Research And Implementation Of Pulmonary Fibrosis Auxiliary Diagnosis System Based On Lung Parenchyma Segmentation

Pulmonary fibrosis is one of the four major diseases threating to human health,seriously affecting the patient’s life and work. There is no effective method to treat it, soearly detection and control of fibrosis degree is important. Doctors often find the illnessby lung function testing, but patients are likely to be misdiagnosed, leading to misstreatment time. Therefore developing a system that can detect fibrosis disease is verynecessary. This requires accurate segmentation of lung CT image and there dimensionalreconstruction for accurately fibrosis quantitative analysis, in order to assist doctorsdiagnose and treat disease as soon as possible, which is the key and foundation of futureresearch.The thesis focuses on medical image segmentation and3D reconstructionalgorithm. With the help of medical image processing technology, we built a pulmonaryfibrosis auxiliary diagnosis system on Matlab platform, which is based oncharacteristics of lung image information for quantitative analysis and patient datarecord. The main research contents in this thesis are as follows:①At the respect of image segmentation, the thesis firstly introduced the commonmethods of medical image segmentation: threshold method, region growing method,edge detecting and methods based on certain theory, then analyzed the advantages anddisadvantages of each algorithm by segmenting lung CT image using various algorithmsin Matlab environment. Aimed at the problem of traditional methods which are difficultto satisfy the subject requirements, the thesis proposed an interactive lung parenchymasegmentation algorithm based on improved Snake model. The original image ispre-segmented by threshold algorithm to get roughly boundary, then live-wire model isused to get initial contour, finally Snake model is utilized to extract the lungparenchyma. The thesis did some related improvements on the disadvantages ofLive-wire algorithm and Snake model. Using the evaluation method of imagesegmentation algorithm, we evaluated and analyzed the improved Snake algorithm.②On reconstruction issues, the thesis firstly introduced surface rendering andvolume rendering of3D reconstruction, then analyzed the characteristics of the twomethods, and finally realized3D reconstruction of lung image sequences using volumerendering method in Matlab environment.③On aided detection of pulmonary fibrosis, according to pulmonary parenchyma segmentation results, the thesis quantitatively analyzed pulmonary fibrosis from scalecharacteristics, lung field area and so on. According to three-dimensional model doctorcan view fibrosis lesions from different angles.④The thesis realized the design and development of the system in the Matlabenvironment. It mainly had five modules such as image reading and display, imagesegmentation, three-dimensional reconstruction，quantitative analysis and databaserecord.By researching image segmentation and reconstruction, we constructed a lungfibrosis auxiliary diagnosis system and achieved good results, which can help doctorsquantitatively analyze pulmonary fibrosis and accurately diagnose advanced fibrosis,mild or severe pulmonary fibrosis.