Abstract
Purpose :
Visualizing vision-threatening disease (VTD) with 3D optical coherence tomography angiography (OCTA) imaging and using stereolithographic 3D printing to create physical representations of those structures can be useful as both an educational and clinical tool. Additionally, 2D retinal color photos were given depth using convolutional techniques, which may provide meaningful representation and localization of VTD on 3D models.
Methods :
We collected imaging data from 3 subjects with VTD as well as 1 with normal retinal anatomy at a Type 1 Diabetes Mellitus conference. OCTA was captured using an OptoVue RTVue XR Avanti (Fremont, CA), supplemented by a Canon OCT-A1 Xephilio with 3-micron resolution (Tokyo, Japan). Retinal photographs were captured by a Canon 21 MP non-mydriatic retinal camera CR-2 Plus AF (Tokyo, Japan).
Firstly, manufacturer-provided STL (“stereolithography”) images were made printable on a Formlabs Form 2 3D printer (Somerville, MA) using PreForm 3.0.0.
We then wrote software to transform the proprietary OCTA file formats to DICOM part 10 format, while also applying transformations to the samples to improve apparent signal-to-noise ratio and extract layer-by-layer detail. The DICOM files were then converted to STL meshes with 3D Slicer (v4.10.2) before printing in the same fashion as above. Retinal photos were transformed into volumetric representations by convolutional operations on individual color channels, adding artifical depth to the 2D images.
Results :
We successfully created a workflow that allowed the transformation of several existing and emerging imaging standards from volumetric to mesh to physical representations. This enabled our group to manufacture the first scan-accurate physical model of VTD. Printing 1:1 6mm square scans is possible in <2hr, but to represent the full resolution of the scan in our physical models and optimize tactile sensation, we printed most models at 10-20:1 scale (~12-24hr print), giving unprecedented insight into fine features.
Conclusions :
Further making the processing steps transparent to clinicians would allow individuals with visual impairment to understand their conditions from a more complete perspective, allow surgeons to plan and simulate surgical procedures with much greater fidelity, and allow more intuitive investigation of complex anatomic pathologies.
This is a 2020 Imaging in the Eye Conference abstract.