Authors Paper Title Abstract (Concise) Challenges Contribution
Li et al. Multimodal Surveys various techniques for Ensuring accurate Comprehensive
Medical Image fusing medical images from alignment between overview of existing
Fusion: A Review different modalities (e.g., MRI, images from different methods and their
CT, PET). Highlights the modalities. Developing applications in
importance of accurate robust fusion algorithms medical diagnosis.
registration and effective feature that preserve important
extraction. information while
minimizing artifacts.
Ghassemi Deep Learning Focuses on the application of Limited availability of Provides insights
et al. for Multimodal deep learning techniques for large, annotated datasets into the potential of
Medical Image medical image fusion, including for training deep learning deep learning for
Fusion: A Review convolutional neural networks models. Ensuring the advancing medical
(CNNs) and autoencoders. interpretability and image fusion.
Discusses the potential of deep explainability of deep
learning to improve fusion learning-based fusion
performance. methods.
Zhang et Multimodal Proposes a novel CNN-based Training deep learning Introduces a new
al. Medical Image method for fusing MRI and PET models can be approach to
Fusion Using images. Demonstrates improved computationally medical image
Convolutional image quality and diagnostic expensive and fusion using deep
Neural Networks accuracy compared to traditional time-consuming. learning, with
methods. Ensuring the potential for
generalizability of the improved clinical
proposed method to outcomes.
different types of medical
images.
Viola et Multimodal Image Provides a comprehensive Emphasizes the need for Offers a broad
al. Fusion for review of multimodal image robust and reliable fusion overview of the
Medical fusion techniques in medical methods that can state-of-the-art in
Diagnosis: A diagnosis, covering various improve diagnostic multimodal medical
Survey modalities and applications. accuracy and clinical image fusion.
Discusses the challenges and decision-making.
limitations of existing methods.
Hong et A Novel Proposes a new fusion method The proposed method Introduces a new
al. Multimodal based on the non-subsampled may be computationally approach to
Medical Image shearlet transform and sparse more complex than some medical image
Fusion Method representation. Demonstrates traditional methods. fusion with potential
Based on improved image quality and for improved
Non-Subsampled feature preservation compared performance.
Shearlet to traditional methods.
Transform and
Sparse
Representation
Shen et Multimodal Reviews recent advances in Highlights the need for Provides a
al. Medical Image deep learning-based multimodal more research on comprehensive
Fusion with Deep medical image fusion. interpretability, overview of deep
Convolutional Discusses the potential and explainability, and learning
Neural Networks: challenges of deep learning in generalizability of deep applications in
A Review and this area, and outlines future learning models for medical image
Future Directions research directions. medical image fusion. fusion and identifies
promising areas for
future research.