Implementation of a VGG-19 and Discrete Wavelet Transform Combined Multimodal Fusion Technique

The integration of diverse medical imaging modalities facilitates the identification of diseases. Medical imaging is a critical component of medical research and diagnosis, providing detailed information about the structure and function of the body. In some cases, imaging approaches that utilize a single modality may not capture the complete set of the diagnostic data necessary for reliable physician evaluations. The objective of this study is to enhance the clarity of medical imagery and facilitate more precise disease identification. The proposed approach involves a multimodal medical image fusion technique that integrates Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) data. The suggested technique involves three sequential steps: image registration, image merging, and image segmentation. Image registration is a process that aligns CT and MRI images by utilizing procedures that are based on landmarks to ensure that pixel-level correlation is maintained. To preserve both structural and functional characteristics from the input pictures, the fusion procedure makes use of deep learning-based transfer learning in conjunction with the VGG-19 network and Discrete Wavelet Transform (DWT). Lastly, the watershed algorithm is employed to extract and highlight Regions of Interest (ROIs), such as tumors, during the segmentation process. The suggested method substantially increases picture clarity, maintains essential characteristics, and boosts the precision of tumor segmentation, as demonstrated by the results of the experiments.