Design And Implementation Of Algorithms For Lung Contour Distortion Detection

Automatic location of lesions in medical images is a key technique of computer aided therapy.In the current approach,doctors are required to diagnose medical patterns in an artificial way,without a doubt,which requires a huge labor cost.In the existing medical database,there is a large amount of data.Computer aided diagnose can reduce the workload and improve the efficiency of diagnosis,this topic has been an important one for discussion.Many image segmentation algorithm and matching model were proposed,such as based on image gray algorithm,based on image gradient algorithm and based on template match-ing model.However,thanks to the complexity of medical images,the whole process still requires doctors to complete image segmentation and positioning interactively,which can not achieve large-scale medical image processing.In this paper,a set of automatic diagnostic algorithms for the contour distortion diag-nosis of lung CT image sequences is established.The algorithm consists of contour mask extraction algorithm and contour distortion estimation algorithm.In fact,there always a lot of diseases in the structure of the diseased lungs,and the contour distortion is only a part of the disease.However,no matter what kind of disease is needed to be diagnosed,it is necessary to obtain the mask of the lung contour firstly.That is to say,the contour mask extraction algorithm provides a precondition for all the other algorithms.Based on the input of the former algorithm,the contour distortion algorithm is used to establish the feature vector of the contour.In the training procedure,the contour information is marked by interaction.The final diagnosis model is obtained by machine learning,which is used to Highlighted the diseased part of contours.