Font Size: a A A

Based On The CT Images In Terms In The Four Kinds Of Lung Segmentation Algorithm Performance Evaluation Studies

Posted on:2010-03-13Degree:MasterType:Thesis
Country:ChinaCandidate:P H ShiFull Text:PDF
GTID:2178360302968585Subject:Medical imaging and nuclear medicine
Abstract/Summary:PDF Full Text Request
ObjectiveLung segmentation with four different algorithms to analyze the lung segmentation, segmentation evaluation criteria adopted to explore the effect of the four algorithms.MethodsThrough 166 different lung CT images, prepared by the author of the image processing algorithm to calculate the standard evaluation of segmentation algorithm parameters to explore the relevance of the effect of the four algorithms. These algorithms are: otsu and morphology based on a combination of methods, based on the connected region, based on clustering, based on the growth law. Uniformity of the use of intra-regional and inter-regional contrast of these two indicators to the image of the partition after the quantitative analysis. Finally, the use of these two indicators, the four lung segmentation algorithm for real quantitative analysis.ResultsThe substance of the lung CT image segmentation research that otsu and morphology based on a combination of methods, based on the connected region, based on clustering and growth based on the law.According to the evaluation index of Segmentation algorithm, that is, the contrast between the region and regional analysis of the internal uniformity of comparison, the four algorithms for inter-regional contrast the largest numerical average is based on the combination of otsu and morphological methods, but based on morphology and otsu combined with the stability of the method is the best.On the stability of inter-regional contrast, the stability of the four algorithms from the good to the bad order of them(the smaller of the variance, the better of the stability): Otsu and morphology-based combination (variance 0.0116)
Keywords/Search Tags:otsu, morphology, connectivity regions, clustering, region grow, uniformity of intra-regional, inter-regional contrast
PDF Full Text Request
Related items