (SD:0.40) with r=0.53 (p<0.01). The average pachymetry, pre-operative ACD, and lens thickness as estimated by algorithm derived OCT data had correlation coefficients of r=0.872, r=0.910 and r=0.942 respectively in comparison to values obtained by optical low coherence reflectometry, with p<0.01 for all metrics.

Conclusion
Algorithm derived a3D-OCT predictions preformed at the time of ReLACS can provide an accurate estimate of TLP and other biometric parameters that is at least as good if not better than theoretically calculated multivariate estimates. Further refinements in this algorithm can potentially result in more accurate postoperative refractive outcomes.