Keywords : Edge Preservation ��� Histogram Equalization
MULTI MODALITY MEDICAL IMAGE FUSION BY COMBINING ENTROPY DATA WITH 2D DISCRETE WAVELET TRANSFORM (2D DWT) AND ENTROPY PRINCIPLE COMPONENT ANALYSIS (EPCA)
European Journal of Molecular & Clinical Medicine,
2020, Volume 7, Issue 2, Pages 5695-5711
In the recent past, multi-modality medical image fusion method plays significant role in Computer Aided Diagnosis (CAD) for accurate and efficient classification and diagnosis of brain tumor disease. Conventional methods may introduce artifacts, loss of pixels in segmentation of tumor region or low quality fusion images. The multimodality medical images should be properly filtered all available noises and enhanced in terms of contrast and brightness. In this paper, a novel methodology for image fusion is introduced by combining entropy data based on Entropy Principle Component Analysis (EPCA) in 2D-Discrete Wavelet Transform (2D-DWT). The image restoration methodology is applied on multi-modality medical image to remove various noises such as salt and pepper noise, random noise and Gaussian noise. The 2D-Adaptive Bilateral Gabor Filter (2D-ABGF) is implemented for image restoration method. The filtered image is given for image enhancement for improving quality of image. The Edge Preservation – Histogram Equalization (EPHE) methodology is applied to improve contrast and brightness of the image. The 2D DWT algorithm is used to decompose the images into low and high frequency coefficients. The EPCA fusion rule is used to apply fusion rule. The frequency bands are applied for combining entropy data of multi-modality images for efficient and accurate fusion. The inverse 2D DWT is used to convert fused coefficients to estimate the final fusion image. The novel fusion rule is used based on combining entropy data of multi-modal images to reduce the dimension of image dictionary and cost of computation. The various fusion parameters such as structural similarity metric, image fusion index of quality, standard deviation, entropy and mutual information are compared with existing methodologies to prove the accuracy and efficiency of proposed methodology.