Hippocampal Contour Extraction Based On Region Growing And Level Set And Three Dimensional Reconstruction

Image segmentation technique is an important part in computer vision detection. In recent years, as one of its important branches, medical image segmentation has beening attentioned by domestic and foreign researchers. It is the basis of two-dimensional observation, three-dimensional reconstruction and subsequent analysis of normal tissue and pathological tissue. The following work is affected by the quality. With the characteristics of fuzzy> multi-mode and uncertainty, medical images segmentation is so hard. With the development of medical imaging technology, there are a large number of imaging modalities. On one hand, More sufficient information is provided by image data, on the other hand a higher segmentation methord to capture the feature is put forward, so medical image segmentation is a difficult, but also a meaningful research fields.The volume of the hippocampus is used as the diagnosis standard of the diseases in clinical, such as Epilepsy, Alzheimer’s disease, Cognitive impairment, Schizophrenia and other neurological diseases. Because the3D reconstruction of the hippocampus is based on the2D segmentation,2D segmentation of the hippocampus is crucial. There are few studies about the contour extraction and the three-dimensional reconstruction, so it’s significant to do some researches on these fields.The main work of this thesis is as folio wings:(1) The area of the hippocampus is small, in order to reduce the complexity, the artificial clicked point was used as the center to shear a square area from the slice in this paper. For the sake of avoiding the interference of the noises, the gray of the seed point’s was replaced by5*5field gray mean of the seed point in region growing.(2) The gray mean of the grown region was applied in traditional region growing, a more accurate gray reference was provided. The standard deviation of the grown region was put forward as the threshold in region growing, the adaptive region growing was realized. Based on the disorderly phenomena of the region growing results, the mathematical morphology method was joined to treat the boundary, a better boundary was obtained.(3) The boundary was set as the initial contour of the level set in this paper, it gave the prior information to the segmentation; the boundary was easily staying at the local extreme values, In order to cope with this phenomena, a new level set based on target gray differences was proposed. Gray energy variations was joined in the traditional energy functions, then the accuracy was improved, the whole efficiency of the segmentation was improved. Image sequences of the hippocampus were segmented in this way, the results of experiments showed that this method had a better adaptability.(4) Finally, volume rendering method was chose to reconstruct the hippocampus, a good reconstruction result was obtained, and a valuable hippocampus contour was provided for the doctors.