The extraction method of tooth preparation margin line based on S‐Octree CNN

The tooth preparation margin line has a significant impact on the marginal fitness for dental restoration. Among the previous methods, the extraction of margin line mainly relies on manual interaction, which is complicated and inefficient. Therefore, we propose a method to extract the margin line with the convolutional neural network based on sparse octree(S-Octree) structure. Firstly, the dental preparations are rotated to augment the dataset. Secondly, the preparation models are treated as the sparse point cloud with labels through the spatial partition method of the S-Octree. Then, based on the feature line, the dental preparation point cloud is automatically divided into two regions by the convolutional neural network (CNN). Thirdly, in order to obtain the margin line, we adopt some methods such as the dense condition random field (dense CRF), point cloud reconstruction and back projection to the original dental preparation model. Finally, based on the measurement indicators of accuracy, sensitivity and specificity, the average accuracy of the label predicted by the network model can reach 97.43%. The experimental results show that our method can automatically accomplish the extraction of the tooth preparation margin line.

[1]  Pengfei Xu,et al.  Model-Guided 3D Sketching , 2019, IEEE Transactions on Visualization and Computer Graphics.

[2]  Sha Sha,et al.  Robust designs for compound 4 adjoining points circle detection , 2011 .

[3]  Zhou Ming-Quan,et al.  Convolutional Neural Networks in Image Understanding , 2016 .

[4]  Yang Liu,et al.  O-CNN , 2017, ACM Trans. Graph..

[5]  Trevor Darrell,et al.  Caffe: Convolutional Architecture for Fast Feature Embedding , 2014, ACM Multimedia.

[6]  Volker Ahlers,et al.  Combining contour detection algorithms for the automatic extraction of the preparation line from a dental 3D measurement , 2005, SPIE Medical Imaging.

[7]  Zhang Yi,et al.  Big Data Analysis by Infinite Deep Neural Networks , 2016 .

[8]  Sergey Ioffe,et al.  Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift , 2015, ICML.

[9]  Ronald M. Summers,et al.  Segmentation label propagation using deep convolutional neural networks and dense conditional random field , 2016, 2016 IEEE 13th International Symposium on Biomedical Imaging (ISBI).

[10]  Mingyue Ding,et al.  Deep learning based classification of focal liver lesions with contrast-enhanced ultrasound , 2014 .

[11]  Vladlen Koltun,et al.  Efficient Inference in Fully Connected CRFs with Gaussian Edge Potentials , 2011, NIPS.

[12]  Naif Alajlan,et al.  Deep learning approach for active classification of electrocardiogram signals , 2016, Inf. Sci..

[13]  Vladlen Koltun,et al.  Parameter Learning and Convergent Inference for Dense Random Fields , 2013, ICML.

[14]  Cheng Li,et al.  A Radiomics Approach With CNN for Shear-Wave Elastography Breast Tumor Classification , 2018, IEEE Transactions on Biomedical Engineering.

[15]  Wang Wen-ping Guo Si-yuan Li lin Wang Ming-li Mo Li-ping,et al.  Extraction, Separation and Structural Analysis of Water-Soluble Polysaccharides from Chaenomeles cathayensis , 2008 .

[16]  Peng Qun-sheng,et al.  Smooth feature line detection for meshes , 2005 .

[17]  Stephen Lin,et al.  Automatic Feature Learning to Grade Nuclear Cataracts Based on Deep Learning , 2014, IEEE Transactions on Biomedical Engineering.

[18]  Peng Qun-sheng,et al.  Smooth feature line detection for meshes , 2005 .

[19]  T. Vos,et al.  Global, regional, and national incidence and prevalence, and years lived with disability for 328 diseases and injuries in 195 countries, 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016 , 2017 .

[20]  Dinesh Manocha,et al.  Ray-Strips: A Compact Mesh Representation for Interactive Ray Tracing , 2007, 2007 IEEE Symposium on Interactive Ray Tracing.

[21]  Ronald M. Summers,et al.  Deep Learning in Medical Imaging: Overview and Future Promise of an Exciting New Technique , 2016 .