Comparison of Medical Image Fusion Algorithm for Preserving the Edge Information Based On Improved Wavelet Coefficient Contrast

now a day, multimodality medical image fusion has drawn lots of attention with the increasing rate at which multimodality medical images are available in many clinic application fields. Radiotherapy plan, for instance, often benefits from the complementary information in images of different modalities. CT scans and MRI scans contains details regarding soft and hard tissues. For medical diagnosis, CT provides the better information on denser tissue with less distortion, while MRI offers better information on soft tissue with more distortion .For Medical applications, these CT and MRI images needs to be fused with high efficiency for diagnosis purpose. Hence medical image fusion has been very important. In our paper, we have described methods for medical image fusion-simple averaging fusion rule based on wavelet, maximum pixel replacement based on wavelet transform ,contrast based image fusion, discrete wavelet packet based image fusion and discrete packet based image fusion using contrast. Depending upon the requirement of fused image and application, desired method is used. Discrete Wavelet Transform is used as it separates out the low frequency band and high frequency band and further sub bands (i.e. LL, LH, HL, HH bands). The approximation coefficient i.e. is most commonly fused via uniform averaging. This is because approximation coefficients are interpreted as the mean intensity value of the source images with all salient features encapsulated by the detail coefficient sub-bands at their various scales (Piella, 2003). Therefore, fusing approximation coefficients by averaging maintains the appropriate mean intensity needed for the fusion result with minimal loss of salient features. In case of simple averaging fusion rule based on wavelet transform, all the sub bands are fused via uniform averaging. But high frequency contains both edge information as well as noise, and due to averaging noise also gets introduced in the final fused image. In case of maximum pixel replacement based on wavelet transform method, detail coefficients are fused by replacing the maximum pixel value in the fused image. In this method though maximum edge information is preserved but, more noise is also introduced. Entropy of such fused image is higher. In our contrast based image fusion, contrast corresponding to detail coefficient is calculated and preserved in the final image. This lead to increase in contrast and edge information in fused image. Power Signal to Noise Ratio (PSNR) of such fused image is higher than other two methods. In discrete wavelet packet based image fusion method, LL band is again decomposed into L, H, V, D bands and then mean fusion rule is applied to obtain the fused image ,using inverse of the image obtain by fusion rule. Similarly in contrast based method, contrast corresponding to each pixel is calculated and depending upon the contrast, fused image is obtained. We have compared the results of their above methods and can be concluded that each method has its own application and use, which depends upon the requirement of fused image. In image fusion techniques, contrast and edge information are generally lost. Since these are the salient features from medical diagnosis point of view, needs to be preserved. Hence with use of contrast based image fusion, these parameters can be preserved. The visual experiments and the quantitative analysis demonstrate that the contrast medical image fusion method can preserve the important structure information such as edges of organs, outlines of tumours compared to other image fusion methods. This characteristic makes the contrast based methods a promising application in medical diagnosis. Keywords— Medical image fusion, wavelet coefficient contrast, edge preservation, performance evaluation, Medical diagnosis, discrete wavelet packet transform.