Abstract
Purpose :
Biomechanical simulation facilitates the quantitative study of factors contributing to strabismus and the effectiveness of surgical treatment. However, creating three-dimensional (3D) eye models for simulations that include patient-specific characteristics of the EOMs can be laborious, subjective, and hard to reproduce. We developed an automated pipeline to generate patient-specific 3D eye models.
Methods :
A 3D eye model creator was implemented as a plugin in Autodesk Maya, a 3D animation and modeling application. The input to the pipeline was outlines of the extraocular muscles, globes, and orbital walls traced in the quasi-coronal (perpendicular to the orbital axis) and quasi-sagittal (parallel to the orbital axis) stacks of magnetic resonance images (MRI) from 7 strabismic patients. Each MRI stack of one eye contained ~20 images with 0.3125 mm pixel resolution and 2 mm slice thickness. When sagittal MRI was unavailable, coronal MRI was resliced in ImageJ to indirectly generate sagittal images, in which the inferior oblique (IO) muscle and superior oblique (SO) tendon anterior to the trochlea were traced. 3D EOM geometric paths were created by computing the centroids of the muscle traces and optimally fitting a B-spline curve to those centroids. Interpolation was performed at locations where the EOMs were not tracible. The globe center and size were automatically determined by tracing the globe.
Results :
The IO muscle path was realistically modeled from resliced sagittal MRI. Seven 3D patient-specific eye models were automatically and efficiently generated from MRI traces within 5 minutes each. The (SO) EOM and tendon were successfully modeled, from the tendon traces, and wrapped around the globe. The four rectus EOMs demonstrated visually realistic paths.
Conclusions :
An automated pipeline has been implemented to realistically model the geometry of the EOMs from orbital MRI even when available only in the quasi-coronal plane. Such an objective tool significantly reduces the time and effort in building 3D patient-specific eye models for biomechanical simulation.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.