Dissertation > Excellent graduate degree dissertation topics show

Research and implementation of algorithms brain MR image segmentation

Author: ZhaoZuo
Tutor: JieMei
School: University of Electronic Science and Technology
Course: Signal and Information Processing
Keywords: MR image segmentation intensity inhomogeneity correction skull stripping KNN
CLC: TP391.41
Type: Master's thesis
Year: 2011
Downloads: 87
Quote: 3
Read: Download Dissertation

Abstract


The research for magnetic resonance images of brain has always been one of the hottest fields. And image segmentation is one of the most critical processing steps before image analysis. For the medical image segmentation, a good accuracy of results is very important and helpful for the following image processing. Therefore, it’s meaning for the segmentation of brain tissues for magnetic resonance images. However, the current segmentation results which are extracted by the computer itself or lots of manual intervention are not accepted for the bad accuracy in medical practical applications. So, in this thesis, we focus on the studies of the segmentation of brain tissues for MR images. The main works are as follows:1. Analyzed the principle of MR technique, and summed up the characteristics of MR images, also the current common methods of brain tissue segmentation for MR image are summarized.2. Introduce the purpose and significance of the MR images preprocess. Study a skull stripping algorithm based on morphology in detail. First, the MR image is processed with an anisotropic diffusion filter to smooth non-essential gradients. We then apply a edge detector to the filtered image to obtain the boundaries image. And then the largest connected region is identified using a sequence of morphological and connected component operations. For most of brain MR images, satisfactory results are obtained using this method. But those MR images containing eyeballs, jaw or other complex organization is hardly segmented accurately by our method. Also, for the images of injured brain or brain lesions, it is difficult to get the right results.3. Study a skull stripping algorithm based on surface model in detail. And introduce the principle of the algorithm. A triangular tessellation of a sphere’s surface is initialized inside the brain, and allowed to slowly deform, following the“internal force”that keep the surface well-spaced and smooth, and the“external force”that make the surface attempting to move towards the brain’s edge. Finally, the skull is removed. Compared with the algorithm based on morphology, this method can obtain a smoother surface model due to the“internal force”. For the images of injured brain or brain lesions this method can obtain a better result. But, this algorithm is only used for 3D MR images, and its application is limited.4. Summarize current popular intensity inhomogeneity correction methods. Study a algorithm for intensity inhomogeneity correction and segmentation of MR images which based on energy minimization. The intensity inhomogeneity(bias field) is modeled as a linear combination of a set of basis functions, thereby the clinical images are parameterized by the coefficients of the basis functions. And then define an energy function which depends on those coefficients. The coefficients can be determined through minimizing the energy function. Finally, get the corrected image and segmentation results. Through the experiment, we have proved that the algorithm can effectively correct the intensity inhomogeneity . And the algorithm is not sensitive to the initial values. Also have a high convergence speed and precision segementation result. But this method can not solve the problem of partial volume effect in MR images.5. We proposed a KNN-based intensity inhomogeneity correction and segmentation method to solve the problem of partial volume effect in MR images. This method determined the membership function using the K nearest neighbor algorithm, which make full use of the pixel neighborhood information. So, this method can eliminate partial volume effect effectively. Because this algorithm is based on the energy minimization method, it also have the same advantages with the method mentioned in section 4: not sensitive to initial values, high convergence speed, and precision segmentation.

Related Dissertations

  1. Transportation network path and neighbor selection study,U116.2
  2. Financial Index Forecast Based on Data Mining,F830
  3. Feasibility Study on Discriminating Adulterated Milks by Near Infrared Spectroscopy,TS252.7
  4. Computer-aided Method of Detecting Lung Fissure Based on CT Images,TP391.41
  5. Ontology-based medicine named entity recognition technology research,TP391.1
  6. Research on Relation Extraction of Person Entity in News Webpage,TP393.092
  7. Subspace-based face recognition algorithm,TP391.41
  8. Research on Web Content Filtering Based on Concepts of Collection,TP393.092
  9. Content-based spam filtering technology research,TP393.098
  10. Research of Spatial Outliers Detection Algorithm,TP311.13
  11. The Research and Implementation of Web Page Classification in Enterprise Search Engine,TP393.092
  12. Text Categorization of Chinee Small Sample Based on Graph Model,TP391.1
  13. Research and Realization on Correlation Techniques of Topic Search-Specific Engine,TP391.3
  14. Production Forecast for Daliudi.S1 Gas Reservoir,P618.13
  15. Preparation and Electric Properties of Lead-Free Sodium Potassium Niobate Based Piezoceramics,TM282
  16. On Research for Chinese Automatic Text Categorization Technology Based on VSM Model and Feature Selection,TP391.1
  17. Research on Web Text Clustering and Classification Algorithm,TP391.1
  18. Lead-free Piezoelectric Ceramics (K_ (0.44)) Na_ (0.52) Li_ (0.04)) (Ta_ (0.2) Nb_ (0.8)) O_3-xCeO_2 Preparation and Properties,TM282
  19. Neighborhood Rough Set and Its Classification Algorithm,TP18
  20. Research on the Joint Classification Based on Support Vector Machine and K-nearest Neighbor,TP18
  21. A Surface Reconstruction of Point Clouds Based on Spacial Status Octree,TP391.41

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device
© 2012 www.DissertationTopic.Net  Mobile