|
Currently, the medical image segmentation in medical image processing plays a very crucial role , its purpose is to segment the image area of special significance , and extract the main features of the data , so as medical image processing and analysis provide a strong basis for is the basis for further diagnosis . However, since the prevalence of medical image noise, and low resolution . Therefore, it is the field of medical image processing is a classic problem. Breast cancer is a malignant tumor of serious harm to women's health , one of the world's female breast cancer incidence rate in the first place, which seriously endangers the health of women has become highly aroused world attention. Mammography X- ray image segmentation of breast lumps in the early diagnosis , early treatment has a very important significance , so as to breast cancer prevention and treatment efforts to gain time , to a certain extent, reduce morbidity and mortality. Breast lumps are Mammography X- ray image of a lesion occurs on the main performance, a standard technique for monitoring breast tumor segmentation and quantification of breast cancer is a very important step , because medical image is affected by many factors, in breast X-ray Mo automatic target mass image segmentation is generally more difficult . This article , on the existing home and abroad breast mass segmentation method conducted in-depth research and comparison, on this basis, puts forward a morphological marker controlled watershed algorithm and without re-initialization of the level set segmentation algorithm combined with a two-step France , to achieve automatic segmentation of breast lumps . Specifically , this paper made the following aspects of work : First , mark controlled by morphological watershed algorithm mammography X-ray images of the initial segmentation to obtain the initial rough outline. Then, as the acquired contours without re- initialization of the level set algorithm initial contour , secondary segmentation, contour obtained as the final contours. Our method can avoid the morphological marker controlled watershed segmentation algorithm is incomplete , and the experimental results in the MIAS and DDSM database proved the effectiveness of the algorithm .
|