Image Fusion in Medical Imaging: Techniques, Clinical Applications, and Workflow Integration

Image fusion in medical imaging is the process of combining complementary information from two or more imaging modalities into a single composite image that is more diagnostically useful than any individual input. This article explains the core techniques, clinical applications, and practical workflow considerations for students, researchers, analysts, and life-science professionals who need to select, implement, or evaluate fusion methods in medical imaging projects.

Medical imaging modalities provide different views of the same anatomy or pathology. Computed tomography (CT) excels at bone and dense tissue visualization, magnetic resonance imaging (MRI) provides superior soft-tissue contrast, positron emission tomography (PET) reveals metabolic activity, and ultrasound offers real-time functional information. No single modality captures all clinically relevant features, which is why fusion has become a standard tool in diagnosis, treatment planning, and image-guided interventions. The practical challenge is choosing the right fusion approach for a given clinical question and modality pair, then integrating it into a reproducible workflow.

At a Glance: Image Fusion Decision Framework

The table below provides a practical starting point for selecting an image fusion approach based on the clinical question and the modalities being combined. This framework is derived from published evidence on fusion applications across oncology, neurology, cardiology, and interventional radiology.

Core Principles of Medical Image Fusion

What Fusion Accomplishes

Image fusion integrates complementary information from multimodal images to produce a new composite image that is more informative for visual perception than any individual input image. Radiologists need more informative and high-quality images to diagnose diseases, and fusion plays a vital role in biomedical image analysis by combining the strengths of different modalities while minimizing their individual limitations. For example, lung cancer diagnosis currently relies on multimodality image fusion to find anatomical and functional information about tumors and metabolic measurements to identify cancer stage and metastatic disease. The success of this approach comes from combining PET and CT imaging advantages while reducing the impact of their respective weaknesses.

The Role of Registration

Before any fusion can occur, the images must be spatially aligned through a process called registration. This is a technically challenging and time-consuming step in clinical settings. Registration can be rigid, affine, or deformable depending on whether the anatomy has changed shape between acquisitions. Deformable image fusion is particularly important in interventional settings where tissue shifts or respiratory motion alters anatomy between scans. The quality of the fused image depends directly on the accuracy of the underlying registration, and poor registration produces misleading fusion results regardless of the sophistication of the fusion algorithm.

Spatial Resolution and Contrast Tradeoffs

Different modalities offer different spatial resolutions and contrast mechanisms. CT provides high spatial resolution for bone and calcified structures, while MRI offers superior soft-tissue discrimination. PET and single-photon emission computed tomography (SPECT) provide functional information at lower spatial resolution. Fusion methods must handle these resolution mismatches without introducing artifacts or losing clinically relevant detail. Some fusion approaches preserve the highest spatial resolution from one modality while overlaying functional information from another, which is the principle behind PET-CT display fusion.

Fusion Techniques and Their Evolution

Spatial Domain Methods

Spatial domain fusion methods operate directly on pixel intensities. Principal component analysis (PCA) is a classical approach that extracts relevant information from large datasets based on eigenvalue decomposition. When applied to image fusion, PCA identifies the most informative components from each input image and combines them into a fused output. A cascaded PCA approach combined with shift-invariant wavelet transforms has been shown to improve information content, preserve edges, and enhance fused image quality for MRI and CT brain images. The wavelet transform operating in the complex domain with shift-invariant properties brings out directional and phase details that standard spatial methods miss.

Transform Domain Methods

Transform domain methods convert images into a different representation before fusion. Wavelet transforms are the most common example, decomposing images into different frequency bands that can be fused separately. The dual-tree complex wavelet transform with a maximum fusion rule enhances average information and morphological details. These methods are particularly effective at preserving edges and directional features because they capture information at multiple scales and orientations.

Sparse Representation Methods

Sparse representation (SR) is a robust signal modeling technique that has demonstrated success in multi-dimensional medical image fusion. A fundamental limitation of existing SR models is their lack of directionality, which restricts their efficacy in extracting anatomical details from different imaging modalities. Complex sparse representation (ComSR) addresses this by independently representing multi-dimensional signals over directional dictionaries along specific directions, allowing precise analysis of intricate details. This approach provides a unified framework for both 2D and 3D fusion tasks, which is important because most existing methods focus on either 2D or 3D problems. Experimental results across six multimodal fusion tasks involving 93 pairs of 2D source images and 20 pairs of 3D source images demonstrated superiority over 11 state-of-the-art 2D fusion methods and 4 representative 3D fusion methods in both visual quality and objective evaluation.

Deep Learning Approaches

Deep learning has transformed medical image fusion by enabling end-to-end learning of fusion rules from data. Generative adversarial networks (GANs) have been applied to fuse CT and high-resolution isotropic 3D T1-Gd MRI image sequences to generate an image with CT bone structure and MRI soft-tissue contrast. The MedFusionGAN approach uses one generator network and one discriminator network trained in an adversarial scenario, with content, style, and L1 losses used to preserve texture and structure information. This method successfully generates fused images with MRI soft-tissue and CT bone contrast, outperforming both traditional and deep learning methods on six out of nine quantitative metrics while achieving the highest spatial resolution without adding image artifacts.

Attention mechanisms and multi-scale architectures have further advanced fusion quality. A multibranch and multiscale neural network based on semantic perception uses unsupervised image segmentation to extract semantic information that is integrated into the fusion process. Multiple attention mechanisms selectively emphasize important features and integrate them across different modalities and scales. A joint loss function combining content loss, structural similarity loss, and semantic loss guides the network in preserving image brightness and texture while maintaining semantic relevance.

Fourier Domain Fusion

Fourier domain fusion offers an intuitive methodology for combining and visualizing relevant diagnostic features from multiple signals. In X-ray phase-contrast breast imaging, a Fourier domain fusion algorithm combines attenuation, phase shift, and scattering information into a single image. This approach minimizes the noise component while maintaining visual similarity to a conventional X-ray image, with noticeable enhancement in diagnostic features, details, and resolution. Radiologists experienced in mammography validated that the combination of all relevant diagnostic features present in the input images was also present in the fused image.

Clinical Applications Across Medical Domains

Oncology and Tumor Imaging

PET-CT fusion has become standard practice for lung cancer diagnosis, staging, and metastasis detection. The combination of PET and CT imaging advantages while minimizing their respective limitations enables simultaneous assessment of anatomical tumor extent and metabolic activity. However, the registration of two different modalities remains time-consuming and technically challenging in clinical settings. Deep learning techniques are being explored to automate the fusion procedure with better image quality while preserving essential clinical information.

For brain tumor analysis, AI-based approaches using machine learning and deep learning algorithms have performed credibly in tumor detection, segmentation, and classification tasks. The major challenges before clinical application include dataset bias, limited generalization, lack of explainability, and high computational costs. A three-factor taxonomy covering diagnostic tasks, learning strategies, and data modalities helps organize the landscape of AI methods applied to brain tumor imaging, with a focus on adult diffuse gliomas and secondary coverage of brain metastases, meningiomas, and pediatric tumors.

Image-Guided Interventions

Intraprocedural image fusion has demonstrated measurable clinical benefit in interventional radiology. A single-center comparative analysis of CT-guided thermal ablation of liver tumors found that using intraprocedural CT-CT deformable image fusion to visually assess applicator placement before ablation and ablation completeness after treatment was associated with improved local tumor progression-free survival. Two-year local tumor progression-free survival was significantly improved with fusion for both hepatocellular carcinoma and colorectal liver metastases. On univariate regression analysis, the use of image fusion was a predominant factor significantly associated with improved outcomes in both patient groups.

Real-time fusion of ultrasound with PET-CT has been applied to guide thermal ablation of PET-positive liver metastases, with contrast enhancement adding further value to the procedure. This approach combines the real-time capabilities of ultrasound with the metabolic information from PET to improve targeting of metabolically active tumor tissue.

Cardiovascular Imaging

Three-dimensional fusion display of CT coronary angiography and myocardial perfusion combines anatomical coronary information with functional perfusion data. This approach supports the assessment of whether coronary stenoses identified on CT angiography correspond to clinically significant perfusion defects.

Deep learning and AI in cardiovascular imaging offer tools for automatic segmentation, quantification of changes, and risk stratification. Multimodal approaches have improved diagnostic accuracy and reproducibility in cardiac imaging, myocardial perfusion assessment, valve defect detection, and coronary event prediction. Explainable AI enhances transparency and clinical confidence, while deep learning enables faster image acquisition and processing without compromising precision.

Neurological Imaging

Stroke lesion segmentation in MRI is impeded by heterogeneous lesion morphology and the computational expense of modeling global context in three-dimensional data. Diffusion-weighted imaging, apparent diffusion coefficient, T2-weighted imaging, and T2-star sequences each offer complementary information for cerebral infarction assessment. Adaptive multimodal fusion approaches that employ dynamic cross-attention to spatially weight and merge signals from all four MRI sequences have achieved superior results compared to other state-of-the-art algorithms. Interpretability analysis confirms that model attention corresponds to true ischemic areas across modalities, offering visual evidence of diagnostic reliability.

For rotator cuff tear classification, a hierarchical sequence-aware multimodal framework demonstrated that instead of indiscriminately aggregating entire clinical protocols, multimodal fusion is optimized by selecting complementary imaging series. A streamlined three-sequence subset with clinical metadata achieved peak performance, outperforming the full four-sequence protocol. Metadata utility was configuration-dependent, assisting only fluid-sensitive combinations.

Venous Interventions

Intraoperative imaging for venous pathologies has become increasingly important with the evolution of dedicated venous stents and novel interventional devices. Most venous interventions are performed using a combination of standard 2D fluoroscopy, digital-subtraction angiography, and intravascular ultrasound. Latest generation CT and MRI scanners provide high-resolution 3D and 4D information about venous vasculature. Novel MRI techniques such as 3D time-resolved MR venography and 4D flow sequences provide quantitative information and help visualize intricate flow patterns. Image fusion integrates high-fidelity information from multiple imaging techniques to overcome the limitations of current intraoperative imaging. For example, the limitations of standard 2D fluoroscopy and luminal angiography can be compensated for by perivascular and soft-tissue information from MRI during complex venous interventions.

Practical Workflow for Implementing Image Fusion

Step 1: Define the Clinical Question

Before selecting a fusion method, specify what diagnostic or therapeutic decision the fused image will support. Different clinical questions require different fusion strategies. Tumor staging may require PET-CT fusion to combine metabolic and anatomical information. Radiotherapy planning may require CT-MRI fusion to combine bone definition with soft-tissue contrast. Intraprocedural guidance may require real-time fusion of ultrasound with pre-acquired CT or PET data. The clinical question determines which modalities are needed and what information must be preserved in the fused output.

Step 2: Assess Modality Compatibility

Evaluate whether the available imaging modalities provide complementary or redundant information for the clinical question. Complementary modalities such as CT and MRI provide different tissue properties that can be meaningfully combined. Redundant modalities may not justify the additional complexity and computation time of fusion. Also assess spatial resolution differences, field of view coverage, and whether the patient anatomy is likely to change between acquisitions due to motion, positioning, or treatment effects.

Step 3: Select the Fusion Approach

Choose the fusion method based on the modality pair, the clinical scenario, and the available computational resources. Classical methods such as PCA and wavelet transforms are well understood and computationally efficient but may not capture complex relationships between modalities. Sparse representation methods offer improved directional analysis for multi-dimensional data. Deep learning methods can learn optimal fusion rules from data but require substantial training datasets and computational resources. Consider whether the method has been validated on the specific modality pair and clinical application you are addressing.

Step 4: Implement Registration and Preprocessing

Registration is the critical prerequisite for meaningful fusion. Select the appropriate registration model based on whether the anatomy is expected to remain rigid or deform between acquisitions. Preprocess images to normalize intensity ranges, correct for artifacts, and resample to a common voxel size. Document all preprocessing parameters to ensure reproducibility.

Step 5: Validate Fusion Quality

Evaluate the fused image using both qualitative and quantitative measures. Qualitative assessment by experienced readers is essential for clinical acceptance. Quantitative metrics should assess structural similarity, contrast preservation, distortion level, and edge preservation. Multiple metrics are needed because no single metric captures all aspects of fusion quality. Compare the fused result against each input image to verify that clinically relevant features from both modalities are preserved.

Step 6: Document and Report

Maintain detailed records of the fusion workflow, including software versions, registration parameters, fusion algorithm settings, and quality metrics. This documentation supports reproducibility and enables comparison across patients and studies. When fusion is used in research, report the specific methods and parameters so that other investigators can replicate the analysis.

Records and Measurements for Fusion Quality

Quantitative Metrics

Nine quantitative metrics are commonly used to quantify the preservation of structural similarity, contrast, distortion level, and image edges in fused images. Structural similarity measures how well the fused image preserves the structural information from the source images. Contrast metrics assess whether the fused image maintains the contrast differences visible in the inputs. Distortion metrics quantify artifacts introduced by the fusion process. Edge preservation metrics evaluate whether boundaries and fine details survive the fusion.

Qualitative Assessment

Radiologist evaluation of feature content in each input and the fused image is essential for clinical validation. In the Fourier domain fusion study for breast imaging, radiologists experienced in mammography applied the fusion method to mastectomy samples and evaluated the feature content of each input and the fused image. This assessment validated that the combination of all relevant diagnostic features present in the input images was also present in the fused image.

Reproducibility Records

For research applications, document the complete workflow from image loading through processing to final output. Workflow generators such as RADIUMA support reproducible analysis by allowing users to construct workflows that begin with loading multi-modality DICOM images and continue through processing, diagnosis, and prognostic modeling. Executable and graphical workflows can be saved, modified, or extended later and shared with other researchers, allowing reproducible analysis using the same parameters and settings across datasets.

Common Failure Patterns in Image Fusion

Registration Errors

The most common cause of poor fusion results is inaccurate registration. When anatomy shifts between acquisitions due to patient motion, respiratory motion, or treatment effects, rigid registration produces misaligned fused images. Deformable registration can compensate for some tissue shifts but may introduce unrealistic distortions if the deformation model is too flexible. Always visually inspect registration results before proceeding with fusion.

Modality Imbalance

In multimodal analysis, different modalities often present unique challenges and characteristics that lead to imbalances within datasets. This significantly impedes model training and generalization due to varying convergence rates of different modalities and suppression of gradients in less dominant modalities. Fusion approaches that implement modality-specific balancing factors can mitigate this problem by generating balanced modality pairs based on image orientations of different modalities.

Loss of Fine Detail

Traditional fusion methods often suffer from poor image quality and loss of crucial details due to inadequate handling of semantic information and limited feature extraction capabilities. This is particularly problematic when fusing modalities with very different spatial resolutions. Deep learning approaches with attention mechanisms and multi-scale architectures are designed to address this limitation by capturing a comprehensive range of image details and contextual information.

Artifact Introduction

Fusion algorithms can introduce artifacts that were not present in either input image. These artifacts may appear as blurring, ringing, or intensity distortions that could be misinterpreted as pathology. The MedFusionGAN approach was noted for fusing source images with the highest spatial resolution without adding image artifacts, highlighting that artifact-free fusion is an important quality criterion.

Overfitting in Deep Learning Approaches

Deep learning fusion methods trained on limited datasets may not generalize to different imaging protocols, patient populations, or disease types. The GLIS-RT dataset used for MedFusionGAN included different imaging protocols and patients with various brain tumor types to improve model generalization. When applying deep learning fusion methods, verify that the training data is representative of the intended clinical application.

Limitations and Interpretation Boundaries

Validation Status

Many fusion techniques are validated in research settings but lack comprehensive clinical utility evidence. A review of AI methods for brain tumor analysis acknowledged that comprehensive clinical utility evidence remains an open research direction. Similarly, AI and deep learning in cardiology require validation in prospective, multicenter studies before widespread clinical adoption. Distinguish between methods with established clinical use and those that are promising but not yet clinically validated.

Computational Requirements

Deep learning fusion methods require substantial computational resources for training and inference. High computational costs are identified as a barrier to clinical application. Consider whether the available infrastructure can support the computational demands of the chosen fusion approach, particularly for real-time or intraoperative applications.

Explainability

Lack of explainability is a significant barrier to clinical adoption of deep learning fusion methods. Clinicians need to understand why a fusion algorithm produced a particular result, especially when the fused image influences treatment decisions. Interpretability tools such as Grad-CAM can provide visual evidence that model attention corresponds to clinically relevant regions, enhancing transparency and clinical confidence.

Data Sharing and Reproducibility

Medical image fusion research often involves sensitive patient data. The NIH Genomic Data Sharing Policy provides a framework for responsible data sharing that balances the benefits of data sharing with the need to protect participant privacy. The FAIR Guiding Principles emphasize that data should be findable, accessible, interoperable, and reusable. When developing fusion workflows, consider how data sharing and reproducibility requirements affect the choice of methods and documentation practices.

Safety and Regulatory Context

Clinical Decision Support

AI-based fusion methods are positioned as clinical decision-support tools that assist radiologists in improving diagnostic accuracy and supporting the interpretation of neuroimaging data. They are not intended to replace human interpretation. The role of fusion is to provide supplementary information that supports more accurate clinical decision-making, not to make autonomous diagnostic decisions.

Intraprocedural Use

When fusion is used to guide interventions, the accuracy of the fused image directly affects patient safety. In thermal ablation procedures, fusion is used to assess applicator placement before ablation and ablation completeness after treatment. Incorrect fusion could lead to incomplete tumor ablation or damage to surrounding healthy tissue. Verify fusion accuracy before relying on it for procedural decisions.

Professional Escalation Criteria

Fusion results that are inconsistent with clinical expectations should be escalated to the responsible clinician. Specific escalation criteria include: fused images showing apparent anatomical distortion that cannot be explained by known pathology, registration failures that cannot be corrected with available tools, fusion artifacts that obscure clinically relevant structures, and quantitative quality metrics that fall below acceptable thresholds. In these cases, the fusion result should not be used for clinical decision-making until the issue is resolved.

Frequently Asked Questions

What is the difference between image registration and image fusion?

Registration is the spatial alignment of two or more images so that corresponding anatomical structures overlap. Fusion is the combination of aligned images into a single composite image. Registration is a prerequisite for fusion, and the accuracy of fusion depends directly on the accuracy of registration. Poor registration produces misleading fusion results regardless of the sophistication of the fusion algorithm.

Which fusion method should I choose for PET-CT images?

PET-CT fusion typically uses display overlay or hardware co-registration to combine metabolic activity from PET with anatomical localization from CT. This approach is standard for lung cancer staging and metastasis detection. The main challenge is the time-consuming and technically demanding registration of the two modalities. Deep learning techniques are being explored to automate the fusion procedure while preserving essential clinical information.

How do deep learning fusion methods compare to traditional methods?

Deep learning methods can learn optimal fusion rules from data and often outperform traditional methods on quantitative metrics. The MedFusionGAN approach outperformed both traditional and deep learning methods on six out of nine quantitative metrics while achieving the highest spatial resolution without adding image artifacts. However, deep learning methods require substantial training data and computational resources, and their lack of explainability can be a barrier to clinical adoption.

What quality metrics should I use to evaluate fused images?

No single metric captures all aspects of fusion quality. Use multiple metrics that assess structural similarity, contrast preservation, distortion level, and edge preservation. Nine quantitative metrics are commonly reported for this purpose. Qualitative assessment by experienced readers is also essential for clinical acceptance, particularly to verify that clinically relevant features from both modalities are preserved in the fused image.

Can image fusion be used for real-time procedural guidance?

Yes, real-time fusion of ultrasound with pre-acquired CT or PET data has been applied to guide thermal ablation of liver metastases. Intraprocedural CT-CT deformable image fusion has been shown to improve local tumor progression-free survival after thermal ablation of liver tumors. These applications require fast registration and fusion algorithms that can operate within the time constraints of an interventional procedure.

What are the main barriers to clinical adoption of fusion methods?

The main barriers include dataset bias, limited generalization, lack of explainability, and high computational costs. Many fusion techniques are validated in research settings but lack comprehensive clinical utility evidence from prospective, multicenter studies. Registration remains time-consuming and technically challenging in clinical settings. These factors must be addressed before fusion methods can be widely adopted in routine clinical practice.

How should I handle missing modalities in a fusion workflow?

Missing modalities are a common challenge in clinical practice. Adaptive multimodal fusion approaches that employ dynamic cross-attention can spatially weight and merge signals from available modalities while adapting to missing inputs. Some frameworks specifically address missing-modality robustness. When a required modality is unavailable, document the limitation and consider whether the remaining modalities provide sufficient information for the clinical question.

What documentation should I maintain for reproducible fusion analysis?

Document the complete workflow including software versions, registration parameters, fusion algorithm settings, preprocessing steps, and quality metrics. Workflow generators that support saving, modifying, and sharing executable workflows enable reproducible analysis using the same parameters and settings across datasets. This documentation supports both research reproducibility and clinical quality assurance.

Related Bioinformatics Guides

• Epigenetics and Computational DNA Methylation Analysis: Mechanisms, Methods, and Veterinary Applications
• In Silico Design of Peptide-Based Viral Entry Inhibitors Targeting Class I Fusion Proteins
• Cryo-EM Image Processing and 3D Reconstruction
• Multi-Omics Integration Strategies
• Computational Strategies in Structure Based Drug Design

References and Further Reading

• EMBL-EBI Training. European Bioinformatics Institute.
• NCBI Data Resources. National Center for Biotechnology Information.
• Genomic Data Sharing Policy. National Institutes of Health.
• The FAIR Guiding Principles. Scientific Data.
• Multi-Dimensional Medical Image Fusion With Complex Sparse Representation.. IEEE transactions on bio-medical engineering, 2024.
• MedFusionGAN: multimodal medical image fusion using an unsupervised deep generative adversarial network.. BMC medical imaging, 2023.
• Intraoperative Imaging and Image Fusion for Venous Interventions.. Methodist DeBakey cardiovascular journal, 2018.
• A multibranch and multiscale neural network based on semantic perception for multimodal medical image fusion.. Scientific reports, 2024.
• [Medical Image Displaying Techniques for Clinical].. Igaku butsuri : Nihon Igaku Butsuri Gakkai kikanshi = Japanese journal of medical physics : an official journal of Japan Society of Medical Physics, 2015.
• Enhancing multimodal medical image analysis with Slice-Fusion: A novel fusion approach to address modality imbalance.. Computer methods and programs in biomedicine, 2025.
• Fourier domain image fusion for differential X-ray phase-contrast breast imaging.. European journal of radiology, 2017.
• Improved medical image fusion based on cascaded PCA and shift invariant wavelet transforms.. International journal of computer assisted radiology and surgery, 2018.
• Hierarchical Multimodal Fusion of Multi-Sequence MRI and Clinical Metadata for the Classification of Rotator Cuff Tears.. 2026.
• A Comprehensive Review of Artificial Intelligence for Brain Tumor Analysis: Taxonomy, Robustness, and Open Challenges in Neuro-Oncology.. 2026.
• Deep Learning and Cardiovascular Diseases: An Updated Narrative Review.. 2026.
• A State-of-the-Art Review on Multimodal Deep Learning for Medical Diagnosis: Integrating Text and Image. 2026.
• Artificial Intelligence-Assisted Pathogen Detection: Algorithms, Biosensing Platforms, and Applications.. 2026.
• The role of multimodality in clinical disease diagnosis: advances, challenges, and opportunities.. 2026.
• Adaptive multimodal fusion via Gated Parallel Mamba architecture for ultra-high-precision stroke lesion segmentation in medical imaging.. 2026.
• CT-guided Thermal Ablation of Liver Tumors Using Intraprocedural CT-CT Fusion for Applicator Position and Ablation Completeness Assessment: a Single-Center Comparative Analysis. Cardiovascular and Interventional Radiology, 2025.
• RADIUMA: A Shareable Executable Workflow Generator for Medical Image Analysis. 2025 IEEE Nuclear Science Symposium (NSS), Medical Imaging Conference (MIC) and Room Temperature Semiconductor Detector Conference (RTSD), 2025.
• [Exploration and clinical application of the "digital and intelligent surgery" diagnosis and treatment workflow for oral and maxillofacial tumors].. Beijing da xue xue bao. Yi xue ban = Journal of Peking University. Health sciences, 2026.
• Three-Dimensional Fusion Display of CT Coronary Angiography and Myocardial Perfusion. 2015.
• Multimodal Classification on PET/CT Image Fusion for Lung Cancer: A Comprehensive Survey. ECS Transactions, 2022.
• Real-Time US-18FDG-PET/CT Image Fusion for Guidance of Thermal Ablation of 18FDG-PET-Positive Liver Metastases: The Added Value of Contrast Enhancement. Cardiovascular and Interventional Radiology, 2018.
• Advanced Image Fusion Techniques for Medical Imaging: Trends, Applications, and Future Directions. Advanced Image Fusion Techniques for Medical Imaging Trends Applications and Future Directions, 2025.
• A Comparative Review of Multimodal Fusion Techniques for Medical Image Diagnosis. 9th International Conference on Engineering and Emerging Technology Iceet 2023, 2023.
• Medical Image Processing: A Multimodal Fusion Technique. Mathematical Methods in the Digital Age, 2024.
• A review on multimodal medical image fusion: Compendious analysis of medical modalities, multimodal databases, fusion techniques and quality metrics. Computers in Biology and Medicine, 2022.
• Development of Multimodal Fusion Technique for Medical Images. 2nd IEEE International Conference on Advanced Technologies in Intelligent Control Environment Computing and Communication Engineering Icatiece 2022, 2022.

This article is educational and does not replace validated analysis plans, institutional policy, clinical interpretation, or specialist review.