Multiview Volume and Temporal Difference Network for Angle-Closure Glaucoma Screening from AS-OCT Videos

Figure 2

The pipeline of our MT-net architecture. The input AS-OCT videos are aligned to reduce the video jitter. Then, the ACA extraction and image difference are carried on for the two subnetworks: (a) multiview volume network and (b) temporal difference network. Finally, a soft voting-based ensemble model is adopted to incorporate the two subnetworks to output the final classification results.