(IJCST-V5I1P12) :ms. Trupti B. Pawal, Mr. S. N. Patil

International Journal of Computer Science Trends and Technology (IJCST) Volume 5 Issue 1, Jan Feb 2017
RESEARCH ARTICLE OPEN ACCESS
3-D Discrete Shearlet Transform Based MRI, CT and PET Image

Fusion: a Review
Ms. Trupti B. Pawal, Mr. S. N. Patil
Electronics department, PVPIT Budhgaon
India
ABSTRACT
Traditional two dimensional (2-D) fusion method suffers from the loss of the between slice information of the third dimension. The
popularly used average- maximum fusion rule can capture only the local information but not any of the global information. In this
paper, a 3D Discrete Shearlet Transform is proposed for 3D medical image fusion and to overcome the drawbacks of previous
methods. A parallel study is done among wavelet, contourlet transform and 3D discrete shearlet transform. The experimental results
demonstrate that better fusion results can be obtained by the proposed method.
Keywords:- Wavelet transform, contourlet transform, 3D discrete shearlet transform (3DDST)
I. INTRODUCTION As compared with source images, the fused images are more
suitable for perception and comprehension. Frequently, the
Image fusion has become a common term used within multiple sensor data are excess and complementary. The fusion
medical diagnostics and treatment. Here fusion occurs when of such data reduces the uncertainty and increases the exactness.
multiple images of patient are registered and merged to provide Fourier transform is unsuccessful to analyze a non-stationary
additional information. Due to the great need in practical signal. The MRI and CT images can be processed by utilizing
applications, different fusion technologies have been developed the wavelet transform. Wavelets in 2-D are good at eliminating
in recent years. The multimodal medical image fusion is a noise for efficient analysis and isolating discontinuities at the
promising research in area of biomedical imaging system. Using edges, however wont get the smoothness along the contours and
medical images from multiple biomedical image sensors, wont hold directionality and anisotropy. Contourlet transform
increases robustness and enhances accuracy in medical research mostly utilized for the reduction and denoising of images. But
and clinical diagnosis. However medical image fusion can be the contourlet transform has more computational complexity. A
broadly defined as the combination of visual information proper continuum theory is missing in this approach [2].
contained in any number of input medical images in a single
fused image without introducing distortion and information loss Most of the methods are implemented in two dimensional (2-
[4]. For example, the combination of the positron emission D) space. Due to the loss of between slice information the
tomography (PET) and computed tomography (CT) images can results are not of the same quality as those of three dimensional
be used to concurrently view the tumor activity by visualizing (3-D) methods. The traditional 3D image fusion methods suffer
the anatomical from bad image representations. The edges and contours in the
and physiological characteristics in oncology. The images cannot be well represented by these methods. The
fusion of CT and magnetic resonance imaging (MRI) is helpful average maximum fusion rule can capture only the local
for the neuronevigation in skull base tumor surgery. The information but not any of the global information [1].
combination of PET and MRI is useful for the hepatic metastasis
[1]. The different fusion methods can be generally classified into II. LITERATURE SURVEY
three levels: pixel level, feature level and decision level. Medical
Multimodal medical volumetric data fusion using 3-D
image fusion usually employs the techniques at the pixel level. discrete shearlet transform and global to local rule discussed by
According to whether multiscale decomposition (MSD) is Lei Wang, Bin Li, Lianfang Tian. A 3D discrete shearlet
applied, the pixel level fusion methods can be roughly classified transform is proposed to overcome the drawbacks of previous
into MSD based and non- MSD based methods. methods. A global to local fusion rule is proposed to get not only
local but also global information [1].
ISSN: 2347-8578 www.ijcstjournal.org Page 70

G. Thamarai Selvi and R.K. Duraisamy discussed a novel 3D In shearlet transform, shearlet function is parametreized by
digital shearlet transform (3 D D S T) based image fusion scaling, shear and translation parameter. With the help of these
technique using MR and CT Images for brain tumor detection. parameters discrete shearlet transform provides a multiresolution
They have discussed a novel 3 D D S T for image sequence analysis useful for the development of fast algorithmic
fusion and denoising. If MR and CT images contain noise, the implemetations. Shearlets have properties such as they are well
noises may be transferred into the fusion image pixels [2]. localized, in fact, they are compactly supported in the frequency
Biswajit Biswas, Somoballi Ghoshal, Amlan Chakrabarti domain and have fast decay in the spatial domain, Shearlets
discussed a medical image fusion by combining SVD and satisfy parabolic scaling, Shearlets exhibit highly directional
shearlet transform where they first decomposed the source image sensitivity, they are spatially localized [6].
at second level by shearlet transform and used SVD method on The 3D discrete shearlet transform can be depicted as the
those individual decomposition levels to select higher singular cascade of multi-scale decomposition, based on the Laplacian
value from the shearlet image to preserve the effective pyramid filter, emulated by a phase of directional filtering [1].
information [3]. The major innovation of 3DDST is to be denoised the images by
Biswajit Biswas, Amlan Chakrabarti, and Kashi Nath Dey utilizing the direction filtering. The directional filtering design
discussed in medical image fusion using regional statistics of endeavors to reproduce the frequency decomposition provided
shift-invariant shearlet domain a new scheme of medical image by utilizing a process based on the pseudo-spherical Fourier
fusion based on shearlet transform. They have compared their transform. Hence, reduces the computation complexity of the
technique with the other methods to prove that the shearlet 3DDST and improves the visual quality. The shearlet
transform has a good performance in regards to image fusion methodology function declares at different scales and locations
[4]. and according to different orthogonal transformations controlled
A medical image fusion scheme based on shearlet transform and by shearing matrices [2].
human feature visibility discussed by Nemir Ahmed Al-Azzawi.
The image fusion algorithm is an effective, efficient, and CT PET/MRI
feasible. The visual and statistical comparisons demonstrate that Image Image
the discussed method can preserve the important structure
information. This quality makes the discussed method
promising applications in medical diagnosis [5]. Multiscale Decomposition (3DDST)
Shiji T.P, Rekha Lakshmanan, Vinu Thomas and Suma Mariam
Jacob, Thara P discussed a despeckling of breast ultrasound
images using level adaptive thresholding of discrete shearlet
transform coefficients. The shearlet transform is very effective
in removing speckle noise even with a simple thresholding High pass Subbands Low pass Subbands
scheme. The proposed method produces better quality
ultrasound images for subsequent computer-assisted image
analysis by medical experts [6].
III. METHODOLOGY Fused Subband
3-D Discrete Shearlet Transform (3DDST): Shearlets (Global to Local Fusion Rule)
emerged in recent years among the most successful frameworks
for the efficient representation of multidimensional data. Indeed, Inverse Multiscale Decomposition
many other transforms were introduced to overcome the
limitation of traditional multiscale transforms due to their poor
ability of capturing edges and other anisotropic features. Fused Image
However, shearlet transform stands out since it has many
advantages uniquely. It has a single or finite set of generating
functions; it provides optimally sparse representations for In this framework, the source images are firstly decomposed
multidimensional data, it allows a unified treatment of the into different levels and different directions in each level. That is
continuum and digital realms. With these advantages, shearlet they are decomposed into highpass subbands and lowpass
transform has been widely utilized in many image processing subbands. Then,
tasks such as denoising, edge detection , and enhancement.

the low-pass subbands and high-pass subbands are combined

under the fusion rules to get the fused subband. Finally, the
fused results are obtained by the inversion of the corresponding
MSD tool. Thus, the fusion performance is highly determined by
the MSD tools and the fusion rules [1].
The contourlet transform, wavelet transform are two most of
the popular MSD tools in image fusion. These methods,
however, often produce undesirable side effects in the final
fusion results, the reduced contrast, which may result in the
wrong diagnosis. The reason is that wavelet-like tools
decompose the source images into only three high-pass
subbands, and the limited high-pass subbands result in that
wavelets cannot well represent the sharp image features. As one
of the state-of-the-art MSD tools, the shearlet transform has been
reported to be the better MSD tool than the discrete wavelet
transform in image fusion for it decomposes the source images
into more than the wavelet-like vertical, horizontal and diagonal
high-pass subbands. Therefore, more directional information can
be captured. In addition, compared with the curvelet transform, Fig 2: Fusion results of MRI and PET images using 3D discrete shearlet
transform a) five slices of MRI b) five slices of PET c) Fusion result of MRI and
contourlet transform, which have been successfully introduced PET
into medical image fusion in the shearlet transform has better
mathematical properties. For example, different from the From fig. 1 and 2 we can observe a 3D discrete shearlet
contourlet transform, the number of directions for shearing the transform has better results than previous methods.
images is not restricted. Furthermore, compared with the
inversion of the contourlet transform, the implementation of the
inversion of the shearlet transform is more efficient V. CONCLUSION
computationally, more details can be found.
This paper proposes a novel 3D discrete shearlet transform
IV. EXPERIMENTAL RESULTS for medical image fusion, denoising and to overcome the
drawbacks of wavelet and contourlet transform. The proposed
3DDST can be more effective, efficient and feasible for the
purpose of denoising MRI, CT and PET images. This method
can preserve the important information from source image.
REFERENCES
[1] Lei Wang, Bin Li, Lianfang Tian Multimodal
Medical volumetric data fusion using 3-D discrete
shearlet transform and global to local rule, IEEE
Transactionson Biomedical Engineering. Vol. 61, NO, 1,
January 2014.
[2] G. Thamarai Selvi and R.K. Duraisamy, A novel 3-D

digital shearlet transform based image fusion technique
using MR and CT images for brain tumor detection
Middle-East Journal of Scientific Research 22 (2): 255-
260, 2014, ISSN 1990-9233
Fig.1 MRI brain images a) MRI T1 b) MRI T2 c) MRI Pd. Fig. j, f, n shows
fusion results of T1 & T2, T1 & Pd, T2 & Pd respectively using 3D discrete
shearlet transform. [3] Biswajit Biswas, Somoballi Ghoshal, Amlan
Chakrabarti,Medical image fusion by combining SVD
and shearlet transform, 2nd International Conference on
Signal Processing and Integrated Networks (SPIN), 2015.

[4] Biswajit Biswas, Amlan Chakrabarti, and Kashi Nath Dey, Journal of Computer Applications (975-8887) Volume
Medical image fusion using regional statistics of shift- 125 No.12, September 2015.
invariant shearlet domain, IEEE Conference on
Biomedical Engineering and Sciences, 8 - 10 December [6] Shiji T.P, Rekha Lakshmanan, Vinu Thomas and Suma
2014. Mariam Jacob, Thara P Despeckling of Breast
Ultrasound Images Using Level Adaptive Thresholding
[5] Nemir Ahmed Al-Azzawi, Medical image fusion based of Discrete Shearlet Transform Coefficients , The
on shearlets and human feature visibility, International International Conference on Information Science 2014
(ICIS'14)

(IJCST-V5I1P12) :ms. Trupti B. Pawal, Mr. S. N. Patil

Uploaded by

Copyright:

Available Formats

(IJCST-V5I1P12) :ms. Trupti B. Pawal, Mr. S. N. Patil

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(IJCST-V5I1P12) :ms. Trupti B. Pawal, Mr. S. N. Patil

Uploaded by

Copyright:

Available Formats

International Journal of Computer Science Trends and Technology (IJCST) Volume 5 Issue 1, Jan Feb 2017

RESEARCH ARTICLE OPEN ACCESS

3-D Discrete Shearlet Transform Based MRI, CT and PET Image

ISSN: 2347-8578 www.ijcstjournal.org Page 70

III. METHODOLOGY Fused Subband

ISSN: 2347-8578 www.ijcstjournal.org Page 71

the low-pass subbands and high-pass subbands are combined

[2] G. Thamarai Selvi and R.K. Duraisamy, A novel 3-D

ISSN: 2347-8578 www.ijcstjournal.org Page 72

ISSN: 2347-8578 www.ijcstjournal.org Page 73

You might also like