Pixel level image fusion based on hybrid technique for multimodal medical images

Abstract

Image fusion is widely acknowledged as a useful tool for enhancing overall system performance in a variety of application areas such as battlefield surveillance, camouflaged ordnance detection, non- destructive testing defect detection, remote sensing, traffic control, machine learning and health care applications to name few. There has been a considerable increase in the number of scholarly literatures on the fusion of medical images in recent years. The significant growth can be largely attributed to diversity of complementary images from different sensors, advancement in low cost and high-performance medical imaging and computing technology, and extended usage of medical diagnostic instruments due to the belief placed by medical practitioner on the clinical improvements resulting from wide variety of image fusion methods. There are many distinct medical imaging modalities (CT, MRI, PET, SPECT etc.), and each has special qualities of its own. There are, however, drawbacks to the information gathered from single- modality medical imaging. Medical diagnosis cannot be aided by extensive lesion information from single- modality imaging, which inevitably results in annoyance and poor clinical diagnosis performance. Medical image fusion is a method for resolving this issue; it does so by merging the benefits and supplementary data of several models of source images, eliminating redundant data, and providing a more thorough, accurate lesion description to support specialists in diagnosis and decision-making. In nutshell, this research work summarises the state-of-art image fusion methods, identified the current issues aiming to offer hints and conceptual backing for subsequent research, thereby developing pixel-level hybrid techniques for multi-modal medical image fusion and highlighting potential development directions. newline