A Comprehensive Analysis of Medical Image Fusion Techniques:

A Detailed Review

T. M. Hayat

1,*

and Sai Madhavi D.

2,†

Ballari Institute of Technology and Management, Ballari, Visvesvaraya Technological University, Belagavi-590018, India

RaoBahadur Y Mahabaleshwarappa Engineering College, Ballari, Visvesvaraya Technological University,

Belagavi-590018, India

Keywords: CT, Image Fusion, Image Processing, MRI.

Abstract: Image fusion involves merging a collection of images of the same scene to create a single composite image.

Its purpose is to generate a more visually appealing image or to extract additional valuable information from

it. The main aim of image fusion is to produce a new image that contains high-quality data, which cannot be

obtained through other means. This process combines multisensor, multiview, and multitemporal data to

create a single, comprehensive image. Image fusion techniques have been applied in various fields, including

remote sensing, astronomy, and medical imaging. In medical imaging, image fusion has been particularly

useful for simultaneously evaluating CT, MRI, and PET images to find what type of disease or its effect. In

this paper, we present a novel literature review on image fusion techniques applied to medical images. Our

findings suggest that image fusion can greatly improve the clinical reliability of disease diagnosis and

analysis, and we anticipate strong growth in this field in the near future.

1 INTRODUCTION

Image fusion is a progression method of merging a set

of images of the same scene into one composite

image. Fusioning is done in order to get an enhanced

image or to enhance some useful information from it.

Image fusion in medical field has seen significant

growth several years and it incorporates a broad range

of techniques in the field of image fusioning.

The fusion process aims to address medical

conditions or diseases by analyzing images of the

human body, organs, and cells. With advancements in

computer-aided imaging techniques, this process

helps medical experts make more informed decisions

in a shorter amount of time. By using fusion methods

on multi-sensor and multi-source images, a wider

range of features can be used for medical analysis

which is leading to more precise information

processing and the ability to uncover details that may

be invisible to the human eye. Additional Information

is obtained from fused images, which lead to locate

abnormalities, more accurately. Image filters are

indeed a fundamental concept in image processing,

Research Scholar

†

HOD and Prof

Figure 1: Information fusion system at all three levels of

processing.

and they are used to enhance, manipulate, or extract

information from images. There are various types of

image filters, such as spatial filters, frequency filters,

and edge detection filters, each serving a specific

purpose in image processing. In spatial domain

filtering, a filter is applied directly to the pixels of the

Hayat, T. and Madhavi D., S.

A Comprehensive Analysis of Medical Image Fusion Techniques: A Detailed Review.

DOI: 10.5220/0012603200003739

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 1st International Conference on Artiﬁcial Intelligence for Internet of Things: Accelerating Innovation in Industry and Consumer Electronics (AI4IoT 2023), pages 147-152

ISBN: 978-989-758-661-3

147

image, and it can be utilized for smooth or sharpen

images, remove noise or enhance specific features.

Spatial filters are often used for real-time image

processing applications, such as video or camera

feeds, as they are faster than frequency domain filters.

Image processing has various applications, as you

have mentioned. It is used in computer vision tasks,

such as object recognition, face detection, and motion

tracking. In security, image processing can be used

for surveillance systems and access control. It is also

used in entertainment and gaming industries, for

creating special effects or developing interactive

games. One of the most important uses of image

processing is in medical imaging. Medical images,

such as X-rays, CT scans, or MRI, are used to

diagnose and treat various diseases and conditions.

Image processing can help improve the accuracy of

medical image analysis, by enhancing the contrast

between different tissues, segmenting specific

regions of interest, or detecting abnormalities.

Figure 2: (a) MRI Image used as source image (b) CT

Image.

Deep learning models can also be trained on huge set

of database of medical images, to automatically

detect and diagnose certain conditions, such as cancer

or fractures. This can greatly improve the speed and

accuracy of medical diagnosis, and help doctors make

more informed decisions about patient care.

Medical Image Fusion.

Medical image fusion comprises of processing and

grouping of multiple images acquired from single or

multiple imaging modalities. The key aim of this

process are to increase the quality of medical images,

decrease randomness and redundancy, and increase

their clinical applicability for diagnosis and

evaluation of medical problems. Several multimodal

medical image fusion algorithms and devices have

shown great progress in improving the accuracy of

clinical decisions based on medical images. This

overview classifies the process of fusioning of

medical image research based on his three factors:

(a) commonly used image fusion methods,

(b) relevant imaging modalities;

Despite numerous open technical and scientific

challenges, medical image fusion process has shown

encouraging results in augmenting the clinical

dependability of medical images for diagnostics as

well as analysis. It is a rapidly growing scientific area

with the potential for significant advancements in the

future years The fusion of medical images controls

the noteworthy and complementary information of

various images those are retrieve from the different

sources that used for identify the diseases and better

treatment.

Figure 3: Modalities and Algorithms of image fusion

studies.

The field of medical image analysis is distributed

into six different categories as depicted below:

Table 1: Categories of Medical Image Analysis.

Categories

Description

Post-acquisition

Prior to diagnosis, images are often subjected to preprocessing techniques such as denoising and

renovation to improve their quality and make them usable.

Segmentation

The accurate diagnosis of medical images such as CT scans of abdomen or MRI scans of brain

which requires the identification and definition of important features, such as organs, within

the image. This process, known as delineation, is crucial for effective analysis and interpretation.

Registration

The process of registering or aligning captured images with a model or previous image is a

crucial requirement in computer-assisted surgery.

AI4IoT 2023 - First International Conference on Artiﬁcial Intelligence for Internet of things (AI4IOT): Accelerating Innovation in Industry

and Consumer Electronics

148

Computation

In various computer-assisted therapies, there is also a need for the calculation of physical

quantities and the execution of additional computational tasks such as fusion and compression.

Visualization

It is crucial for medical images to be displayed, so that medical professionals can accurately

diagnose diseases.

Security

Personal medical health information is highly sensitive and must be properly secured through

methods such as watermarking. This ensures that only legal users have access to the information

and that it is accurately linked to the correct medical record for the appropriate patient.

Medical image fusion can be categorized as shown

below: -

1.) Multi View Image Fusion: In this fusion

images are taken from different viewpoints but have

same modality and at the same time.

2.) Multi Modal Image Fusion: In this fusion

different sensors like CT, MRI and PET etc. are used

to collect the images (IJARIIE, n.d.). Clinical

precision are improve by using the Multimodal

medical image fusion algorithms and devices.

3.) Multi Temporal Image Fusion: In this type of

fusion, images are extracting at different times for

finding the changes between images.

4.) Multi Focus Image Fusion: In this type of

fusion images are taken from a 3D scene continually

with various focal lengths (Mishra and Bhatnagar,

2014 ).

2 LITERATURE REVIEW ON

MEDICAL IMAGES

P. James and B.V. Dasarathy (James and Dasarathy,

2014) stated that multimodal fusion of medical

images has shown significant improvement in clinical

diagnosis of disease. Fusion using Multimodel of

medical images has shown great improvement in

clinical diagnosis of disease. This review article

provides practical techniques and summarizes the

challenges in medical image fusion. In this white

paper, while there are numerous scientific challenges

and open technologies available, medical image

fusion is a highly useful technology for improving

clinical consistency, identification, and analysis.

Daniel Ruijters (Ambrosini et al., 2017) proposed

that medical scanning technology gives a wide

spectrum of valuable and harmonizing information

about a patient's physiology, anatomy and pathology,

but the optimal exploitation of this wealth of

information is a tough job.

Deron Rodrigues et al., (Rodrigues et al., 2014)

have explained that image fusion has become

important part in medical field for diagnosis or

analysis of disease. The paper describes the

introduction of image fusion methods by using

wavelet transform and the comparison between the

performance of the various types of wavelet basis

families used.

Hari Om Shanker Mishra and Smriti Bhatnagar

(Mishra and Bhatnagar, 2014) have explained that

fusing techniques are used for image enhancement in

several imaging techniques like Computed

Tomography (CT) and Magnetic Resonance Imaging

(MRI).

Hiral Rameshbhai Patel and Raviraj Chauhan

(IJARIIE, n.d.) proposed a decomposition and

reconstruction method to improve image quality. The

Discrete Ripplet Transform is an advanced

directionality and localization transform for such

edges, and the combination of DWT and DRT yields

better images than DWT.

Jan Flusser et al. (Zitova and Flusser, 2003),

stated as fusion process of various images is used in

numerous applications, such as astronomy, multi-

sensor fusion, medical imaging, military, remote

sensing, security and surveillance fields. used in these

applications. Image fusion is used in many

applications such as remote sensing and medical

fields, and this pattern is primarily used in CT and

MRI images where the more accurate the image, the

more useful the information. Many approaches have

been developed for medical image fusion.

M.D. Nandeesh and Dr. M. Meenakshi (Casey

and Damper, 2010) studied about image fusion

techniques with their performance evaluation

analysis. They used Discrete Wavelet Transform,

Curvelet Transform, Principle Component Analysis,

Stationary Wavelet Transform techniques etc.

Madhusmita Sahoo (Ambrosini et al., 2017)

explains a modern fusion method to enhance the

information content of the fused image. The

technique uses wavelet transform, maximum

selection rule, windowing technique and GLCM

based segmentation.

Mc Cassey et, al. (Casey and Damper, 2010) uses

image fusion algorithm to acquire the sincere feasible

depth-of-field in macro-photography by using typical

digital camera images. Macro photography has some

primary problems, one of the most critical is the

difficulty of insufficient lighting. Mayank Agrawal et

al. (Agrawal et al., 2010) proposed a fusion algorithm

A Comprehensive Analysis of Medical Image Fusion Techniques: A Detailed Review

149

for multispectral magnetic resonance imaging that

preserves both component and edge information and

provides better performance associated to existing

fusion algorithms.

Medha Balachandra Mule and Padmavathi N.B

(Kotian et al., n.d.) have done analysis of different

medical imaging modalities used in fusion. They have

explained and compared different image fusion

techniques using the quality metrics Peak Signal to

Noise Ratio (PSNR) and Root Mean Square Error

(RMSE).

Nayera Nahvi and Deep Mittal (Niranjan and

Patel, n.d.) have explained a new algorithm for

multimodal medical image fusion based on DWT

technique. The algorithm escalates the quality of

multimodality medical image fusion and the output

reveal the efficiency of fusioning process.

P.Ambika Priyadharsini and M.R. Mahalakshmi

(Priyadharsini et al., n.d. ) proposed that SVD is a

substitute image fusion method, which improves the

content of medical images by merging two or more

multimodal medical images.

Paul Hill et al. proposed DT-CWT (Hill et al.,

2005) techniques for image fusion in remote sensing,

robotics and medical applications. This method gives

better qualitative and quantitative output compared to

previous wavelet fusion techniques.

Periyavattam Shanmugam Gomathi and

Bhuvanesh Kalaavathi (Gomathi and Kalaavathi,

2016) states a comparative study of image fusion of

MRI and CT images based on various wavelets

transforms techniques is performed. The final fused

image is tested by using many performance metrics to

evaluate which wavelet gives the best output.A new

generation of high resolution satellite images with

less than 1 meter spatial resolution in panchromatic

mode is now available. This paper compares the

output of three different techniques to fuse the

multispectral information and panchromatic data of

Quick Bird satellite imagery.

Richa Singh et al. (Lawson et al., 2020) proposed

a fusion algorithm that uses Redundant Discrete

Wavelet Transforms to combine pairs of

multispectral magnetic resonance imaging such as

Proton Density, T2 and T1 brain images.

Helonde and Prof. M.R. Josh (Holende et al.,

2010) explained that image fusion plays an important

role in digital image reconstruction as re-processing

steps. Medical image fusion helps in easy diagnostics

and reduces the time gap between the diagnosis of the

disease and the treatment.

Dr. S. Manikanda Prabu et al. (Prabu and

Ayyasamy, 2014) used Lifting Wavelet Transform

(LWT) based three different medical image fusion

approaches and performed comparative analysis. The

IAV method was found to be more suitable for

medical image fusion than other approaches in

wavelet domain.

Walid Aribi et al. (Arabi et al., 2012), developed

new methods to evaluate the quality of medical

images based on the multi resolution fusion. These

methods are evaluated by objective technical quality.

Zhi-haiXu et al. (Jing, 2009) proposed a new

fusion algorithm based on wavelet transform by

analyzing the three fusion operators. The algorithm

was validated by CT/PET images.

Zhijun Wang et al. (Wang et al., 2005) presented

a complete outline of the General Image Fusion (GIF)

method is a useful framework for categorizing and

evaluating image fusion methods. One such method

is MRAIM, which stands for Multiresolution

Analysis-based Image Fusion using Morphological

Reconstruction and Iterative Method.

The field of Image-Guided Therapy (IGT) is

rapidly growing and has seen success with the

commercialization of advanced IGT systems by

several small companies. However, in meetings

between IGT investigators, it was determined that

there are several key areas that require collaborative

effort from the community to improve patient care.

Image fusion is an interesting field for the

researcher. Various techniques like wavelet transform

HIS and PCA based methods are proposed by many

author or researchers.

Figure 4: Fusion of MRI and CT Image a) Using Daubechies(db) b) Using Coiflets (coif)c) using Bi-orthogonal (bior) d) By

Symlets (sym) (e) By Reverse Bior (rbio) (f)By Discrete Meyer (dmey).

AI4IoT 2023 - First International Conference on Artiﬁcial Intelligence for Internet of things (AI4IOT): Accelerating Innovation in Industry

and Consumer Electronics

150

Table 2: Different Fusion Strategies.

METHOD

TYPE OF IMAGE

FUSION STRATEGIES

MORPHOL

OGY

KNOWLE

DGE

MRI, CT, ULTRA SOUND,

MAMMOGRAM, PET

MORPHOLOGY FILTERA,

LEARNING SYSTEMS,

EXPERT SYSTEMS

WAVELET

CT, PET, MRI, ULTRA

SOUND, SPECT

DISCRETE WAVELET TRANSFORMS,

STATIONARY WAVELET, MULTI-

WAVELET TRANSFORM

ANN

CT, PET, MRI, ULTRA

SOUND, MRA, SPECT

NEURAL NETWORKS, CLUSTERING

NEURAL NETWORKS

FUZZY

LOGIC

CT, PET, MRI, ULTRA

SOUND, MRA, SPECT

IMAGE FUZZIFICATION,

DEFUZZIFICATION, NEURO FUZZY

NETWORKS,

3 CONCLUSION

The field of medical diagnostics and monitoring is

rapidly advancing with the growth of latest

technologies and scientific advancements. However,

the use of medical images to aid in these processes is

not without challenges. These challenges can be

technological, scientific, and societal in nature.

One of the challenges is related to the quality of

imaging features. In order to achieve a

comprehensive understanding of a medical condition,

multiple imaging modalities are often used. However,

these modalities may produce images with different

qualities and characteristics. Image fusion techniques

can be used to improve the quality of imaging features

by integrating information from multiple modalities.

However, the key challenge in applying image

fusion algorithms to medical images is to confirm that

the medical relevance is maintained and that they aid

in achieving enhanced clinical outcomes. This

requires careful consideration of the specific medical

application, as well as the imaging techniques used.

Despite these challenges, image fusion techniques

hold great promise for improving the quality of

medical imaging and aiding in diagnostics and

monitoring of medical conditions. As such, ongoing

research in this area is critical for the advancement of

medical science and for the betterment of patient care.

REFERENCES

Wang, Z., Ziou, D., Armenakis, C., Li, D., & Li, Q. (2005).

A comparative analysis of image fusion methods. IEEE

transactions on geoscience and remote sensing, 43(6),

1391-1402.

Prabu, S. M., & Ayyasamy, S. (2014). An efficient

watermarking algorithm based on DWT and FFT

approach. International Journal on Computer Science

and Engineering, 6(6), 211.

Zitova, B., & Flusser, J. (2003). Image registration

methods: a survey. Image and vision computing,

21(11), 977-1000.

http://ijariie.com/AdminUploadPdf/Medical_Image_Fusio

n_Using_Combine_Approach_of_DWT_and_DRT_ija

riie2635.pdf

Holende et al., A special issue on biologically inspired

information fusion, Inform. Fus. 11 (1) (2010) 1.

M.C. Casey, R.I. Damper, Editorial: special issue on

biologically-inspired information fusion, Inform. Fus.

11 (1) (2010) 2–3.

Aribi, W., Khalfallah, A., Bouhlel, M. S., & Elkadri, N.

(2012, March). Evaluation of image fusion techniques

in nuclear medicine. In 2012 6th International

Conference on Sciences of Electronics, Technologies of

Information and Telecommunications (SETIT) (pp.

875-880). IEEE.

Jing, L. (2009). Scrambling analysis of ciliates (Doctoral

dissertation, Master’s thesis, University of

Saskatchewan, Saskatoon).

J. Greensmith, U. Aickelin, G. Tedesco, Information fusion

for anomaly detection with the dendritic cell algorithm,

Inform. Fus. 11 (1) (2010) 21–34.

Lawson, K. A., Sousa, C. M., Zhang, X., Kim, E., Akthar,

R., Caumanns, J. J., ... & Moffat, J. (2020). Functional

genomic landscape of cancer-intrinsic evasion of

killing by T cells. Nature, 586(7827), 120-126.

Gomathi, P. S., & Kalaavathi, B. (2016). Multimodal

medical image fusion in non-subsampled contourlet

transform domain. Circuits and Systems, 7(8), 1598-

1610.

Mishra, H. O. S., & Bhatnagar, S. (2014). MRI and CT

image fusion based on wavelet transform. International

Journal of Information and Computation Technology,

4(1), 47-52.

Hill, P. R., Bull, D. R., & Canagarajah, C. N. (2005,

September). Image fusion using a new framework for

complex wavelet transforms. In IEEE International

Conference on Image Processing 2005 (Vol. 2, pp. II-

1338). IEEE.

A Comprehensive Analysis of Medical Image Fusion Techniques: A Detailed Review

151

Priyadharsini, P. A., Mahalakshmi, M. R., & Chelliah, D.

S. A Novel NSCT Based Medical Image Fusion

Technique.

Alex Pappachen James and Belur V. Dasarathy. 2014.

Medical image fusion: A survey of the state of the art.

Inf. Fusion 19 (September, 2014), 4–19.

https://doi.org/10.1016/j.inffus.2013.12.002

Niranjan, R., & Patel, R. A Review of Image Fusion

Techniques.

Kotian, S. M., D’souza, S., & Gadagkar, A. V. A Review

on Different Techniques for Medical Image Fusion.

Agrawal, M., Tsakalides, P., & Achim, A. (2010, August).

Medical image fusion using the convolution of

meridian distributions. In 2010 Annual International

Conference of the IEEE Engineering in Medicine and

Biology (pp. 3727-3730). IEEE.

Rodrigues, D., Virani, H. A., & Kutty, S. (2014).

Multimodal image fusion techniques for medical

images using wavelets. Image, 2(3), 310-313.

Ambrosini, Pierre & Ruijters, Daniel & Niessen, W.J. &

Moelker, Adriaan & Walsum, Theo. (2017). Fully

Automatic and Real-Time Catheter Segmentation in X-

Ray Fluoroscopy.

AI4IoT 2023 - First International Conference on Artiﬁcial Intelligence for Internet of things (AI4IOT): Accelerating Innovation in Industry

and Consumer Electronics

152