Related Experiment Videos
A Review of Deep Learning Methods for Multimodal Medical Image Fusion
1Department of Management Science and Technology, University of Patras, 26334 Patras, Greece.
Sensors (Basel, Switzerland)
|July 28, 2026
Summary
This survey reviews deep learning methods for multimodal medical image fusion (MMIF), integrating diverse imaging data for better diagnosis. It covers CNNs, GANs, Transformers, and emerging techniques, offering a guide for future research.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computer Vision
Background:
- Multimodal medical image fusion (MMIF) integrates data from various imaging modalities to enhance diagnostic accuracy.
- Deep learning (DL) approaches have rapidly advanced MMIF since 2017, yet a comprehensive review is needed.
- Existing reviews lack up-to-date coverage of DL-based MMIF techniques and emerging methodologies.
Purpose of the Study:
- To provide a comprehensive survey of deep learning-based MMIF methods.
- To analyze DL frameworks, loss functions, evaluation metrics, and datasets used in MMIF.
- To identify challenges and future research directions in DL-based MMIF.
Main Methods:
- Categorization of MMIF approaches by DL frameworks (CNNs, Autoencoders, GANs, Transformers).
- Review of emerging DL methods like diffusion models and Mamba-based approaches.
- Analysis of datasets, evaluation metrics, and quantitative experiments on representative methods.
Main Results:
- Detailed review of mainstream and emerging DL architectures for MMIF.
- Summary of commonly used datasets and evaluation metrics.
- Proposal of unified evaluation metrics for standardized comparison.
Conclusions:
- Deep learning significantly enhances MMIF, with rapid advancements in network architectures.
- Standardized evaluation and exploration of novel methods like diffusion models are crucial.
- This survey provides a foundational understanding and roadmap for DL-based MMIF research.