The Research Of Biologic Medical Image Based On Computer Intelligent Indentify Key Technology

The application of Digital Image Process becomes a hot spot in medical science field for the passed few years. It develops at full speed and provides the possibility to make two dimension images into three-dimensional reconstruction. After identifying the nerve inner function bind types in nerve slice, three-dimensional visualization of peripheral nerve starts to reconstruct for two dimension image, it can realize the direction rule of peripheral nerve inner function bind directly and totally. There was no relevant method to automatic identification for medical nerve slice staining image in inner function bind types at present, so it used to be identify the function bind types by experienced doctor artificially, that resulted in a lot of work loan and low error precision. While inner nerve function bind was the key step before nerve slice image was built, it decided three-dimensional visualization effect of nerve inner function bind. This project is an interdisciplinary subject research among artificial intelligent, pattern recognition and biomedicine, realizing the three-dimensional visualization, thus it has important theory research significance and practical value.This issue was proposed on the basis of the key technology research of peripheral nerve function bind three-dimensional digital simulated system (No.9151008901000006); it’s a project which cooperates with Sun Yat-Sen University affiliated first hospital and was supported by Guangdong province natural science fund. This paper’s aim was to explore and research the automatic identification of the peripheral nerve function bind types. Research contents as follows:Firstly, analyzing the relevant preprocessing technology about medical science image, this paper took nerve slice dye image as an object, in order to denoising.Secondly, aiming at different texture specificity in different kinds of types nerve function bind in peripheral nerve slice image, we took grey mean value and variances as texture evaluate parameters to analyze and extract characteristic.Thirdly, traditional K-means cluster algorithm has limitation, aiming to the boundedness, so we proposed a kind of modified K-means cluster algorithm and applied the automatic identification field which different nerve function bind types in medicine science nerve slice dye image.At last, integrating the image border extract and image registration function module etc., designing and developing the nerve slice image three-dimensional reconstruct processing platform based on VS2005, in order to realize the interactive operation of three-dimensional visualization in peripheral nerve slice image.