301
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254 – 4143 Edición Especial Special Issue Marzo 2020
IMPLEMENTATION OF DIFFERENTIAL EVOLUTION
ALGORITHM TO PERFORM IMAGE FUSION FOR
IDENTIFYING BRAIN TUMOR
Pothiraj Sivakumar
Department of Electronics and Communication Engineering,
Kalasalingam Academy of Research and Education,
Virudhunagar District, Tamil Nadu, (India).
E-mail: siva@klu.ac.in ORCID: https://orcid.org/0000-0003-1328-8093
Subbiah Parvathy Velmurugan
Department of Electronics and Communication Engineering,
Kalasalingam Academy of Research and Education,
Virudhunagar District, Tamil Nadu, (India).
E-mail: s.p.velmurugan@klu.ac.in ORCID: https://orcid.org/0000-0002-3314-1454
Jenyfal Sampson
Department of Electronics and Communication Engineering,
Kalasalingam Academy of Research and Education,
Virudhunagar District, Tamil Nadu, (India).
E-mail: jenyfal.sampson@klu.ac.in ORCID: https://orcid.org/0000-0001-8007-3995
Recepción:
05/12/2019
Aceptación:
30/12/2019
Publicación:
23/03/2020
Citación sugerida:
Sivakumar, P., Velmurugan, S. P., y Sampson, J. (2020). Implementation of dierential evolution
algorithm to perform image fusion for identifying brain tumor. 3C Tecnología. Glosas de innovación
aplicadas a la pyme. Edición Especial, Marzo 2020, 301-311. http://doi.org/10.17993/3ctecno.2020.
specialissue4.301-311
Suggested citation:
Sivakumar, P., Velmurugan, S. P., & Sampson, J. (2020). Implementation of dierential evolution
algorithm to perform image fusion for identifying brain tumor. 3C Tecnología. Glosas de innovación
aplicadas a la pyme. Edición Especial, Marzo 2020, 301-311. http://doi.org/10.17993/3ctecno.2020.
specialissue4.301-311
302 http://doi.org/10.17993/3ctecno.2020.specialissue4.301-311
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254 – 4143 Edición Especial Special Issue Marzo 2020
ABSTRACT
Automated mechanization for curing a disease is a reliable and protuberant method. A
disease in brain can be detected by Magnetic Resonance Imaging (MRI). In this context,
image fusion is a method for creating an image by merging pertinent data from 2 or
more images. The resultant image will be highly useful than the individual input images
to retentive the vital characteristics of every image. Multiple image fusion is a signicant
method employed in image processing techniques. In this study, dierential evolution (DE)
algorithm-based image fusion has been performed with MRI and computed tomography
(CT) images. The simulation works have been carried out to evaluate the dierent quality
measurements of DE on image fusion.
KEYWORDS
De-speckling, Brain tumor detection, CT, DE, Image fusion, MRI.
303 http://doi.org/10.17993/3ctecno.2020.specialissue4.301-311
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254 – 4143 Edición Especial Special Issue Marzo 2020
1. INTRODUCTION
Brain tumours is harmful to humans, due to the atypical availability of cells inside the
brain. Brain function will be interrupted and be deadly. Benign and malignant tumors
are frequently identied. Benign tumors are not as harmful as malignant tumors, because
they can grow rapidly. Medical imaging methodologies such as MRI, CT, Ultrasound,
X-ray etc. are employed to display the internal body parts for diagnosing (Rowden, 2019).
Among them MRI is widely employed and it oers accurate brain images and cancer cells.
So, brain tumor can be detected via MRI images. This study concentrates on detection
of brain tumor through image fusion. Image fusion is a process of merging two or more
images into a single compound image that contains the information of the source images
without clamor. Multi-modular recuperative image fusion has been employed to recognize
the wounds. In biomedical image processing image fusion has got more attention in the
past decade (Daneshvar & Ghassemian, 2010; Wang, Li, & Tian, 2014). MRI and CT
images held more practical information than biomedical images. The aim of image fusion
is to obtain the information at each pixel without damaging the pixel associations of the
particular image.
In this context, previously, a complex wavelet modication for image fusion has been
proposed to attain the optimal combination using Lifting wavelet transform (LWT),
Multiwavelet transform (MWT), Stationary wavelet transform (SWT) and spatial domain
(Principal component analysis (PCA) approaches (Singh & Khare, 2014). Similarly,
undecimated wavelet has been implemented, where the image is crumbled into two
successive scrutinizing errands (Ellmauthaler et al., 2013). An aable fusion technique
using SWT and NSCT has been presented, where the input image is rotten by SWT and
NSCT. The coecients of SWT and NSCT are combined to form the fused image (Li
& Liu, 2009). A new framework has been proposed where the images considered with
SWT primarily and the overall textural topographies have been attained via gray level co-
occurrence matrix (Singh & Khare, 2014; Huang et al., 2014; Shi & Fang, 2007). Hence, a
scheme for fusing MRI and CT images using DE based Debauchee’s wavelet Transform
(DE-DWT) has been attempted in this study.
304 http://doi.org/10.17993/3ctecno.2020.specialissue4.301-311
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254 – 4143 Edición Especial Special Issue Marzo 2020
2. MATERIALS AND METHODS
As a part of image fusion, pre-handling of images have been performed using DE. DE
has been employed to create the ssion rubrics. The preprocessing steps involved in image
fusion have been illustrated in Figure 1.
Input CT
Image
Input MRI
Image
Enhanced
CT Image
Enhanced
MRI Image
Medium FilterMedium Filter
Fusion using
DE
Fused output
Image
Performance
Metrics
Figure 1. Flowchart of proposed approach.
Primarily, the informative source images such as CT and MRI images have been collected.
Subsequently, the source images have been converted into dark scale and resized. The
enhancement of quality of the images has been performed using imadjust order available in
MATLAB simulation. Commotion dismissal has been carried out by using median channel.
This is an excellent method in ejecting salt and pepper commotions of biomedical images.
It happens due to the movement of antiquities.
Performance indices such as Mean Square Error (MSE), Peak Signal to Noise Ratio (PSNR),
contrast and homogeneity have been estimated. The amount of clamor available in the
image is denoted as PSNR. It is used to indicate the obtained fused image has tumbled-
down or not. MSE value need to be low and PSNR value need to be high.
305 http://doi.org/10.17993/3ctecno.2020.specialissue4.301-311
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254 – 4143 Edición Especial Special Issue Marzo 2020
(1)
(2)
Contrast reinstates the data associated with the pixel with the adjacent pixel. It has been
calculated as follows.
(3)
Image A
Image B
Decompose into
m by n sized
blocks
Partitioned Image A
Partitioned Image B
Calculate
sharpness
Calculate
sharpness
Compare
Results
Fused Image F
Fitness evaluation
Differential Evolution
Algorithm
Optimise m and n
Mutation
Crossover
Selection
Figure 2. Flowchart for image fusion using DE.
Table 1. Best parameters of DE.
Parameter Value
Number of population 100
Maximum generation 100
Crossover probability 0.5
Scaling factor 0.9
Homogeneity has been used to estimate the intimacy of components availed in gray level
concurrence matrix (GLCM).
3. DIFFERENTIAL EVOLUTION ALGORITHM
Price and Storn introduced DE as a population-based stochastic direct search technique.
The implementation procedure of DE has been adopted from Aslantas and Toprak (2014).
306 http://doi.org/10.17993/3ctecno.2020.specialissue4.301-311
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254 – 4143 Edición Especial Special Issue Marzo 2020
The steps involved in DE based image fusion have been illustrated in Figure 2. The best
control parameters for DE have been provided in Table 1.
The performance indices such as MSE, PSNR, Contrast, Entropy and Homogeneity have
been presented in Table 2.
Table 2. Performance indices of DE on image fusion.
SET MSE PSNR Entropy Contrast Homogeneity
1 12447 5.4875 10.1457 0 1
2 10253 6.0248 11.2488 0 1
3 17305 8.1027 12.5761 0 1
4. RESULTS AND DISCUSSIONS
MRI and CT images have been fused together using DE. The ultimate objective of image
fusion is to acme the valuable data from various input images. The adaptive fuzzy clustering
rule has been employed to fragment the region of interest (ROI) to isolate the tumor from
the resultant fused image. It will group the various grade intensity segments of the fused
image. The segments with huge grade intensity are marked as the tumor, and they have
been isolated using thresholding method (Chabira, Skanderi, & Aichouche, 2013).
Figures 3 (a), (b), (c) and (d) provide the information about the CT and MRI images which
are processed for fusion. Figures 3 (a) and (b) displays the gray scale CT and MRI images
respectively. Figures 3 (c) and (d) demonstrate the median ltered CT and MRI images
respectively. DE-DWT has been involved in the fusing mechanism. Using the fusing rules,
fusing rules, the input images have been combined. Diverse levels have been xed to decide
clamor data adversity in the image. Figures 4 (a) and (b) demonstrate decomposed CT
and MRI images. Figure 5 illustrates the resultant fused image with decent idiosyncratic
enrichment. By following the DE-DST rule, least value of CT is combined with the least
level decomposed MRI image to form the fused image.
307 http://doi.org/10.17993/3ctecno.2020.specialissue4.301-311
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254 – 4143 Edición Especial Special Issue Marzo 2020
Figure 3. (a) Gray scale CT image (b) Gray scale MRI image (c) Median ltered CT image (d) Median ltered
MRI images.
Figure 4. (a) Decomposed CT image (b) Decomposed MRI image.
Figure 5. Resultant fused image.
308 http://doi.org/10.17993/3ctecno.2020.specialissue4.301-311
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254 – 4143 Edición Especial Special Issue Marzo 2020
5. CONCLUSION
A typical muscles grown in brain disturb brain activities and that is referred as brain tumor.
Biomedical image processing aims to recognize precise data through images with minimum
error. Detection of brain tumor via MRI images is not easy due to the intricacy of brain. A
pixel based image fusion procedure using DE-DWT has been proposed in this study. The
simulations have been carried out with CT and MRI images. The performance indices
such as entropy, MSE, PSNR, contrast and homogeneity imply the eectiveness of the
proposed DE-DWT approach.
309 http://doi.org/10.17993/3ctecno.2020.specialissue4.301-311
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254 – 4143 Edición Especial Special Issue Marzo 2020
REFERENCES
Aslantas, V., & Toprak, A. N. (2014). Multi focus image fusion by dierential evolution
algorithm. In Proceedings of the 11th International Conference on Informatics in Control,
Automation and Robotics - Volume 2: ICINCO, 312-317, Vienna, Austria. https://doi.
org/10.5220/0005061103120317
Chabira, B., Skanderi, T., & Aichouche, B. A. (2013). Unsupervised change detection
from multitemporal multichannel SAR images based on stationary wavelet
transform. In 7th International Workshop on the Analysis of Multi-temporal Remote Sensing
Images (Multi-Temp). https://doi.org/10.1109/Multi-Temp.2013.6866025
Daneshvar, S., & Ghassemian, H. (2010). MRI and PET image fusion by combining
IHS and retina-inspired models. Information Fusion, 11(2), 114-123. https://doi.
org/10.1016/j.inus.2009.05.003
Ellmauthaler, A., Pagliari, C. L., & da Silva, E. A. B. (2013). Multiscale Image
Fusion Using the Undecimated Wavelet Transform With Spectral Factorization and
Nonorthogonal Filter Banks. IEEE Transactions on Image Processing, 22(3), 1005-1017.
https://doi.org/10.1109/TIP.2012.2226045
Huang, P. W., Chen, C. I., Lin, P. L., Ping, C., & Hsu, L. P. (2014). PET and MRI brain
image fusion using wavelet transform with structural information adjustment and
spectral information patching. In 2014 IEEE International Symposium on Bioelectronics and
Bioinformatics (IEEE ISBB 2014), 1-4. https://doi.org/10.1109/isbb.2014.6820901
Li, Y., & Liu, G. (2009). Cooperative Fusion of Stationary Wavelet Transform and Non-
subsampled Contourlet for Multifocus Images. In Second International Symposium
on Computational Intelligence and Design, 1, 314-317. https://doi.org/10.1109/
ISCID.2009.86
Rowden, A. (2019). Types, symptoms, and treatment of a brain tumor. https://www.
medicalnewstoday.com/articles/315625.php
Sahoo, T., & Patnaik, S. (2008). Cloud Removal from Satellite Images Using Auto
Associative Neural Network and Stationary Wevlet Transform. In First International
310 http://doi.org/10.17993/3ctecno.2020.specialissue4.301-311
3C Tecnología. Glosas de innovación aplicadas a la pyme. ISSN: 2254 – 4143 Edición Especial Special Issue Marzo 2020
Conference on Emerging Trends in Engineering and Technology, 100-105, Nagpur, Maharashtra,
India. IEEE. https://doi.org/10.1109/ICETET.2008.99
Shi, H., & Fang, M. (2007). Multi-focus Color Image Fusion Based on SWT and IHS. In
Fourth International Conference on Fuzzy Systems and Knowledge Discovery (FSKD 2007), 2,
461-465. https://doi.org/10.1109/FSKD.2007.414
Singh, R., & Khare, A. (2014). Fusion of multimodal medical images using Daubechies
complex wavelet transform – A multiresolution approach. Information Fusion, 19, 49-
60. https://doi.org/10.1016/j.inus.2012.09.005
Wang, L., Li, B., & Tian, L. F. (2014). Multi-modal medical image fusion using the inter-
scale and intra-scale dependencies between image shift-invariant shearlet coecients.
Information Fusion, 19, 20-28. https://doi.org/10.1016/j.inus.2012.03.002
