Abstract
Purpose :
We created a new high-resolution (50-MHz) 3D ultrasound biomicroscopy (3D-UBM) imaging system to enable new assessments of the anterior segment structures (e.g., cornea, anterior chamber, iris, trabecular-iris angle, and ciliary body and processes). Here, using 3D-UBM image volumes, we report automated 360-degree assessment of the trabecular-iris angle (TIA), an important biomarker for glaucoma patients.
Methods :
The 3D-UBM system automatically linearly scans a 2D UBM probe across the eye and collects several hundred 2D images to create a high-resolution volume. To assess TIAs, processing steps are: (1) Automatically align images to reduce effects of eye motion. (2) Reduce noise using advanced 3D algorithms. (3) Use interactive visualization software to create reformatted, rotational views about the optic-axis. (4) Use a semantic-segmentation convolutional neural network (CNN) to segment the anterior chamber. (5) Use a modified, automated TIA method to assess the angle for each of 360-degree rotational views. (6) Generate statistical reports and an en face heat map of TIAs.
Results :
CNN segmentations of the anterior chamber gave folded “leave-one-eye-out” excellent Dice score of 0.96±0.01. Most errors arose from ultrasound artifacts at the front of the anterior chamber rather than from the corners where errors would affect the TIA. Automated measurements tended to lie within the spread of 4 readers, giving insignificant difference to readers (p = 0.996). In normal, in vivo eyes, the mean angle was within expected clinical range. Good analysis and measurement repeatability was determined by comparing results from different operators of the software on a single eye and by repeated measurements of a cadaver eye, respectively. For example, 360 measurements at different rotational angles around a single normal eye gave 36.5±5.8 deg and COV of 0.15.
Conclusions :
Using 3D-UBM, one can acquire 3D data sets and measure TIA in a highly automated fashion. Clearly, the method has sufficient precision to assess a condition such as angle closure glaucoma. Potentially, the ability to assess TIA over the entire eye with heat maps (Figure) could inform a more precise assessment of drainage.
This abstract was presented at the 2019 ARVO Imaging in the Eye Conference, held in Vancouver, Canada, April 26-27, 2019.