A Systematic Literature Review on MultimodalMedical Image Fusion

• 全文链接
• 请求原文

作   者：

Shatabdi Basu;  Sunita Singhal;  Dilbag Singh; 

作者机构：

India;  Department of Radiology;  NY;  USA;  Center of Biomedical Imaging;  New York University Grossman Schoolof Medicine;  Manipal University Jaipur;  New York;  Ajmer Express Highway;  Jaipur303007;  Computer Science and Engineering;  Rajasthan; 

关键词：

Image modality;  Medical image fusion rules;  Multimodal medical image fusion;  Evaluation metrics;  Systematic literature review; 

期刊名称：

Multimedia tools and applications

i s s n：

1380-7501

年卷期：

2024 年 83 卷 6 期

页   码：

15845-15913

页   码：

摘   要：

Medical image fusion is a relevant area with widespread application in disease diagnosisand prediction with easily available image scans of Computed Tomography, Positron EmissionTomography, Magnetic Resonance Imaging, and Single Photon Emission ComputedTomography. Each diagnostic imagemodality has its advantages and limitations.MultimodalMedical Image Fusion aims to combine more than one image of the same or different modalityto enhance the image content and providemore information about diseases.We performeda Systematic Literature Review according to the methodology outlined in Kitchenham Charterand based on our search string, we extracted 844 studies from four electronic databasespublished between 2017 and 2021. Around 175 studies were selected for further in-depthanalysis using inclusion and exclusion criteria.We further divide this review article into fivesections that (a) Identify the most frequently used input image decomposition methods (b)Describes the most common fusion rules applied on decomposed sub-bands (c) Discusses theoptimization algorithms used to improve the efficiency of the fusion scheme (d) Examinesthe modalities which are subjected to image fusion techniques in the medical domain (e)Identifies the evaluation metrics used to judge the effectiveness of image fusion technique.The result of the comparative study of five sections highlights that the majority of studies usemultiscale decomposition methods, and hybrid and neural network-based fusion rules whilethe CT-MRI combination was mostly used as an input dataset. The review also indicated theprevalent use of particle swarm optimization and non-reference metrics in the majority ofstudies. Our results suggest that medical image fusion can improve the quality and accuracyof medical images for diagnosis and treatment planning. Further research can be conductedto handle potential research gaps outlined in this review and optimize medical image fusionfor clinical applications.