Abstract
Purpose: :
To utilise 3-D Magnetic Resonance Imaging (MRI) representation of the eye for in vivo measurement of anterior and posterior ocular volume in human ametropia.
Methods: :
Thirty nine young adult participants, 25 myopes, 8 emmetropes (spherical equivalent refraction SER >-0.25 <+0.50D) and 6 hyperopes, group SER mean (± sd) -1.84D± 3.86, were scanned using a Siemens Trio 3-tesla whole-body (head-coil) MRI scanner. The technique acquires a T2-weighted MR image (voxel thickness 1.0x1.0x1.0 mm) which is optimised to reveal the fluid-filled chambers of the eye. Images were analysed using a modified (by KDS) version of freeware software, mri3dX. Voxels were labelled using a 3-D flood-filling algorithm and automatically shaded to delineate fluid-based intraocular structures for each eye. To determine total eye volume, manual editing of ~26 successive axial, sagittal and coronal slices was used to account for non-fluid structures (e.g. the lens) and errors in automatic shading. To differentiate anterior from posterior volume, coronal slices were unshaded progressively from the corneal pole to the posterior pole of the lens. Longitudinal axial length was calculated from a 3-D surface model generated by automatic segmentation, meshing algorithms and an iterative shrink wrap process and compared to partial coherent interferometry (PCI, IOLMaster Zeiss). Editing errors were small; means (± sd) for 10 repeated edits of the same 3-D image were: axial length 24.54 mm (± 0.44); posterior volume 6308.5 mm3 (± 22.28); anterior volume 1135.7 mm3 (± 20.00).
Results: :
The correlation between axial length measured by MRI (y) and PCI was high (RE: y = +0.95x+1.79, R2 0.91, p<0.001; LE: y = +0.88x+3.68, R2 0.79, p<0.001) with respective mean differences (mm) RE: +0.61 (±0.50); LE:+0.84 (±0.72). The correlations between MRI axial length (y) and SER were consistent with previous literature (RE: y = -0.34x+24.50, R2 0.73, p<0.001; LE: y = -0.31x+24.70, R2 0.60, p<0.001). Correlations between posterior volume (y) and SER were a corollary of those for axial length but weaker (RE: y = -218.46x+6490.60, R2 0.53, p<0.001; LE: y = -189.06x+6646.90, R2 0.41, p<0.001). Of special note was that correlations between anterior volume (y) and SER were negligible (RE: y = -0.87x+1056.60, R2 0.001, p=0.86; LE: y = +0.92x+1064.40, R2 0.001, p=0.86).
Conclusions: :
Eye volume anterior to the posterior lens pole is not a significant structural correlate of ametropia. That posterior volume is less potent as a structural correlate than longitudinal axial length may reflect regional variations in posterior eye conformation across refractive groups.
Keywords: myopia • refractive error development • imaging/image analysis: clinical