A novel framework for efficient identification of brain cancer region from brain MRI

International Journal of Electrical and Computer Engineering (IJECE)
Vol. 9, No. 2, April 2019, pp. 1410~1417
ISSN: 2088-8708, DOI: 10.11591/ijece.v9i2.pp1410-1417  1410
Journal homepage: http://iaescore.com/journals/index.php/IJECE
A novel framework for efficient identification of brain cancer
region from brain MRI
Parvati Angadi1
, M. Nagendra2
, Hanumanthappa M.3
1
Rayalaseema University, Kurnool, India
2
Department of Computer Science & Technology, S K University, India
3
Department of Computer Science, Bangalore University, India
Article Info ABSTRACT
Article history:
Received Dec 22, 2017
Revised Jul 17, 2018
Accepted Sep 2, 2018
Diagnosis of brain cancer using existing imaging techniques, e.g., Magnetic
Resonance Imaging (MRI) is shrouded with various degrees of challenges.
At present, there are very few significant research models focusing on
introducing some novel and unique solutions towards such problems of
detection. Moreover, existing techniques are found to have lesser accuracy as
compared to other detection schemes. Therefore, the proposed paper presents
a framework that introduces a series of simple and computationally cost-
effective techniques that have assisted in leveraging the accuracy level to a
very higher degree. The proposed framework takes the input image and
subjects it to non-conventional segmentation mechanism followed by
optimizing the performance using directed acyclic graph, Bayesian Network,
and neural network. The study outcome of the proposed system shows the
significantly higher degree of accuracy in detection performance as
compared to frequently existing approaches.
Keywords:
Accuracy
Brain cancer
Magnetic resonance imaging
Optimization
Tumor detection
Copyright © 2019 Institute of Advanced Engineering and Science.
All rights reserved.
Corresponding Author:
Parvati Angadi,
Rayalaseema University,
Kurnool, India.
Email: parvati_angadi@yahoo.com
1. INTRODUCTION
Brain cancer is considered as one of the most lethal form of the cancer in human being as where the
complications of diagnosis and treatment is exponentially high [1]. Normally, brain cancer is a stage of
cancer where a brain region has presence of either lesions or tumor [2]. Lesions are required to be repaired in
order to stop being causing further internal hemorrhage. Similarly, brain tumor could be benign or
malignant [3]. However, in both the cases, an expensive surgery with complicated surgical equipment,
procedures, and highly skilled surgeon is required. All this just begins with the diagnosis stage where
normally Magnetic Resonance Imaging (MRI) is highly preferred in initial stage of analysis [4]. However,
apart from MRI, there are other imaging techniques too viz. Computed Tomography (CT) scan, Positron
Emission Tomography (PET) scan [5]. Although, with the inclusion of technology, radiological studies have
significantly improved. However, they have a limitation too [6]. At present, all the imaging techniques could
just assists in capturing the region suspected from cancer from very internal-depth (3D Doppler scan with
multi-slices), however, there is no such machines or technology who offers conclusive inference of the
resultant images. It only depends on the experience of physician that the inference is made considering a
slight probability that human error. From research viewpoint, there has been a series of research work being
carried out toward cancer detection problem; however majority of the existing research work has actually not
focused on brain MRI images. There are very less number of significant research work towards identification
of brain tumor in recent times. The most recent survey is carried out 3 year back by Ashwathy [7] to prove
that not much progress has happened in this area of cancer detection. The complication in brain cancer is

Int J Elec & Comp Eng ISSN: 2088-8708 
A novel framework for efficient identification of brain cancer region from brain MRI (Parvati Angadi)
1411
very different than other forms of cancer e.g. i) it is highly grayscale image where the tumor regions are very
less identified accurately from raw images, ii) performing preprocessing on MRI image could lead to loss of
some significant information until and unless it is not assured that it has some noise or errors, iii) normally
brain images are highly symmetrical in nature that causes a greater degree of challenges when it comes to
feature extraction process, iv) usage of training or machine learning algorithm could result in generation of
higher quantity of false positive until and unless a proper stopping function is designed or develop a strategy
that can recognize the elite outcomes. Such complications are found very less to be assumed in existing
system that causes increasing number of research effort in this direction [8], [9]. It is also seen that
segmentation is part and parcel of almost every detection techniques in existing system; however,
the techniques are too straightforward. Existing segmentation focuses on just foreground details whereas in
order to perform clustering and classification for bigger data set, the background information is equally
essential. There is a lot of scope of improvement in addressing the problem of detecting brain tumor which
we find very less in number in existing system and thereby it poses a significant research gap.
Therefore, this manuscript offers a very simple optimization technique of brain tumor detection
where there is split emphasis on computational efficiency as well as system performance concerning
accuracy. The potential features of Neural Network, a Bayesian network, watershed transform-based
techniques were harnessed to optimize the accuracy performance. Usage of such techniques results in non-
iterative and progressive enhancement of the extraction of information that redefines the value of features
being extracted that finally results in optimal performance in the identification of brain tumor region
extraction from the MRI images of the brain. Section 1.1 discusses the existing literature followed by the
discussion of research problems in Section 1.2 and proposed solution in 1.3. Section 2 discusses algorithm
implementation followed by the discussion of result analysis in Section 3. Finally, the conclusive remarks are
provided in Section 4.
This section discusses about the existing research approach being carried out towards identification
of brain cancer. The study carried out by Anitha and Murugavalli [10] have presented a classification
technique for investigating brain tumour cells using K-Nearest Neighbor algorithm. Usage of sophisticated
image capturing devices was introduced in the study of Baghaei et al. [11] for identification of lesions from
the phantom brain images that has exhibited the capability to identify less than 7 mm size of lesions. Similar
direction of the study towards detection process was also carried out by Beddad and Hachemi [12] who have
used fuzzy clustering method for classification along with the level set algorithm. Segmentation has been
marked as role player in majority of the existing studies. Work done by Demirhan et al. [13] has presented a
simple segmentation process using self-organizing map and vector quantization. The technique has also used
wavelets for extracting features. Similar problem towards segmentation was also addressed by Hamamci et
al. [14] on contrasted image using cellular automata. Haritha [15] have studied various detection methods
where k-means algorithm is found to offer better computational performance. Problems pertaining to
segmentation and detection was also addressed by Islam et al. [16] using Gabor filters and enhancing
AdaBoost algorithm. Kavitha et al. [17] have used feed-forward algorithm and radial basis function to
enhance the detection process. Nagtode et al. [18] have used discrete wavelet transform, Gabor transforms,
and neural network for classification of cancer. Usage of morphological operators and segmentation for
detecting cancer was seen in the work of Nandi [19]. Adoption of time-resolved fluorescence is used by Nie
et al. [20] for identification of tumour margin in brain. Perez et al. [21] have studied various techniques of
identifying cancer stages of brain from MRI images. Resaeieh et al. [22] have investigated microwave
imaging for identification of brain tumor followed by process of performance verification. Salwe et al. [23]
have introduced an adaptive process of wavelet implementation where fine windowing operation as well as
thresholds was used for detecting tumor cells of brain. Adoption of entropy factor toward similar detection
problem was seen in the work carried out by Somwanshi et al. [24] for enhancing the segmentation
performance. Dictionary learning was used adopted for incorporating adaptiveness in the detection process
followed by usage of sparse reconstruction process as seen in the work carried out by Su et al. [25].
Subashini and Sahoo [26] used back propagation as well as pulsed_coupled neural network for detection
brain tumor cells. Thara and Jasmine [27] used fuzzy K-means clustering as well as C-means clustering in
order to perform segmentation that is further followed by neural network for image classification. Usage of
morphological operations for addressing similar problem was also seen in the work of Vishnumurthy
et al. [28]. Zhang et al. [29] have used microwave imaging and especially Vivaldi antenna for detecting brain
tumor. Hence, existing system offers various approaches with benefits as well as limitation. The next section
outlines the research problems identified.
The significant research problems are as follows:
a. Existing approaches mainly subjects the input image to either segmentation, clustering, classification,
training algorithms without much emphasizing on processing more.

 ISSN: 2088-8708
Int J Elec & Comp Eng, Vol. 9, No. 2, April 2019 : 1410 - 1417
1412
b. Existing segmentation algorithms only focus on the foreground whereas background information is
ignored resulting in cases of false positives.
c. Level of enhancement towards morphological operation is very much limited only to erosion and dilation
whereas more potential of the morphological operator could be harnessed.
d. Lesser focus on complex features extraction is another cause of false positive especially when subjected
to machine learning algorithms.
Therefore, the problem statement of the proposed study can be stated as “It is highly
computationally challenging task to formulate a framework that uses cost-effective techniques to extraction
substantial information from the image that could assist in superior identification of brain cancer." The next
section discusses the proposed solution.
The proposed study is an extension of our prior contribution [30] where we presently focus on
using series of smaller and computationally cost-effective techniques to leverage the accuracy of brain
cancer detection.
The methodology involved in designing the first case study is as briefly discussed:
a. The rationale behind Selection of Brain MRI Image: The first part of our study chooses to work on brain
MRI images owing to the complication of the type of oncological cancer. Brain cancer is still the most
challenging cancerous condition with increased mortality rate, and hence accuracy in diagnosis is highly
required. We believe that owing to the complicated form of the brain MRI images; there are possibilities
of performing inaccurate feature extraction that may call for false alarm. Therefore, owing to the
criticality of the disease condition, we choose to consider brain MRI image as our first case study.
b. Dataset used: MR-TIP database.
c. Process Involved in Design Methodology: The proposed system takes the input of Brain MRI image and
applies gradient magnitude for better definition of possible edges Figure 1. This is followed by applying
watershed-based transformation scheme for assisting in better segmentation scheme. The processed image
is further marked for both foregrounds as well as a background. Finally, a watershed algorithm is applied
to perform proper segmentation of the foreground object. A Bayesian Network (BN) is constructed along
with Directed Acyclic Graph (DAG) followed by histogram calculation and region merging. This entire
process significantly assists in extracting enough information about the normal and abnormal condition of
an object. This stage results in the formation of features that are finally subjected to the supervised
training carried out by Artificial Neural Network (ANN).
Finally, the detection process is carried out to detect the location of cancer, and the study outcome is
assessed using conventional accuracy parameters, e.g., True/false- positive/negative, Sensitivity, Recall rate,
Specificity, Precision, F1-Score. The next section discusses the algorithm implementation for developing the
proposed detection mechanism.
Input Brain MRI
Image
Gradient
magnitude
Watershed
Transform
Mark
Foreground
Remove dark
Spots
Regional
Maxima
Superimpose
Mark
Background
Watershed(Seg
mentation)
Construct DAG Construct BN
Histogram
Calculation
Region Merging
Extract FeaturesANN TrainingANN Detection
Figure 1. Methodology adopted for 1st case study
2. ALGORITHM IMPLEMENTATION
This section discusses the algorithm that performs identification of the cancerous region for a given
MRI image of a brain. The algorithm takes the input image I that after processing yields to the final image Idet
with suspected cancerous regions. The significant steps of the algorithm are as follows:

1413
Algorithm for cancer detection from brain MRI image
Input: I (input brain MRI)
Output: Idet (Image detected with cancer)
Start
1. Obtain
22
yxm IIg  , where (Ix, Iy)=ϕ(ΔI)
2. Iwsμ(gm)
3. Imfmr1(struct(f1(Io), f2(Ie)))
4. Idsmr2(struct(Io))
5. Irmf3(α(β(γ(θ(struct(Imf))))))
6. Imbf4(β, γ)
7. Isi μ(minima(gm(Imb)))
8. Construct IDAG, IBN
9. For i=1:max(Isi)
10. h1=H[I(R)]
11.End
12. For j=1:size(IDAG)
13. c=BC/∑(ψ1+ ψ2)
14. Irmf5(IBN, c)
15. End
16. Ifeatstats(I(R))
17. Apply ANN training (Ifeat)
18. IdetUse GT for cancer detection
End
The steps of the above mentioned algorithm is as discussed below: Initially, the input image I
undergoes a processing to obtain gradient magnitude gm (Line-1). For this purpose, image filter ϕ is applied
to the image I where a Sobel operator Δ is applied to extract Ix and Iy. The next part of the algorithm is to
apply the watershed transform μ on the accomplished gradient magnitude gm of input image (Line-2) in order
to obtain the transformed image Iws. The next step of the algorithm is to mark the foreground of an image
(Line-3). For this purpose a function f1 is used for opening the image morphologically and a function f2 is
used for eroding the image with respect to the disk morphological structure struct. Finally, a method mr1 is
used for reconstructing the morphological factors of an image I with respect to Ie in order to obtain finally
marked image Imf. This process is also followed by removal of the dark spots by applying morphological
closing operator mr2 on Io with respect to the structure struct (Line-4). The next part of the algorithm
performs dilation operation θ on foreground marked image Imf with respect to the structure struct (Line-5).
This operation is further succeeded by reconstruction process γ carried out morphologically that is further
encapsulated by function for computation of complement image β. Finally, we apply morphological function
of region maxima function α and this is further encapsulated by the function f3 used for superimposing the
foreground in order to obtain superimposed image of regional maxima Irm. Now the algorithm applies a
principle of background marking and for this it obtains the binary image from the prior accomplished steps
using the function β and γ (Line-6).
The obtained image is than explored for its distance between two transformed points in the image
that is finally followed by the operation of the watershed transformation operator f4 (Line-6) in order to obtain
the image where the marked background is obtained Imb. After obtaining both foreground and background
image, the next step will be to apply the processing for initiating the segmentation process. For this purpose,
initially the image with gradient magnitude gm and the concatenation of background image Imb and image
with regional maxima Irm and edges and this processing offers second stage gradient magnitude followed by
minima-based operation (Line-7). This operation is further subjected to watershed algorithm μ for better
segmentation performance in order to obtain the segmented image Isi. For better computational performance,
we apply direct acyclic graph IDAG and Bayesian network IBN in order to further optimize the detection
performance (Line-8).
The mechanism applied for constructing DAG and Bayesian Network is as follows: The first step in
construction acyclic graph is to give labels to all the regions and in order to do this a loop is created for all
the possible value of labels (extracted from watershed transform) where all the regions are extracted along
with morphological dilation operation from where the unique labels are extracted logically. The next step is
to apply a probabilistic graphical model, i.e., Bayesian network on obtained directed acyclic graph
concerning sizes of the nodes. The next step of the algorithm is to perform region merging operation. In order
to do so, a loop is created for all the maximum values of segmented image Isi (Line-9), where regions R are

 ISSN: 2088-8708
1414
checked for their equivalency with the maximum values. A histogram operator H is then applied to the image
with only region information to finally extract the histogram h1 (Line-10). This operation is further followed
up by computation of merged regions. For this purpose a loop is created for all the maximum values obtained
by finding the size of direct acyclic graph, where all the connected components for each segments are found.
A histograph of the current node is obtained followed by the computation of the Bhattacharya Coefficient BC
(Line-13). Finally, all the connected components c is computed using a simple mathematical expression that
uses Bayesian network once again as shown in Line-14. This process results in Irm that is a final region
merging image. The function f5 represents region merging operation and the variables ψ1 and ψ2 represents
cell with 1x1 dimension and 1x dimension of all the connected components respectively. This step is
followed by extraction of statistical features stats on the regions of an image in order to obtain an image
matrix Ifeat (Line-16). Finally, neural network is applied on these features (Line-17) where the ground truth
GT images were used for assessing the regions that are possibly infected by cancer in resultant image Idet
(Line-18). Therefore, the proposed algorithm implements various smaller steps in order to finally perform
detection of cancerous portions for a given MRI image of a brain. The next section discusses about the results
being obtained
3. RESULTS ANALYSIS
This section discusses the result being obtained after implementing the concept written in the prior
section. The complete analysis is carried out on images obtained from MR-TIP database [31] where the
dimension of the images are approximately 618x630 with the occupation of 36.2 KB of memory. The visual
outcomes of the proposed system are shown in Figure 2 where it can be seen that input images after
subjecting to gradient magnitude loses its contrast and highlights for edge-based information Figure 2(b).
Applying watershed algorithm further assists in an effective segmentation process Figure 2(c) and hence it
highlights all the possible colors for the transformed image. In the process of segment, it was found that
foreground marking Figure 2(d) offers internal changes and not much on superficial changes followed by
region merging Figure 2(e). A closer look into the region merging operation in Figure 2(e) shows that the
suspected portion of cancer could be located within the region. Therefore, it can be said that true outcome
starts coming only after region merging operation. Further implying the direct acyclic graph and Bayesian
network, the identification performance of the cancerous regions has tremendously increased as well as it
offers better inference of the outcome of the processed image. At the same time, all the missing pixel
elements are successfully handled as all the encoding is carried out using Bayesian network. Even from the
accuracy viewpoint, the proposed system offers highly enhanced accuracy level as it can mitigate the
problem of overfitting to maximize the outcome of identification. Finally, applying neural network training
has further optimized the outcome the accuracy level of the proposed system with the shortest iteration of the
learning process. The complete optimization was carried out using damped least square algorithm.
For better study effectiveness, we compare the outcome of the proposed study with that of Support
Vector Machine (SVM) and K-Nearest Neighbor (KNN) algorithm that is frequently used in existing system
for classification purpose. The numerical outcome is shown in Table 1.
Table 1. Comparative Analysis of Existing System
Parameters SVM KNN Proposed System
Sensitivity/Recall rate 0.45721 0.32872 0.653024
Specificity 0.97695 0.93786 0.999948
Precision 0.98675 0.97658 0.999920
Accuracy 95% 93% 99%
The outcome shows that proposed system using neural network offers better accuracy as compared
to SVM and KNN. The prime reason behind this is SVM and KNN offers higher computational problem
when it comes to performing classification of increased size of data. Moreover, both the algorithm performs
optimization till a threshold value whereas in neural network the optimization is more applicable on
minimization of errors. This will mean that error minimizing operation is better for neural network in every
aspect. Therefore, proposed system will neural network offers enhanced accuracy as compared to existing
system.

1415
(a) (b) (c)
(d) (e) (e)
Figure 2. (a) Input image, (b) Gradient magnitude, (c) Watershed transform, (d) Mark foreground,
(e). Region merging, (f) Detection of tumor
4. CONCLUSION
This paper has presented a simple and cost-effective technique for enhancing the accuracy level of
brain cancer detection using a series of operations. The reason behind its cost effectiveness is because none of
the algorithms used in proposed system consumes maximum memory or is dependent on higher number of
trained dataset. Most significantly, the response time of proposed system is very less as compared to the
existing system. The proposed system also introduces a unique mechanism to perform segmentation where the
importance is being shared equally to for marking both foreground and background. Usage of directed acyclic
graph as well as Bayesian Network has assisted to improve the extraction of unique information that was used
for constructing features. Finally, such features when fed to neural network, we find much-enhanced accuracy
performance. The complete analysis is found to offer higher accuracy level in contrast to existing methods.
Our future work will be to continue the presence research and include more complications in diagnosis using
novel optimization schemes.
REFERENCES
[1] Adekunle M. Adesina, Tarik Tihan, Christine E. Fuller, Tina Young Poussaint, "Atlas of Pediatric Brain Tumors,"
Springer, 2016.
[2] Jürg Hodler, Rahel A. Kubik-Huch, Gustav K. von Schulthess, "Diseases of the Brain, Head and Neck, Spine
2016-2019: Diagnostic Imaging," Springer, 2016.
[3] Alessandro Crimi, Bjoern Menze, Oskar Maier, Mauricio Reyes, "Brainlesion: Glioma, Multiple Sclerosis, Stroke
and Traumatic Brain Injuries," Springer, 2017.
[4] Christopher M. Collins, "Electromagnetics in Magnetic Resonance Imaging: Physical Principles, Related
Applications, and Ongoing Developments," Morgan & Claypool Publishers, 2016
[5] S. Caorsi, A. Massa, M. Pastorino and A. Rosani, "Microwave Medical Imaging: Potentialities and Limitations of a
Stochastic Optimization Technique," in IEEE Transactions on Microwave Theory and Techniques, vol. 52, no. 8,
pp. 1909-1916, Aug. 2004.
[6] Rivka R. Colen, Imaging of Brain Tumors, "An Issue of Magnetic Resonance Imaging Clinics of North America,"
E-Book, Elsevier Health Sciences, 2016.

 ISSN: 2088-8708
1416
[7] S. U. Aswathy, G. G. Deva Dhas and S. S. Kumar, "A Survey on Detection of Brain Tumor from MRI Brain
Images," 2014 International Conference on Control, Instrumentation, Communication and Computational
Technologies (ICCICCT), Kanyakumari, pp. 871-877, 2014.
[8] Roya Babaie Aghdam, Atieh Sadat Bayat Ghiasi, Parastoo Fatemi, Nazanin Sadat Hashemi, "Challenges in Brain
Magnetic Resonance Image Segmentation," American Scientific Research Journal for Engineering, technology,
and Sciences, vol.27(1), 2017.
[9] Ivana Despotović, Bart Goossens, and Wilfried Philips, "MRI Segmentation of the Human Brain: Challenges,
Methods, and Applications," Computational and Mathematical Methods in Medicine, 2015.
[10] V. Anitha and S. Murugavalli, "Brain Tumour Classification using Two-Tier Classifier with Adaptive
Segmentation Technique," in IET Computer Vision, vol. 10(1), pp. 9-17, 2016.
[11] H. Baghaei et al., "Comparison of Brain Phantom Lesion Imaging Capability of the Brain and Whole-Body Modes
of the Transformable HOTPET Camera," in IEEE Transactions on Nuclear Science, vol. 58(3), pp. 730-735, Jun
2011.
[12] B. Beddad and K. Hachemi, "Brain Tumor Detection by using a Modified FCM and Level Set algorithms," 2016
4th International Conference on Control Engineering & Information Technology (CEIT), Hammamet, pp. 1-5,
2016.
[13] A. Demirhan, M. Törü and İ. Güler, "Segmentation of Tumor and Edema Along With Healthy Tissues of Brain
Using Wavelets and Neural Networks," in IEEE Journal of Biomedical and Health Informatics, vol. 19(4), pp.
1451-1458, Jul 2015.
[14] A. Hamamci, N. Kucuk, K. Karaman, K. Engin and G. Unal, "Tumor-Cut: Segmentation of Brain Tumors on
Contrast Enhanced MR Images for Radiosurgery Applications," in IEEE Transactions on Medical Imaging, vol.
31(3), pp. 790-804, Mar 2012.
[15] D. Haritha, "Comparative Study on Brain Tumor Detection Techniques," 2016 International Conference on Signal
Processing, Communication, Power and Embedded System (SCOPES), Paralakhemundi, Odisha, India,
pp. 1387-1392, 2016.
[16] A. Islam, S. M. S. Reza and K. M. Iftekharuddin, "Multifractal Texture Estimation for Detection and Segmentation
of Brain Tumors," in IEEE Transactions on Biomedical Engineering, vol. 60(11), pp. 3204-3215, Nov 2013.
[17] A. R. Kavitha, C. Chellamuthu and K. Rupa, "An Efficient Approach for Brain Tumour Detection Based on
Modified Region Growing and Neural Network in MRI Images," 2012 International Conference on Computing,
Electronics and Electrical Technologies (ICCEET), Kumaracoil, pp. 1087-1095, 2012.
[18] S. A. Nagtode, B. B. Potdukhe and P. Morey, "Two Dimensional Discrete Wavelet transform and Probabilistic
Neural Network used for Brain Tumor Detection and Classification," 2016 Fifth International Conference on Eco-
friendly Computing and Communication Systems (ICECCS), Bhopal, pp. 20-26, 2016.
[19] A. Nandi, "Detection of Human Brain Tumour using MRI Image Segmentation and Morphological Operators,"
2015 IEEE International Conference on Computer Graphics, Vision and Information Security (CGVIS),
Bhubaneswar, pp. 55-60, 2015.
[20] Z. Nie et al., "Integrated Time-Resolved Fluorescence and Diffuse Reflectance Spectroscopy Instrument for
Intraoperative Detection of Brain Tumor Margin," in IEEE Journal of Selected Topics in Quantum Electronics,
vol. 22(3), pp. 49-57, May-Jun 2016.
[21] U. Perez, E. Arana, and D. Moratal, "Brain Metastases Detection Algorithms in Magnetic Resonance Imaging,"
IEEE Latin America Transactions, vol. 14(3), pp.1109-1114, 2016.
[22] S. Ahdi Rezaeieh, A. Zamani and A. M. Abbosh, "3-D Wideband Antenna for Head-Imaging System with
Performance Verification in Brain Tumor Detection," in IEEE Antennas and Wireless Propagation Letters, vol. 14,
no. 3, pp. 910-914, 2015.
[23] S. Salwe, R. Raut and P. Hajare, "Brain Tumor Pixels Detection using Adaptive Wavelet Based Histogram
Thresholding and Fine Windowing," 2016 International Conference on Information Technology (InCITe) - The
Next Generation IT Summit on the Theme - Internet of Things: Connect your Worlds, Noida, pp. 256-260, 2016.
[24] D. Somwanshi, A. Kumar, P. Sharma and D. Joshi, "An Efficient Brain Tumor Detection from MRI Images using
Entropy Measures," 2016 International Conference on Recent Advances and Innovations in Engineering (ICRAIE),
Jaipur, pp. 1-5, 2016.
[25] H. Su, F. Xing and L. Yang, "Robust Cell Detection of Histopathological Brain Tumor Images Using Sparse
Reconstruction and Adaptive Dictionary Selection," in IEEE Transactions on Medical Imaging, vol. 35(6),
pp. 1575-1586, Jun 2016.
[26] M. M. Subashini and S. K. Sahoo, "Brain Tumour Detection using Pulse Coupled Neural Network (PCNN) and
Back Propagation Network," IET Chennai 3rd International on Sustainable Energy and Intelligent Systems
(SEISCON 2012), Tiruchengode, pp. 1-6, 2012.
[27] K. S. Thara and K. Jasmine, "Brain Tumour Detection in MRI Images using PNN and GRNN," 2016 International
Conference on Wireless Communications, Signal Processing and Networking (WiSPNET), Chennai,
pp. 1504-1510, 2016.
[28] T. D. Vishnumurthy, H. S. Mohana and V. A. Meshram, "Automatic Segmentation of Brain MRI Images and
Tumor Detection using Morphological Techniques," 2016 International Conference on Electrical, Electronics,
Communication, Computer and Optimization Techniques (ICEECCOT), Mysuru, India, pp. 6-11, 2016.
[29] H. Zhang, B. Flynn, A. T. Erdogan and T. Arslan, "Microwave Imaging for Brain Tumour Detection using an
UWB Vivaldi Antenna array," 2012 Loughborough Antennas & Propagation Conference (LAPC), Loughborough,
pp. 1-4, 2012.

1417
[30] P. N. Angadi and P. K. Srimani, "DODOR: Dual Optimization for Detection of Oncological Region in
Radiological Image," 2015 International Conference on Emerging Research in Electronics, Computer Science and
Technology (ICERECT), Mandya, pp. 198-203, 2015.
[31] “Brain MRI”, http://www.mr-tip.com/serv1.php?type=db1&dbs=Brain%20MRI, Retrieved on 15th
July, 2017.
BIOGRAPHIES OF AUTHORS
Parvati Angadi, is a research scholar of Rayalaseema University having 20 years of Industry and
Teaching Experience. She has published over 20 papers her area of interest focuses mainly on
Automation.
Dr. M Nagendra, is professor and HOD at Department of Computer Science and Technology,
S.K. University. He has a rich academic experience of over 28 years. His areas of interest are
Computer Networks and Software Engineering. He is an author of over 20 publications and he
mentored 6 Ph.D students. He has been teaching variety of papers in Computer Science for MCA
and M.Sc (CS) courses at Dept of Computer Science and Technology, S.K. University,
Anantpuramu
Dr. Hanumanthappa M, is professor at Department of Computer Science, Bangalore University.
Having 18 years of teaching, administration and industry experience. His research focuses on
Information retrieval, Data mining, Network security and Natural language processing. He is
author of 5 books and over 95 publications. He has adjudicated nearly 50 Ph.D and M.Phil thesis.
He has many distinctions conferred.

A novel framework for efficient identification of brain cancer region from brain MRI

More Related Content

What's hot

Similar to A novel framework for efficient identification of brain cancer region from brain MRI

More from IJECEIAES

Recently uploaded

A novel framework for efficient identification of brain cancer region from brain MRI