The Research In Method Of Measureing Volume Of Intracranial Hematoma Based On CT Images

Intracranial hematoma ,especilly acute intracranial hematoma is one of head injuries endangering health of human.The volume of intracranial hematoma is an important index which determines the patient's operation .Measuring the volume of intracranial hematoma accurately has weighty clinical value.This paper studies the processing and segmention of CT images in this background,then obtaining the method of measuring volume automaticly which meet clinical need.CT(Computed Tomography) utilizes the X-ray's attenuation character and turn the density difference into gray variance .In the image of acute intracranial hematoma, CT is superior to MRI.So,CT image have been generally used for acute intracranial hematoma diagnosis. Thinking of the character of CT imaging ,as well as noise ,artifact ,partial vilume effect,we filter CT images. The paper put forword a new filtering method:nonlinear anisotropis diffusion filtering .It can not only remove noise from medical images but also keep the details and edges of images as well .CT image inflects the density difference of tissues ,then we shall process two steps in segmention of CT images series when measuring the volume of intracranial hematoma.First,removing the shin, muscle ,scalp hematoma whose gray level is similar to intracranial hematoma,we completes the first step of segmention. In removing the non-brain tissue ,we utilizes thresholding region of image logil operation Because the gray level of skull is apparently different from the other tissue in CT images. We can easily obtain the skull edges by experienced threshold ,instead of other complex method . Four-domain growing is used in region growing .it travals the whole brain tissues and intracranial hematoma which are presented by image logic operation .Results show it simple and effective.Second .because the brain tissue and hematoma are different in gray level ,we can remove the brain tissue by threshold method . we obatains the unique image of intracranial hematoma and completes the second step of segmention. Eventually ,we measures the volume of intracranial hematoma by summing all areas of image series . For CT image's noise,we employ two-dimensional entropy threshold method to segment brain tissue and intracranial hematoma.lt not only reflects gray-level of image pixel,but also reflects neighborhood gray-level information of image pixel .So two-dimensional entropy threshold method may restrict noise and remove the tiny framework efficiently.Entropy is a concept of information theory .It proven a new good method of image segmentation by maxmizing the rule function which embodys the infomation quantity of segmention result.However the amount of calculation in two-dimensional entropy threshold method is far too much .In this thesis we employ the improved genetic algorithm .GA is a new search algorithms which is applicabe to concurrent distribution process.We utilize hierarchic genetic algorithm and improved selection operator crossover operator ,mutation operator .Especially to avoid "premature" ,we use adaptive probability of crossover and probability of mutation which are determined by concentrative degree of fitness function.In the lower rank GA we devise various probability of crossover and probability of mutation which produce the diversity of the population .According to the experiment, hierarchic GA is better than common algorithms and simple genetic algorithms.