Improved Watershed Algorithm For Bone Removal From CTA

Image segmentation is the key step from image processing to analysis; it is the basic of image analysis and understanding, therefore the correctness of segmentation results have a great impact on the follow-up operations. Using the image processing techniques in medical field can help doctors to diagnose the disease better. Because of the complexity of medical images, the development and application of the related technology of other medical image processing were restricted by the medical image segmentation techniques.The CTA (Computed Tomography Angiography, CTA) bone removal method was studied in this paper which was based on mathematical morphology. According to the character of CTA images, watershed algorithm was chosen for the CTA bone removal through the depth study of image segmentation method. In order to avoid the over segmentation, a new flooding method was used in the improved watershed algorithm. The improved watershed algorithm can not only accelerate the execution speed and the segmentation results can also be intervened by user. Over segmentation was avoided effectively.An improved watershed algorithm was applied to the CTA bone removal in this paper, including the following facets:(1) According to the character of CTA images, watershed algorithm was chosen for the CTA bone removal through the depth study of image segmentation method and comparative analysis.(2) The theory and methods of mathematical morphology and watershed algorithm were firstly discussed in this paper. Based on the application analysis of the watershed algorithm, it was improved to avoid the over segmentation. For the non-minimum plateaus problem of the watershed algorithm, the marking process was described and the results were explained in the improved algorithm. The effectiveness of the algorithm was inspected by comparing the improved algorithm with several methods of CTA bone removal finally.(3) Finally in this paper, the new algorithm was used to do the bone removal of abdomen CTA, and the results were shown by 3D imaging technology. So the improved watershed algorithm can be inspected again during the application of CTA bone removal. The experiment indicated that the improved watershed algorithm built upon a fast immersion-based watershed algorithm followed by a hierarchical organization of the resulting basins in a tree structure. Each element was processed exactly once with respect to its neighborhood in the specified order. After computing the watershed transform, a merging threshold parameter was given to control the tree partitioning and basin merging in real time to trace the image segmentation result. Therefore, the results of the improved algorithm will have a certain reference value in the medical diagnosis.