Intelligent Fusion Approach for MRI and CT Imaging using CNN with Wavelet Transform Approach

Abstract

Medical image fusion (MIF), with its numerous medical uses for precisely diagnosing medical imaging, has attracted meticulous attention. The fused picture disadvantages from weak contrast, uneven lighting, the existence of noise, and incorrect fusion procedures, leading to an insufficient sparse representation of significant characteristics. Various MIF approaches have been presented to date. This study suggests a bottom-hat-top-hat paradigm for morphology preprocessing to deal with noise and non-uniform light. The wavelet transform approach then effectively restores all important aspects in all dimensions and dimensions by breaking the images down into the Low-Pass (LP) and High-Pass (HP) sub-bands. In order to efficiently capture smooth edges and textures, Different sides of the Convolutional Neuronal Network receive the HP sub-bands This is done through a process of Feature Recognition, Initial Segmentation, And Consistency Confirmation. Whereas the LP sub-bands are merged using local energy fusion, the energy information is recovered utilizing the averaging and selection mode. Qualitative evaluations, both subjective and objective, are used to support the proposed strategy. 12 field experts proved the effectiveness of the proposed methods based on precise details, visual contrasts, distortion in the reconstructed images, and no data redundancy using a customer specific example to make the subjective judgments.