Abstract

In the current era of technological advancement, medical imaging assumes a substantial role in various applications of medical diagnosis and therapy. Medical image fusion is a potentially influential method for integrating many modalities of medical pictures through the application of image processing techniques. However, traditional methods have been unsuccessful in delivering satisfactory image quality evaluations and ensuring the durability of fused images. In order to address these limitations, the present study proposes a three-stage hybrid fusion methodology. This strategy involves the utilization of both Laplacian and Gaussian pyramid decomposition techniques on the source picture as an initial step. Subsequently, a weight-based convolutional neural network (CNN) approach is employed for generating the fusion results. The fusion of frequency bands is achieved through the application of pyramid reconstruction, employing the probabilistic fusion bands. The method described in this study was implemented in the MATLAB R2018a environment and has demonstrated superior quantitative and qualitative analysis capabilities when compared to conventional methodologies.

Description