LensID: A CNN-RNN-Based Framework Towards Lens Irregularity Detection in Cataract Surgery Videos

A critical complication after cataract surgery is the dislocation of the lens\nimplant leading to vision deterioration and eye trauma. In order to reduce the\nrisk of this complication, it is vital to discover the risk factors during the\nsurgery. However, studying the relationship between lens dislocation and its\nsuspicious risk factors using numerous videos is a time-extensive procedure.\nHence, the surgeons demand an automatic approach to enable a larger-scale and,\naccordingly, more reliable study. In this paper, we propose a novel framework\nas the major step towards lens irregularity detection. In particular, we\npropose (I) an end-to-end recurrent neural network to recognize the\nlens-implantation phase and (II) a novel semantic segmentation network to\nsegment the lens and pupil after the implantation phase. The phase recognition\nresults reveal the effectiveness of the proposed surgical phase recognition\napproach. Moreover, the segmentation results confirm the proposed segmentation\nnetwork's effectiveness compared to state-of-the-art rival approaches.\n