The Computer-aided Detection Method For Facial Morphology In Percision Stomatology

The Precision Medicine defines a new medical concept and therapeutic model,amalgamating biological information and big data science, with the goal of right treatment, right time, giving access and personalized information. As a branch of the precision medicine, the Precision Stomatology aims to break through the traditional clinical experience, improve the method and procedure of detection and therapy, finally achieve the purpose for treating the same disease with different methods and treating the different disease with same methods. The Facial morphological measurement and analysis plays an important role in planning therapeutic treatment and evaluating therapeutic effectiveness.Based on anthropometry and medical statistics, the facial morphological measurement and analysis of the stomatology clinical mainly adopts Direct Measurement, Photogrammetry and X-ray Cephalometric Method. Compared to with other two methods, the Photogrammetry has the highest using frequency. Furthermore,it has advantages of intuitive display, easy operating, fast sampling, cost effectiveness,and harmless measuring, etc. However, the existing methods of data processing and analysis in the Photogrammetry use artificial punctuating and statistical analysis which might not be sufficient to meet the demands of the Precision Stomatology, for its low efficiency, and low accuracy.Firstly, we design an improved ASM-based algorithm to measure human faces,which maps the index point measurement to the Facial Feature Point localization.Referring to the standards in Precision Stomatology clinical detections, we measure and analyze the patient's face image automatically and precisely by using the methods of image processing and data mining. Our method optimize both global and local models in existing ASM algorithm by identifying the area of eye and mouth and locating boundary features.Then, we design an improved DBSCAN-based algorithm to analyze the facial morphology, in which the facial morphology analysis is transformed into the multivariate data analysis. In order to eliminate the effect of different measurement index dimension, we a apply a nondimensionalization approach for data preprocessing and then explore a fine-grained facial morphology classification .Finally, we design and implement a facial morphology aided detection system,which can be used for both face measurement and facial morphology analysis in clinical treatments.