Detection And Analysis Algorithm Of Lung Nodules In CT Image Based On Gestalt Vision

With the rapid development of medical technology,doctors hope that they can find focus more accurately by all sorts of means,so as to give the correct diagnosis.One of the most commonly used and intuitive way is to detect and analyze the focal point through CT image.Because medical image sequence generally contains large amounts of data,and possess the characteristic of professionalism,doctors need to test the focal point frame by frame in order to find the focal point.But humans would inevitably feel tired.Additionally,different physician has the different experience.So,the detection effect is directly impacted.In recent years,computer and optical imaging technology have been developing rapidly,we can get images of higher resolution and definition,which make it possible for physicians to detect focal point with the aid of computer,and computer aided detection gradually become a hot research topic.The incidence of lung cancer is high in China,the early detection of lung nodules and early treatment is of great significance to save lives.But medical image sequence is comprehensive data with characteristics of graphics,anatomy,pathology and medical,so it is difficult to apply the common algorithms directly to medical image processing.On the basis of thorough study of gestalt principles of visual,the diagnostic procedure was simulated by computer.The lung nodule detection,analysis process and algorithm were all put forward in this article.For medical image sequence contains large amount of data,the average density projection of focal point segmentation algorithm was applied in the procedure of optimum preprocessing.Make full use of the characteristics that pulmonary nodules are near balls in space,the information was integrated into the local three-dimensional images.The lung nodules and other interferences are all in a complicated environment,so it is difficult to accurately register the suspected area in medical image,then multiple positions of CT image were used to collaboratively detect lung nodules,retaining all suspected area in order to guarantee lower miss rate as much as possible,and the maximum density projection of focal point segmentation algorithm was proposed.Because it is difficult for computer to accurately distinguish a sign of focal point for its features are quite different with each other's and are quite similar with interferences.So,vascular enhancement model was constructed,and a new algorithm was put forward from the aspect of removing blood vessel.The border of focal point was unwrapped,then signs can be distinguished by comparing recursive figures.According to the experimental analysis,in lung nodules detection,accuracy was above 90%.In signs discrimination,accuracy was above 80%,both were higher than current mainstream algorithm.And the algorithm has been applied to clinical,satisfy the standard of clinical.