Using advertisements, 15 normal adult volunteers were recruited and screened by an ophthalmologist (JLD) to verify normal ocular motility, vision, anatomy, and stereoacuity. According to a protocol approved by the Institutional Review Board for Protection of Human Subjects at the University of California, Los Angeles, and conforming to the tenets of the Declaration of Helsinki, all subjects gave written informed consent prior to participation in research procedures.
A scanner (1.5-T Signa; General Electric, Milwaukee, WI, USA) was used to perform MRI with techniques described in detail elsewhere, including foam head stabilization, fixation targets to control gaze position, and a dual-phased surface coil array (Medical Advances, Milwaukee, WI, USA) to improve signal-to-noise ratio.
3,5,9,17 Employing 1.5-T magnetic field strength avoids imaging artifacts problematic for the orbit when imaging with stronger fields. Initially, lower resolution scout images, either axial or sagittal, were used to guide subsequent imaging. High-resolution images were obtained using T2 fast spin echo sequences whose details have been previously published.
17 For each orbit and imaging plane employed, MRI acquisition parameters were identical for all gaze positions, compensating for the quantitative effects of any possible small dimensional errors on interpretation of gaze-related changes in EOMs. For analysis, quasi-coronal images perpendicular to the long axis of the orbit were obtained at 2-mm intervals using an 8-cm field of view (256 × 256 matrix, 312-μm pixel resolution). There were two special modifications to the published MRI protocol: imaging was only analyzed in maximal supraduction, infraduction, abduction, or adduction by the fixating eye; and a larger extent of each orbit, 19 to 24 images planes (38–48 mm), was selected to completely image the entire anteroposterior extent of each analyzed EOM.
We used ImageJ (
http://imagej.nih.gov/ij/; provided in the public domain by the National Institutes of Health, Bethesda, MD, USA) and a computing environment (MATLAB; MathWorks, Inc., Natick, MA, USA) to quantify the digital images. The angles of ocular duction were calculated in degrees by the movement of the globe/optic nerve junction using techniques described in detail elsewhere.
3,5,9,16
Quantitative analysis of MRI was performed only in images free of significant motion artifacts that might blur the margins of EOM contours. Two different techniques were employed to calculate the total EOM volume for each gaze position. The first technique directly measured the EOM volume for all but the terminal portions of the EOM belly. Bellies were manually outlined and cropped in each image slice as far posteriorly and anteriorly as the EOM could be distinguished from surrounding orbital connective tissue and the globe, then saved individually as tagged image file format files. Custom software (MathWorks, Inc.) enumerated the pixels in each EOM cross-section, converted these into calibrated areas (mm2), summed them in all image planes that included an identifiable EOM cross section, and finally converted the summed areas into volumes (mm3) by multiplying by the 2-mm slice thickness.
The second technique added an estimate of the unmeasured terminal EOM volumes posteriorly, near the orbital apex, and anteriorly, near the EOM insertion, where it was impossible to reliably distinguish the EOM cross sections from surrounding tissues (
Fig. 1). Using previously described methods, the position of each measured EOM cross section was calculated relative to globe center using a three-dimensional rotation matrix and extraorbital anatomic landmarks.
16 Then, the anteroposterior distance from globe center to orbital apex was directly measured using ImageJ from either the axial or sagittal scout images. The difference between the measured distance to the orbital apex in longitudinal scans and the anteroposterior location of the most posterior measured EOM cross section represented the length of unmeasured posterior EOM belly. This unmeasured length was divided by 2 mm to estimate the number of additional MRI slices required to completely encompass the apical portion of the EOM. Finally, to estimate the volume contained within this apical segment, the most posterior measured EOM cross-sectional area was carried posteriorly as the cross-sectional area of each of the unmeasured image planes. For example, if the above analysis revealed that a 4-mm length of the posterior EOM belly was not directly measured, the volume of the unmeasured 4-mm posterior segment was estimated by multiplying the most posterior measured cross-sectional area by the 2-mm slice thickness, then by two again to account for the two additional MRI slices required to completely encompass the terminal 4-mm segment.
A similar technique was used to estimate the anteroposterior length of the unmeasured anterior segment of the EOM. To calculate the anteroposterior distance from globe center to EOM insertion, we assumed a corneal diameter of 12 mm and standard locations for the superior rectus (SR) insertion 7.0 mm posterior to the limbus, inferior rectus (IR) insertion 6.6 mm posterior to the limbus, medial rectus (MR) insertion 5.5 mm posterior to the limbus, and lateral rectus (LR) insertion 6.9 mm posterior to the limbus.
19,20 Using measured globe diameters and angles of duction (
Fig. 2), the cord distance between the EOM insertion and most anterior measured EOM cross section was calculated, then divided into 2-mm segments to estimate the number of additional MRI slices required to completely encompass the anterior portion of the EOM belly (
Fig. 1). Then, to estimate the volume contained within this anterior EOM segment, the most anterior measured EOM cross-sectional area was carried forward as the cross-sectional area for the unmeasured image planes. For example, if the above analysis revealed that a 6-mm length of anterior EOM was not directly measured, the volume of the unmeasured anterior portion was estimated by multiplying the most anterior measured cross-sectional area by the 2-mm slice thickness, then again by three to account for the three additional MRI slices required to completely encompass the anterior extent of the EOM.
Volume measurements were then corrected for the obliquity of cross sections relative to the long axis of the EOM (
Fig. 3).
21 Such obliquity results from the generally oblique trajectory of each EOM relative to the long axis of the orbit, as well as from EOM path curvature. The change in position of the EOM area centroid (equivalent to the center of mass) was plotted throughout the anteroposterior extent of the orbit. Changes in the horizontal and vertical coordinates of the bracketing area centroids in consecutive planes were used to compute the horizontal and vertical angular offsets from true orthogonal for each image plane. Measured EOM cross-sectional areas were then multiplied by the cosines of these angles to yield a final total volume corrected for EOM path relative to the image planes.
Total EOM volume was then calculated as the sum of the corrected measured EOM volume and the estimated unmeasured anterior and posterior terminal EOM volumes. Statistical comparisons were made using paired t-tests for both the measured and total EOM volumes between maximal relaxation (duction away from the EOM) and maximal contraction (duction toward the EOM), with a 0.05 level chosen for statistical significance.