Algorithmic Study Of Lenke Classification Of Idiopathic Scoliosis Based On U-net

Adolescent idiopathic scoliosis(AIS)is a three-dimensional spinal deformity,which is the most common spinal deformity in adolescents.It accounts for about 80% of the total number of idiopathic scoliosis,AIS affects 1-3% of the adolescents around the world,not only deforming their appearance but also compromising their motor function,respiratory function,psychological status and even their life quality.Besides,spinal surgery is time-consuming,risky,and the surgical orthopedic instruments are complicated,difficult,and traumatic as well as complicated(major orthopedic surgery),moreover,the preoperative diagnosis is subjective,leading to different diagnosis.Thus,how to make the diagnosis of spinal surgery intelligent,standardized and automated is the significance of this paper.In clinical practice,the spine is classified according to Lenke classification rules.The error between observers and within observers exist in scoliosis diagnosis.Therefore,doctors urgently need a set of automatic and accurate angle measurement method to replace manual operation.This paper mainly studies the implementation of spine segmentation task and automatic Cobb angle measurement based on spine segmentation image and automatic Lenke classification using SVM under the framework of machine learning.The specific work is as follows:The spine was segmented based on the deep learning model.The U-net segmentation network was established and improved by collecting and expanding the data of lumbar,thoracic and coronal radiographs of the spine.The expanded data was input into the segmentation network for training and prediction,and the segmentation results were obtained.Automatic Cobb angle measurement is realized based on the minimum envelope rectangle.In order to achieve automatic Cobb measurement,it is necessary to identify the boundaries of each vertebra of the spine and locate the end vertebra of the vertebra.The angle was calculated by fitting the straight line.Among them,Cobb angle was the largest Angle,and the corresponding vertebra was the identified end vertebra.Automatic classification algorithm based on SVM.In this paper,according to the result of segmentation,extract the center of each vertebral body from the result,and formed a series of point set,put it into the function fitting toolbox,fitting out the corresponding coefficient of spinal curve and the curve of the spine features as the input of the SVM classifier.Through the experiment,it was found that the curve features of the spine can't make a good results.In this paper,the author made a detailed analysis of the curve features,and by adding the angle features of the bending and lateral radiograph,then the automatic classification task is realized.Comprehensive experiments show that the automatic Lenke classification algorithm based on deep learning and machine learning can achieve rapid classification of the spine and successfully achieve intelligent,automatic and standardized surgical diagnosis.It can provide doctors with a set of rapid and accurate preoperative diagnosis system.