The Study Of3D Liver MR Images Segmentation

With the development of medical imaging up to now, it has been widely used invarious related areas of clinical medicine. Using suitable image processing algorithms andprocessing the medical images accordingly, which can provide more efficient andconvenient information for medical image based on diagnosis and other research work, andmedical image segmentation occupies an important place in medical image processing.Interested region can be extracted from the medical image through the segmentationalgorithm and then can be shown separately, so that information lesions or normal tissuestructures can be offered more intuitively, and the result of the segmentation can be usedfor the subsequent processing based on certain purpose, such as image registration,quantitative measurements of the target organization.Magnetic resonance imaging plays an increasingly important role in the currentmedical research and clinical practice, compared with other imaging methods, which hasthe high ability of imaging for soft tissue and internal organs, and it can display humantissue anatomy clearly, and it also has the advantages of multiple parameters (T1, T2, etc.)、multiple imaging. The imaging effect of the MR image can distinguish among variousorganizations well, on this basis, interested area can be segmented more intuitively. Inrecent years, the national multiple liver lesions makes the segmentation of liver based onabdominal scan image be an urgent issue, but the abdominal area contains a large numberof visceral and soft tissues, complicated structure, and the adhesion between viscera andsoft tissue lead to there has infiltrative signs in the result of imaging, thus a large numberof weak edges and pseudo edges are formed, which makes it difficult to segment forinternal. Combined with the relative complex during the magnetic resonance imagingprocess, and some uncertainty in the imaging effect, and the widespread differences indifferent organs and individuals, so extracting the liver from abdominal scan imagesaccurately possesses important theory significance and practical value.In this paper, a variety of algorithms applied in medical image segmentation aresystematically analyzed, and their advantages and disadvantages and application scope were compared and summarized. According to the characteristics of the abdomen imagelevel set algorithm is selected to extract the liver, and the principle、characteristics of levelset algorithm and the various improvements and applications developed by researchersuntil now are described detailedly. Because of the complexity of the human body structureand the individual differences, now there still not has a single image segmentationalgorithm method achieved effective segmentation for all parts of the body, the currentmain research direction is integrating the advantages of various algorithms, combiningwith the morphological characteristics of target segmentation area for hybrid segmentation.Based on that,this subject selected the method of combing threshold segmentationalgorithm and level set method, after analyzing and consider and discuss variousalgorithms and studying the morphological characteristics and imaging characteristics ofthe liver, then adding some other auxiliary segmentation algorithm as auxiliarysegmentation, which can realize the3d image abdomen liver better.In this paper, the main research work is as follows:Firstly, stacking serial section images obtained from medical imaging equipment accordingto the scanning interval and slicing thick, in order to make it more close to the real human bodyto interpolate data among slices, to ensure the authenticity of the data interpolation data shouldbe reduced as much as possible.Secondly, filtering noise for the body data after interpolating, because the level setalgorithm is sensitive to edge image information, to keep the image edge, using gaussianfiltering or anisotropic diffusion filtering both can achieve good effect.Thirdly, extracting liver by using threshold segmentation combined with level set, andadding nonlinear mapping in this step, image enhanced and well speed image produced atthe same time, which makes the overflow phenomenon in the results of segmenting beenavoided.Fourthly, using light projection algorithm and combining with visualization toolkit VTKfor the result of segmentation and the effect of each processing algorithm in the intermediatesteps to do3d reconstruction. The experimental results show that this algorithm chosen obtained the ideal segmentationresults, and make level set algorithm applied into3d segmentation of liver well, and avoid theleak problem of level set algorithm in weak edge and provides the for the follow-up study ofliver.