Abstract
Purpose :
Partial coherence interferometry (IOLMaster) is the gold standard in clinical ophthalmology for measuring axial length and thus is critical in evaluating myopic progression. One alternative for axial length measurement is through MRI. We previously reported consistently longer axial length measurements on IOLMaster versus manual MRI measurements in highly myopic eyes with staphyloma. To better characterize this discrepancy, we compare axial lengths in emmetropic eyes measured by IOLMaster with those measured by a novel 3D MRI algorithm.
Methods :
A prospective study was performed on 22 eyes of 11 young (ages 22 to 28 years) emmetropic volunteers with no known history of ocular disease. All eyes underwent measurement with both IOLMaster and 3D MRI scan (Phillips, 3.0T, fat-suppressed T2-weighted cube, a modified 3D fast-spin echo sequence). Volume renderings of 3D MRI images were generated. To allow for consistent 3D MRI axial length determination, a novel algorithm was developed for a feature-based registration, which included a series of linear transformations, thresholding, voxel normalization, re-sampling and alignment. To compare axial length measurements in the two measurement modalities, we employed linear regression analysis to determine the slope of the MRI:IOLMaster relationship.
Results :
Eyes examined ranged in axial length from 21.69 mm to 24.60 mm on IOLMaster and from 21.55 mm to 23.95 mm on MRI. After performing the linear regression, the MRI:IOLMaster relationship slope was determined to be 0.90 (R2= 0.95, p < 0.0001). An R2 value of 0.95 implies that the axial length measurement by MRI and by IOLMaster are highly correlated. However, since a slope of 1.0 would indicate identical accuracy in measurement, the slope of 0.90 implies that there is a measurement bias that is a result of either an over-estimation by IOLMaster or under-estimation by 3D MRI.
Conclusions :
Axial length measurements differ between 3D MRI and IOLMaster, even in a subset of young, emmetropic eyes free from ocular disease. As MRI is not affected by media issues, it may provide a more veridical measure of length, implying the bias observed may be originate from IOLMaster over-estimating length.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.