Mask-MCNet: Tooth instance segmentation in 3D point clouds of intra-oral scans

Computational dentistry uses computerized methods and mathematical models for dental image analysis. One of the fundamental problems in computational is accurate tooth instance segmentation high-resolution mesh data intra-oral scans (IOS). This paper presents a new model based on deep neural networks, called Mask-MCNet, end-to-end learning 3D point cloud IOS. The proposed Mask-MCNet localizes each by predicting its bounding box simultaneously segments points that belong to individual instance. processes input raw original spatial resolution without employing voxelization or down-sampling technique. Such characteristic preserves finely detailed context like fine curvatures border between adjacent teeth leads highly as required clinical practice (e.g. orthodontic planning). experiments show outperforms state-of-the-art achieving 98% Intersection over Union (IoU) score which very close human expert performance.