Subjective and Objective Evaluation of Noisy Multimodal Medical Image Fusion Using 2D-DTCWT and 2D-SMCWT

This paper focuses on the evaluation of noisy image fusion for medical images obtained from different modalities. In general, medical images suffer from poor contrast and are corrupted by blur and noise due to the imperfection of image capturing devices. In order to improve the visual and quantitative quality of the fused image, we compare two algorithms with other fusion techniques. The first algorithm is based on Dual Tree Complex Wavelet Transform (DTCWT) while the second is based on Scale Mixing Complex Wavelet Transform (SMCWT). The tested algorithms are using different fusion rules in each one, which leads to a perfect reconstruction of the output (fused image), this combination will create a new method which exploits the advantages of each method separately. DTCWT presents a good directionality since it considers the edge information in six directions and provide approximate shift invariant as well as SM-CWT, the goal of PCA is to extract the most significant features (wavelet coefficients in our case) to improve the spatial resolution. We compared the tested methods visually and quantitatively to recent fusion methods presented in the literature over several sets of medical images at multiple levels of noise. Further, the tested fusion algorithms have been tested up to the important level of Gaussian, salt & pepper and speckle noise (350 test). For the quantitative quality, we used several well-known fusion metrics. The results show that the tested methods outperform each method individually and other algorithms proposed in the literature.