Research Of The Automatic Pathology Detection Based On The Cerebral Computerized Tomography Images

Along with the fast development of the medical image technology, medical image processing and research aiming at computer aided diagnosis or computer intelligent diagnosis have became a hot point and the major trend in this field.The problem to be solved firstly is how to implement the automatic pathology detection of medical image. The automatic pathology detection involves image registration, image segmentation, construction of digital maps and such of image processing and analysis technology. The attention of this paper is concentrated on automatic pathology detection that deals with the treatment of computerized tomography images. The automatic pathology detection based on cerebral intensity abnormal pathology is researched combined with the medical professional knowledge.We proposed our images preprocessing method after analysis of the key technology in image processing and analysing, that are improved Demons algorithm which is a non - rigid registration, an automatic segmentation method for cerebral CT image based on prior knowledge and morphology, and the construction method of statistic maps based on intensity and labels. On the basis of these methods, by the computer simulation to diagnose thought of the doctors, we proposed an automatic detection method by means of intensity comparison. This method construct digital map represent of normal persons'image feature, then find the pathology district by the comparison of the patients'image to the intensity average map, overcome the effect of individual differences and non-completed registration by statistic maps based on labels, achieve the automatic pathology detection based on cerebral intensity abnormal pathology by fuzzy inference system based on professional image diagnosis knowledge.