Processing Of Medical Image Based On Fractal Dimension And Multi-Scale Fractal Feature

Combination medical imaging and digital image processing technology, by choosing appropriate processing and analyzing method, high quality medical image can be obtained. This is an effective method to improve the accuracy rate of medical diagnosis.The obtained Medical images can not avoid the signal interference in the transmission process, so the original image there is often a lot of noise, organizational boundaries are not clear, the doctor is difficult to make accurate judgments. In order to reduce the noise in the image and improve the contrast of the image, traditional digital image processing technology is applied in this paper. The image after pre-processing and more suitable for later processing.Two fractal methods were introduced to detect medical image lesions in this paper. First is the box-counting dimension method. Segmented medical image into a large number of small lattices, then calculated the fractal dimensions of each small lattice one by one. There were clear differences in fractal dimension between the edges of the image and background regions. Analyze the value of fractal dimension ( DB ), the set of all fractal dimension DB <2 is the edges of the medical image. The second method selected to detect the medical image after pre-processing is multi-scale fractal feature ( D MF). Selectedε₁ =3andε₂ =5 as measuring scale, using dilation and erosion methods in mathematical morphology to processes the image respectively, Dε₁andDε₂was obtained. Then, the medical image was divided into size 15×15sub areas, to make sure that background, normal regions and lesion regions of the medical image are possessed of different sub areas. The ( DM F)values of different regions in a medical image are normally different. For example, in a medical x-ray image with bone fracture,the ( DM F)values of the background regions are between 0.008 and 0.020; the ( DM F)values of the normal regions are between 0.040 and 0.070; and the ( DM F)values of the knot regions are between 0.100 and 0.200. Extracted singular multi-scale fractal fracture, their set was the margins of the image.This experiment collected one hundred medical sample images from hospital; it can be known that these two methods offered in this paper are efficiency amount to 95%. The results are satisfactory.