Ribs segmentation based on image fusion and wavelet de-noising

With the development of image processing technology, computer-aided diagnosis technology has been widely used in digital chest radiographs lesions detection. As the special nature of medical image, the complexity of image information, and especially the interference of lung vessels and bone cause great difficulties in the texture analysis of lung parenchyma. A new ribs segmentation method is proposed based on the characteristics of medical image. Firstly, the complete region of lung parenchyma was extracted by Otsu after implementation of Gaussian mask and wavelet image fusion algorithm. Secondly, multi-scale wavelet was applied to reduce the noise of the lung fields. Thirdly, ribs were segmented by Gaussian mask. Finally, the edge of the ribs was smoothed. The experiment results show that the algorithm can guarantee to maintain the information of the lungs and ribs. © 2012 IEEE.