Research On Recognition Technology Base On Evidence Theory For Biomedical Image

Peripheral nerve exists in various tissues and organs of the body which is easily injured in people’s daily life, human body tissues and organs injury is often accompanied by the destruction of the peripheral nervous system. The treatment of peripheral nerve injury usually cannot meet with a good result; in that case, people may become disabled. How to improve the effect of the treatment is what people now look for. To fulfill the accurate anastomoses of the same functional fascicular groups inside the impaired nerve is the prerequisite. But now, the treatment of peripheral nerve is based on the information of nerve section by histochemical stain. This method cannot display the change rule of overall length of peripheral nerve, and it dissatisfies the requirements of operation. In recent years, the development of computer science makes the three-dimensional (3D) visualization reconstruction system of peripheral nerve possible. By reconstructing the framework of peripheral nerve, we can easily get the information of nerve section and display the change rule of peripheral nerve which is benefit for treatment of peripheral nerve. The research on the three-dimensional (3D) visualization reconstruction system of peripheral nerve has just begun in and out of states. The solutions of how to make the stain more distinct and how to segment the nerve tract have been provided. But some research should be done on nerve fiber recognition.I did some research on the nerve fiber recognition of the three-dimensional (3D) visualization reconstruction system by using digital image process technique and pattern recognition method. I got some information of nerve fiber image and recognizes the category of nerve fiber. What have been done are as follows.1. By analyzing the benefits of three dyeing methods (Karnorvsky-Roots-Toluidine Blue-ponceau2R/Karnorvsky-Roots-Toluidine Blue-Chromotropic Acid2R/Karnorvsky-Roots-Toluidine Blue-Scarlet red), this paper shows the images of nerve fiber dyeing by different methods and analysis the characteristic of different nerve fiber.2. Nerve fiber image segmentation. Taking into account the characteristic of the nerve fiber, watershed algorithm base on improved seeds has been used in nerve fiber image segmentation which is very benefit in nerve fiber image.3. The feature extraction of nerve fibers and characteristic information fusion. This paper, I make use of the characteristic of the nerve fiber image, and then get the data of characteristic. By using three nerve fiber images dyeing by different method, I fused the information of nerve fiber finally.4. Nerve fiber recognition. By combining evidence theory and multisource information fusion and image feature classification, final classification results of nerve fiber achieved.Different peripheral nerve sliced images have characteristic of multisource which obtain from diverse ways of dyeing. Information fusion method was used in nerve fiber recognition. The results show that nerve fiber classification accuracy can be improved by information fusion method using evidence theory. This method has some advantages that traditional pattern recognition methods are incommensurable. It is simple and compatible.