Research On Algorithm For Lung Segmentation From Chest Radiographs

Medical image processing is a branch of computer field, also an important application of digital image processing technology in biomedical engineering. In computer aided detecting system, it is one of the key and the necessary steps to segment the lung from chest Radiographs accurately and validly. Algorithm for lung segmentation from chest Radiographs is studied in this paper, and the main work and research are as follows:(1) Method for segmenting lung from chest radiographs based on the maximum between-cluster variance algorithm is studied. Because the traditional algorithm has many defects in segmenting chest Radiographs, this paper proposes an improved algorithm that is based on mathematical morphology. Experiment results shows that the improved algorithm is effective.(2) The immune and cloning algorithm is used for getting the better thresholds which is used for segmenting the lung from a chest radiographs. Experiment results shows that the improved algorithm is effective.(3) Method for segmenting lung from chest radiographs based on the fuzzy c-means clustering algorithm is studied. Because the traditional algorithm has many defects in segmenting chest Radiographs, a improved algorithm combining the Gaussian function is proposed in this paper. Besides, lots of experiments are made, and experimental results show the improved algorithm is effective.(4) Traditional and gradient vector flow active contour model studied initially are applied in segmenting lung tissue from chest Radiographs. In this paper, Preliminary experiments are made to verify the feasibility of two models, and the defects of two methods are pointed out, which point out the direction of further research work.