A Comprehensive Review and Comparative Analysis of Pixel-Level Image Fusion Methods for Multi-Modal Medical Imaging

Abstract

The necessity for image fusion in modern computer imaging systems is growing, mostly because of the proliferation and diversification of image acquisition methods. Image fusion refers to the practice of merging relevant information from multiple sensors using mathematical tools to produce a comprehensive composite image. This composite image is designed to be more valuable and practical for human operators and other computer vision applications. This paper provides an extensive survey of pixel-level image fusion methods for multi- modal radiology and diagnostic imaging, evaluating their strengths and weaknesses. By analyzing the performance and constraints of individual methods, we aim to offer a comprehensive view for researchers and practitioners to select the most appropriate fusion strategy for their specific applications. Also, the paper highlights the benefits of preserving supplemental information, avoiding artefacts, and ensuring robustness in the fusion process. Further, numerous image quality assessment metrics are employed to ensure images are reliable for clinical use. Although many methods have been proposed in the literature to accomplish pixel-level image fusion, the hybrid approach has produced high-resolution, low-noise fused images, improving the reliability of radiological studies for medical analysis and diagnosis. This work found that, despite several image fusion methods, there remain several potential hybrid approaches for medical image fusion applications. In recent years, the emergence of deep learning techniques has driven significant advancements in image fusion and shows great potential as a foundation for real-time multimodal fusion. Multi-modal medical image fusion has demonstrated significant success in enhancing the clinical accuracy of decisions derived from medical images. Hence, this review paper will provide readers with insights into recent discoveries in medical image fusion and their implications for the future.