Brain Tumor Diagnosis using Image De Noising with Scale Invariant Feature Transform

International Journal of Trend in Scientific Research and Development (IJTSRD)
Volume 6 Issue 7, November-December 2022 Available Online: www.ijtsrd.com e-ISSN: 2456 – 6470
@ IJTSRD | Unique Paper ID – IJTSRD52272 | Volume – 6 | Issue – 7 | November-December 2022 Page 65
Brain Tumor Diagnosis using Image
De-Noising with Scale Invariant Feature Transform
Namit Thakur1
, Dr. Sunil Phulre2
1
Research Scholar, 2
Associate Professor,
1, 2
Department of CSE, LNCT, Bhopal, Madhya Pradesh, India
ABSTRACT
It is truly challenging for specialists to distinguish mind growth at a
beginning phase. X-ray pictures are more helpless to the commotion
and other natural aggravations. Subsequently, it becomes challenging
for specialists to decide on brain tumor and their causes. Thus, we
thought of a framework in which the framework will recognize mind
growth from pictures. Here we are switching a picture over
completely to a grayscale picture. We apply channels to the picture to
eliminate commotion and other natural messes from the picture. The
framework will deal with the chosen picture utilizing preprocessing
steps. Simultaneously, various calculations are utilized to distinguish
the growth from the picture. In any case, the edges of the picture
won't be sharp in the beginning phases of cerebrum growth. So here
we are applying picture division to the picture to recognize the edges
of the pictures. We have proposed a picture division process and an
assortment of picture-separating procedures to get picture qualities.
Through this whole interaction, exactness can be moved along. This
framework is carried out in Matlab R2021a. The accuracy, Review,
F1 Score, and Precision worth of the proposed model works by
0.16%, 1.99%, 0.47%, and 0.28% for CNN Model.
KEYWORDS: Brain Tumor, classification, Segmentation, Precision,
Recall, F1 Score
How to cite this paper: Namit Thakur |
Dr. Sunil Phulre "Brain Tumor
Diagnosis using Image De-Noising with
Scale Invariant Feature Transform"
Published in
International Journal
of Trend in
Scientific Research
and Development
(ijtsrd), ISSN: 2456-
6470, Volume-6 |
Issue-7, December
2022, pp.65-71, URL:
www.ijtsrd.com/papers/ijtsrd52272.pdf
Copyright © 2022 by author (s) and
International Journal of Trend in
Scientific Research and Development
Journal. This is an
Open Access article
distributed under the
terms of the Creative Commons
Attribution License (CC BY 4.0)
(http://creativecommons.org/licenses/by/4.0)
I. INTRODUCTION
Human body is comprised of many sorts of cells.
Each sort of cell has extraordinary capacities. Most
cells in the body develop and after that partition in a
deliberate approach to frame new cells as they are
expected to keep the body solid and work
appropriately. At the point when cells lose the
capacity to control their development, they separate
time and again and with no request. The additional
phones shape a mass of tissue called a tumor. Brain
tumors are made by unusual and uncontrolled cell
segmentation in cerebrum itself. By and large, if the
development turns out to be over half, at that point
the patient will most likely be unable to recuperate.
Consequently location of brain tumor at its beginning
time with its precise determination is essential.
Distinguishing proof of tumor includes tests like CT
and MRI. X-ray assumes key part in recognizing
region, size and kind of cerebrum tumor. Structure of
Brain: Generally, human cerebrum incorporates three
noteworthy parts controls distinctive activity [3].
Figure 1: Indicate the brain structure
1. Cerebrum-The cerebrum controls getting the hang
of, considering, feelings, discourse, critical
thinking, perusing and composing. It is isolated
into right and left cerebral halves of the globe.
Muscles of left half of the body is controlled by
right cerebral sides of the equator and muscles of
right half of the body is controlled by left cerebral
halves of the globe.
IJTSRD52272

International Journal of Trend in Scientific Research and Development @ www.ijtsrd.com eISSN: 2456-6470
2. Cerebellum-The cerebellum controls
development, standing, adjust and complex
activities.
3. Brain stem-Brain stem joints the brain with spinal
rope. Brain stem controls circulatory strain, body
temperature and breathing and controls some
fundamental capacities.
MR image give definite data about human anatomical
structure and tissues. Likewise MR image is protected
contrasted with CT sweep and X-Ray Image. It
doesn't influence the human body. MR Image gives
data to promote treatment and research territory.
Figure 2: Brain MR Image.
X-ray is essentially used as a piece of the biomedical
to perceive and picture better unpretentious
components in the internal design of the body.
II. BACKGROUND
X-ray pictures are the main apparatus for early
identification of mind cancer. Growth and malignant
growth are a hurtful and shocking sickness for human
existence. In this paper a proposed framework
manages clinical X-ray for characterizing input
computerized picture into typical or unusual cancers,
likewise the kind of strange case that alludes to the
presence of mind growths is additionally analyzed
into harmless cancer or threatening growth. The
proposed mind growth order framework depends on
utilizing Filter descriptor for removing valuable X-ray
highlights for determination clinical X-ray pictures.
(Mohammed Sahib Mahdi Altaei and Sura Yarub
Kamil; 2020)
Brain is an organ that regulates all parts of the body
activities. Detection of glioma from MRI image was
an important method in medical field. In order to
better interpret the medical image segmentation is
generally done as a fundamental step for further
processing. This work proposed a segmentation
algorithm for the MRI image in which the entire work
was structured into two parts. The first section of the
proposed model involved pre-processing of the MRI
image through weiner filter that removes noise after
that extraction of skull portion took place. In second
section of the model, Bio-Geography algorithm was
applied which takes brain portion of pre-processed
input MRI image. (Ashish Kumar Dehariya, Pragya
Shukla; 2020)
Cerebrum cancer is a destructive sickness and its
grouping is a difficult errand for radiologists due to
the heterogeneous idea of the growth cells. As of late,
PC supported finding based frameworks have
guaranteed, as an assistive innovation, to analyze the
cerebrum cancer, through attractive reverberation
imaging (X-ray). (Noreen, S. Palaniappan, A.
Qayyum, I. Ahmad, M. Imran and M. Shoaib; 2020)
The ID and order of growths in the human brain from
MR pictures at a beginning phase assume a crucial
part in determination such illnesses. This work gives
the original Profound Brain network less number of
layers and less complicated in planned named U-Net
(LU-Net) for the recognition of cancers. (Hari Mohan
Rai, Kalyan Chatterjee; 2020)
III. PROBLEM IDENTIFICATION
The essential complaints of my speculation work are
as per the going with:
1. Supervised tumor detection model take an image
in a specific format, but it should be generalize.
2. Some of tumor detection model need prior
information for training, this reduces dynamic
adoption of work.
3. Noise removal steps should be improved for
increasing the detection rate.
IV. RESEARCH OBJECTIVES
1. Reduce the noise present in the image by using
median filter.
2. To study Skull part of the MRI image needs to be
perfectly segment out.
3. Identification of tumor portion from the skull
portion of the MRI image.
4. To study the Accuracy of segmented region
should be increased.
V. PROPOSED METHODOLOGY
The algorithm of the proposed work is as follows.
This method works under four phases.
A. Phase 1
1.1. Read image
In this step, we store the path to our image dataset into
a variable then we created a function to load folders
containing images into arrays.
1.2. Resize image
In this step in order to visualize the change, we are
going to create two functions to displaythe images the
first being a one to display one image and the second
for two images. After that, we then create a function

called processing that just receives the images as a
parameter.
1.3. Remove Noise (De-Noise)
Still, inside the function Processing () we add this
code to smooth our image to remove unwanted noise.
We do this using gaussian blur. Gaussian blur (also
known as Gaussian smoothing) is the result of
blurring an image by a Gaussian function. It is a
widely used effect in graphics software, typically to
reduce image noise. The visual effect of this blurring
technique is a smooth blur resembling that of viewing
the image through a translucent screen, distinctly
different from the bokeh effect produced byan out-of-
focus lens or the shadow of an object under usual
illumination. Gaussian smoothing is also used as a
pre-processing stage in computer vision algorithms in
order to enhance image structures at different scales.
1.4. Segmentation and Morphology (smoothing
edges)
In this step, we step we are going to segment the
image, separating the background from foreground
objects and we are going to further improve our
segmentation with more noise removal.
Phase 2
2.1. Binarize the image using the statistical standard
deviation method
2.2. The complement of the binarized image is
done.
2.3. Two dimensional wavelet decompositions is
done using ‘db1’ wavelet up to level two.
2.4. Re-composition of the image is done using the
approximate coefficient of previous step.
2.5. Interpolation method is used to resize the image
of the previous step to the original size.
2.6. Re-complement of the image in the last step is
done.
2.7. Labeling of the image is done using union find
method.
2.8. The maximum area of all the connected
components is found out which represents the
brain. 2.9. All other components except the
maximum component are removed from the
image.
2.10. The image obtained contains only the brain as 1
pixel.
2.11. Convex hull is computed for these 1 pixel and
the entire pixels inside the convex hull are set
to 1 and outside it are set to zero.
2.12. The image of the previous step is multiplied to
original image pixel wise and thus segmented
brain is obtained.
Phase 3
Now we find out the SIFT descriptors of each source
image of cell array for images of image dataset. SIFT
method perform the following sequence of steps for
find the keypoint descriptors for texture feature.
3.1. Scale-Space Extreme Detection
The initial step of evaluation finds total all scale-space
and different image area in image dataset nodes [4].
It is completely apply effectively by using a
Difference-of-Gaussian (DoG) mapping to represents
potential interest keypoints of feature descriptors
which are scale invariant and orientation in image
dataset nodes [6].
3.2. Keypoints Localization
All candidate area of image in selected ROI(Region of
Interest), a detailed prototypeis fitto analyze keypoints
area and its scale-space [5]. Keypoints of image areain
image ROI are chooses basis on calculate of existing
stability [6].
3.3. Orientation Assignment
One or more orientations task are applied to each
keypoints area based on local image data nodes
gradient directions [2]. Each and every future image
operations are implemented on imagekeypoint dataset
which has been transformed relative to the applied
orientation, scale, and location for each feature
descriptor, hence providing invariance to these
transformations in image data nodes.
3.4. Keypoints Descriptor
The local image gradients value are measured at the
chosen scale space in the Region of Interest (ROI)
around all keypoints in image dataset points [4].
Phase 4
In this phase, algorithm work has following steps.
4.1. First generate random matrix have same
dimension as of input image then combine this
matrix in the image. Here this help in
generating the contour in the image.
4.2. Now find the contour position in the image and
generate contours that help in finding the
segmentation of the image. This creates initial
segmentation for the image.
4.3. Once these contours were found in the image
next is to update the different segments by
finding the nearby distance from the segment
region.
4.4. Now next step is to update the segmented area
by analyzing the nearby pixel values of the
segment.
4.5. Goto step (4.3).

VI. RESULTS AND ANALYSIS
The proposed methodology was implemented in MATLAB software. For this purpose, MATLAB R2021a was
used. The image processing toolkit was used to provide essential image processing functions. The proposed
model was evaluated by implementing it in MATLAB, and the efficiency of the algorithms was analyzed.
Figure 3: Load Brain MRI Image
Figure 4: Brain Threshold Image
Table 1: Compare Precision for Brain Tumor Classification
Classes CFIB[1] CFDB[1] Proposed Model
Glioma 99.67 99.75 99.83
Meningioma 98.3 98.37 99.87
Pituitary 94 97.67 98.18
Figure 5: Graphical Comparison of Precision
Table 2: Compare Recall for Brain Tumor Classification
Glioma 97.67 99 99.62
Meningioma 96.67 97 97.15
Pituitary 99 99.21 99.84

Figure 6: Graphical Comparison of Recall
Table 3: Compare F1-Score for Brain Tumor Classification
Glioma 99 99.3 99.47
Meningioma 97.67 97.81 98.11
Pituitary 97 98 98.78
Figure 7: Graphical Comparison of F1-Score
Table 4: Compare Accuracy for Brain Tumor Classification
Model Accuracy (%)
CFIB[1] 99.34
CFDB[1] 99.51
Proposed Model 99.62
Figure 8: Graphical Comparison of Accuracy
VII. Conclusion
The accuracy of the proposed model is higher than
CFIB [1] (Joined Element based Commencement
Block/Origin CNN Model) and (Consolidated
Component based DensNet Block/DensNet CNN
Model). The accuracy worth of proposed model work
on by 0.16% and 0.08% for CFIB [1] and CFDB [1]
separately.

The review of the proposed model is higher than
CFIB [1] (Consolidated Component based Initiation
Block/Commencement CNN Model) and (Joined
Element based DensNet Block/DensNet CNN
Model). The review worth of proposed model work
separately.
The F1 Score of the proposed model is higher than
CFIB [1] (Consolidated Element based Origin
Block/Beginning CNN Model) and (Joined
Model). The F1 Score worth of proposed model work
individually.
The precision of the proposed model is higher than
CFIB [1] (Consolidated Element based Origin
Block/Beginning CNN Model) and (Joined
Model). The exactness of proposed model work on by
0.28% and 0.11% for CFIB [1] and CFDB [1]
individually.
VIII. SUGGESTIONS FOR FUTURE WORK
The opportunities for distinguishing a mind growth in
the future are that assuming we get a three-layered
picture of the cerebrum with the cancer, then, at that
point, we can gauge the sort of growth as well as the
phase of the cancer. Later on, we will investigate and
apply calibrate methods on pre-prepared models
prepared with a bigger number of layers and may
likewise scratch-based models with information
increase procedures to characterize mind growths. We
will likewise investigate the outfit strategy
(combination of classifiers yield) in light of
calibrating and scratch-based highlights separated
from profound learning models.
REFERENCES
[1] Neelum Noreen, Sellappan Palaniappan, Abdul
Qayyum, Iftikhar Ahmad, Muhammad Imran,
And Muhammad Shoaib, “A Deep Learning
Model Based on Concatenation Approach for
the Diagnosis of Brain Tumor”, IEEE Access,
Special Section on Scalable Deep Learning for
Big Data, 2020.
[2] M. I. Razzak, M. Imran, and G. Xu, ‘‘Efficient
brain tumor segmentation with multiscale Two-
Pathway-Group conventional neural
networks,’’ IEEE J. Biomed. Health Informat.,
vol. 23, no. 5, pp. 1911–1919, Sep. 2019, doi:
10.1109/JBHI.2018.2874033.
[3] A. Rehman, S. Naz, M. I. Razzak, F. Akram,
and M. Imran, ‘‘A deep learning-based
framework for automatic brain tumors
classification using transfer learning,’’ Circuits,
Syst., Signal Process., vol. 39, no. 2, pp. 757–
775, Sep. 2019, doi: 10.1007/s00034-019-
01246-3.
[4] S. Deepak and P. M. Ameer, ‘‘Brain tumor
classification using deep CNN features via
transfer learning,’’ Comput. Biol. Med., vol.
111, Aug. 2019, Art. no. 103345, doi:
10.1016/j.compbiomed.2019.103345.
[5] M. Sajjad, S. Khan, K. Muhammad, W. Wu, A.
Ullah, and S. W. Baik, ‘‘Multi-grade brain
tumor classification using deep CNN with
extensive data augmentation,’’ J. Comput. Sci.,
vol. 30, pp. 174–182, Jan. 2019, doi:
10.1016/j.jocs.2018.12.003.
[6] J. Cheng, Brain Tumor Dataset. Figshare.
Dataset. Accessed: Sep. 19, 2019. [Online].
Available: https://doi.org/10.6084/m9.figshare.
1512427.v5
[7] Y. Gu, X. Lu, L. Yang, B. Zhang, D. Yu, Y.
Zhao, and T. Zhou, ‘‘Automatic lung nodule
detection using a 3D deep convolutional neural
network combined with a multi-scale prediction
strategy in chest CTs,’’ Comput,’’ Comput.
Biol. Med., vol. 103, pp. 220–231, Dec. 2018.
[8] M. Yousefi, A. Krzyżak, and C. Y. Suen,
‘‘Mass detection in digital breast tomosynthesis
data using convolutional neural networks and
multiple instance learning,’’ Comput. Biol.
Med., vol. 96, pp. 283–293, May 2018.
[9] A. Naseer, M. Rani, S. Naz, M. I. Razzak, M.
Imran, and G. Xu, ‘‘Refining Parkinson’s
neurological disorder identification through
deep transfer learning,’’ Neural Comput. Appl.,
vol. 32, pp. 839–854, Feb. 2019, doi:
10.1007/s00521-019-04069-0.
[10] H. Zuo, H. Fan, E. Blasch, and H. Ling,
‘‘Combining convolutional and recurrent neural
networks for human skin detection,’’ IEEE
Signal Process. Lett., vol. 24, no. 3, pp. 289–
293, Mar. 2017.
[11] O. Charron, A. Lallement, D. Jarnet, V. Noblet,
J. B. Clavier, and P. Meyer, ‘‘Automatic
detection and segmentation of brain metastases
on multimodal MR images with a deep
convolutional neural network,’’ Comput. Biol.
Med., vol. 95, pp. 43–54, Apr. 2018.
[12] L. Shao, F. Zhu, and X. Li, ‘‘Transfer learning
for visual categorization: A survey,’’ IEEE
Trans. Neural Netw. Learn. Syst., vol. 26, no. 5,
pp. 1019–1034, May 2015.

[13] L. Zhou, Z. Zhang, Y. C. Chen, Z. Y. Zhao, X.
D. Yin, and H. B. Jiang, ‘‘A deep learning-
based radiomics model for differentiating
benign and malignant renal tumors,’’ Transl.
Oncol., vol. 12, no. 2, pp. 292–300, 2019.
[14] E. Deniz, A. Şengür, Z. Kadiroğlu, Y. Guo, V.
Bajaj, Ü. Budak, ‘‘Transfer learning based
histopathologic image classification for breast
cancer detection,’’ Health Inf. Sci. Syst., vol. 6,
no. 1, p. 18, 2018.
[15] C. Tan, F. Sun, T. Kong, W. Zhang, C. Yang,
and C. Liu, ‘‘A survey on deep transfer
learning,’’ in Proc. Int. Conf. Artif. Neural
Netw. Cham, Switzerland: Springer, 2018, pp.
270–279. [Online]. Available:
https://link.springer.com/book/10.1007/978-3-
030-01418-6, doi: 10.1007/978-3-030-01424-
7_27.

Brain Tumor Diagnosis using Image De Noising with Scale Invariant Feature Transform

Related slideshows

More Related Content

Brain Tumor Diagnosis using Image De Noising with Scale Invariant Feature Transform