Research On Brain MRI Hierarchical Segmentation And Intracranial Pressure Measurment

Measurement of intracranial pressure (ICP) is important in the management of patients with severe head injury, cerebral ischaemia and subarachnoid haemorrhage. Thus, invasive measurement of ICP is used more often. Invasive measurment could lead to various complications. Then noninvasive measurement was needed.The measurement of ICP can tell doctors if cerebral disease would come on. With the help of image processing, doctors can improve the accurateness of medical diagnosis and theropy.MRI (Magnetic Resonance Imaging) is an advanced medical imaging technique providing rich information about the human soft tissue anatomy. However, the amount of data is far too much for analysis/interpretation, and this has been one of the biggest obstacles in the effective use of MRI. Segmentation is one of the most difficult and important steps in digital image processing. Segmentation accuracy determines the eventual success or failure of computerized analysis procedures. In this thesis, we proposed a novel noninvasive measurement of ICP and presented a hierarchical image segmentation algorithm.Firstly, This paper presents a novel mathematical model of noninvasive monitoring intracranial pressure based on physiology and hemodynamics. Not only the circulation of the cerebral blood but also the circulation of the cerebrospinal fluid is considered in this model. The model includes the main biomechanical factors assumed to affect intracranial pressure, particularlycerebrospinal fluid (CSF) dynamics, intracranial compliance, and cerebral hemodynamics. This is consistent with the fact. We can list some state equations on the basis of the model. These equations can be solved using MATLAB. The model is simulated with SIMULINK in this paper. A comparison between the experimentally recordings and simulated intracranial pressure signals shows that they are consistent and that there is a correlation between the intracranial pressure and the arterial pressure. This proves the possible validity of the model. This paper provides an assistant method for clinic noninvasive monitoring intracranial pressure.Secondly, for traditional medical image smooth filtering methods, the details and edges of images would be lessened as well as noise is removed. In the paper, we put forward a new smooth filtering method, which is based on fuzzy theory and pixel's gray degree operation. The new method can not only remove noise from medical images but also keep the details and edges of images well.Thirdly, combining the improved watershed algorithm and the modified fuzzy c-means method, we presented a hierarchical image segmentation algorithm. The improved watershed algorithm is based on the probability function, instead of the traditional gradient. This algorithm is essential to obtain accurate results. The detection of thin structures, usually a major drawback of the watershed transform, is significantly improved through the use of our novel algorithm. Although the image was filtered, the over-segmentation sill existed. To overcome the over-segmentation, this paper utilized a modified fuzzy c-means clustering algorithm. We first initialized the cluster numbers and the cluster centers. Then we optimized the data set to reduce the time for each iteration .At last, we modified the computing steps to lessen storage space of the operation data. All the above steps improved the efficiency of the fuzzy clustering algorithm. Results were presented to show that the hierarchical image segmentation algorithm was faster and more accurate.Finally, a summary of our work and a prospect of future research are given.