A Comprehensive Study on Multimodal Medical Image Fusion Techniques for Health Diagnosis and Challenges

Abstract

The combination of digital medical imaging technology and computer vision is extremely important for the correct diagnosis of the human body. Different imaging technologies such as Computed Tomography (CT), Single Photon Emission Computed Tomography (SPECT), Positron Emission Tomography (PET), Magnetic Resonance Imaging (MRI), and ultrasound scans are able to detect anomalies in different parts of the body; however, the simultaneous diagnosis of different diseases through separate clinical images is a complex process, taking considerable time, especially in the case of patients in critical care. To tackle this issue, the concept of multimodal image fusion techniques has been reintroduced in order to combine the different characteristics of the images from multiple sources in a distinct clinical image, thus allowing for efficient anomaly recognition. This paper aims to present a comprehensive review of some well recognized existing traditional hand-crafted medical image fusion techniques as well as the existing deep learning-based multimodal clinical image fusion techniques proposed by different researchers. Additionally, the paper aims to examine the shortfalls of the above-mentioned techniques based on performance metrics Average Gradient (AG), Spatial Frequency (SF), Mutual Information (MI), QAB/F in a comprehensive manner, thus allowing for a clear differentiation in the existing knowledge gaps, as well as the categorization of the traditional hand crafted solutions and the recent advances in the field of deep learning-based multimodal medical image fusion technology.