A Comparative Review of Various Brain Tumor Detection Techniques

A Comparative Review of Various Brain Tumor Detection Techniques

Shilpa Pathania, Mrs. Mandeep Kaur

Department of Electronics and Communication Engineering , Beant College of Engineering and Technology,

Gurdaspur (Punjab)

Abstract: The brain tumor identification is a vital use of therapeutic picture handling. The writing study has demonstrated that the vast majority of strategies which are in existence has disregarded the low standardized pictures like pictures with commotion or bad brilliance. This paper has presented a comparison between some well known brain tumor detection techniques Recent method which utilized was K-means clustering technique integrated with fuzzy C-mean algorithm and it gave better results as comparison to others. The review has clearly shown that each technique has its own benefits and limitations as well.

Index Terms: Brain Segmentation, Brain Tumor, Brain Tumor Detection Techniques

1. INTRODUCTION

   Image segmentation is the way toward parceling an advanced picture into numerous sections (sets of pixels, otherwise called superpixels). The objective of division is to disentangle as well as change the portrayal of a picture into something that is more important and simpler to investigate. Segmentation is regularly used to find articles and limits (lines, bends, and so forth.) in pictures[1]. All the more exactly, picture division is the way toward doling out a name to each pixel in a picture to such an extent that pixels with a similar mark share certain visual attributes The aftereffect of segmentation is an arrangement of fragments that on the whole cover the whole picture, or an arrangement of forms separated from the picture (see edge identification). Each of the pixels in a district is comparable as for some trademark or registered property, for example, shading, force, or surface. Neighboring areas are altogether unique regarding the same characteristic(s). At the point when connected to a pile of pictures, common in medicinal imaging, the subsequent forms after picture division can be utilized to make 3D recreations with the assistance of addition calculations like walking 3D squares.

   1. Brain Segmentation

      The underlying stride of calculation includes cerebrum division [3]. A couple of procedures have been proposed to play out this operation and some of them are available in virtual items, for instance, Brain Visa, FSL and Brainsuite. Tragically a vast bit of them get flopped by virtue of the closeness of a tumor in the cerebrum, especially if arranged on the edge of the psyche. To deal with this issue, we propose to play out a symmetry examination, in light of the doubt that tumors are all around not symmetrically put in

      both sides of the equator, while the whole mind is generally symmetrical.

   2. Brain Tumor

   A cerebrum tumor, is an intracranial solid neoplasm, a tumor (portrayed as an unordinary improvement of cells) inside the psyche or the central spinal conduit[6]. A couple of tumors are mind diseases. Cerebrum tumors join all tumors inside the human skull (noggin) or in the central spinal conduit. They are made by a weird and uncontrolled cell division, generally speaking in the mind itself, furthermore in lymphatic tissue, in veins, in the cranial nerves, in the cerebrum envelopes (meninges), skull, pituitary organ, or pineal organ. Inside the mind itself, the included cells may be neurons or glial cells (which consolidate astrocytes, oligodendrocytes, and ependymal cells). Cerebrum tumors may in like manner spread from maladies basically arranged in various organs (metastatic tumors). Any tumor is naturally honest to goodness and life-crippling accordingly of its nosy and infiltrative character in the limited space of the intracranial sadness. Nevertheless, mind tumors (notwithstanding debilitating ones) are not always deadly, especially lipomas which are intrinsically favorable.Brain tumors or intracranial neoplasms can be unsafe (undermining) or non-cancer- causing (altruistic); in any case, the implications of hurtful or merciful neoplasms shifts from those normally used as a piece of various sorts of dangerous or non-damaging neoplasms in the body. Its hazard level depends on upon the blend of factors like the kind of tumor, its zone, its size and its state of headway. Since the psyche is all around secured by the skull, the early ID of a cerebrum tumor happens exactly when decisive gadgets are composed at the intracranial wretchedness. By and large area occurs in bleeding edge stages when the proximity of the tumor has brought on unexplained signs. Basic (bona fide) mind tumors are consistently arranged in the back cranial fossa in children and in the front 66% of the cerebral sides of the equator in adults, notwithstanding the way that they can impact any bit of the cerebrum. Detectable quality of signs and symptoms of mind tumors basically depends on upon two segments: the tumor measure (volume) and tumor region. The moment that reactions will get the chance to be particularly clear, either to the individual or people around him or her (symptom onset) is a crucial defining moment over the traverse of the finding and treatment of the tumor. The sign onset in the timetable of the change of the neoplasm depends all things considered, in transit of the

   tumor however a significant part of the time is in like manner related to the change of the neoplasm from "friendly" (i.e. direct growing/late sign onset) to more unsafe (rapidly growing/early reaction onset).Tumors can be good or debilitating, can occur in different parts of the cerebrum, and could possibly be basic tumors.

   Figure 1(A) & (B) Input image; (C) & (D) Tumor detected image

   An essential tumor is one that has begun in the mind, rather than a metastatic tumor, which is something that has spread to the cerebrum from another piece of the body. The occurrence of metastatic tumors are more predominant than essential tumors by 4:1. Tumors could possibly be symptomatic: a few tumors are found in light of the fact that the patient has side effects, others show up by chance on an imaging filter, or at a post-mortem. Figure 1 has demonstrated the information picture and tumor recognized picture.

2. SCOPE OF COMPARATIVE STUDY

   Brain Tumor detection techniques are integrated with digital image processing and it has found to be challenging task in medical research. Scope of comparative study is to improve accuracy of brain tumor detection. This paper has presented a review on various brain tumor detection techniques.There are four main techniques for brain tumour detection as given follows:

   A.Tumor Detection Using Active Contour:

   The system depends on dynamic forms advancing in time as indicated by natural geometric measures of the picture. The advancing forms actually split and union, permitting the synchronous recognition of a few items and both inside and outside limits. This approach depends on the connection between dynamic forms and the calculation of geodesics or negligible separation bends. This geodesic approach for protest division permits associating traditional "snakes" in light of vitality minimization and geometric dynamic shapes in view of the hypothesis of bend advancement. Test aftereffects of applying the plan to genuine pictures incorporating objects with openings and medicinal information symbolism show its energy. The outcomes might be reached out to 3D question division also.

   1. Based on Region Growing

      Region developing is a straightforward district based picture division technique. It is additionally delegated a ixel-based picture division strategy since it includes the choice of beginning seed focuses. This way to deal with division looks at neighboring pixels of beginning "seed focuses" and figures out if the pixel neighbors ought to be added to the area. The procedure is iterated on, in an indistinguishable way from general information grouping calculations.

   2. Using Watershed Segmentation

      A watershed is a bowl like landform characterized by highpoints and ridgelines that slide into lower heights and stream valleys[14]. A dim level picture might be viewed as a topographic help, where the dark level of a pixel is translated as its height in the alleviation. A drop of water falling on a topographic alleviation streams along a way to at long last achieve a nearby least. Naturally, the watershed of an alleviation relates to the furthest reaches of the neighboring catchment bowls of the drops of water. In picture handling, distinctive watershed lines might be processed. In charts, some might be characterized on the hubs, on the edges, or half and half lines on both hubs and edges. Watersheds may likewise be characterized in the persistent space. There are likewise a wide range of calculations to register watersheds. For a division reason, the slope size (i.e., the length of the inclination vectors) is deciphered as height data.

   3. Using neural network

   Artificial neural systems (ANNs) are non-direct information driven self versatile approach instead of the customary model based techniques. They are intense instruments for demonstrating, particularly when the hidden information relationship is obscure.

   Table 1: Comparison of Various Techniques

   Ref no.	Year	Authors	Techniques	Features	Limitations
   [1]	2015	Abdel- Maksoud , et al	Image segmentation approach using K- means clustering technique integrated with Fuzzy C-means algorithm	It is followed by thresholding and level set segmentation stages to provide an accurate brain tumor detection.	This method ignored the poor quality images with noise or poor brightness.
   [2]	2014	Preetha, R., and G. R. Suresh	Performance Analysis of Fuzzy C Means Algorithm in Automated Detection of Brain Tumor." In Computing and Communication Technologies (WCCCT).	This technique focuses on tha fact that the Image Segmentation is fundamental and testing to envision the tissue of human for breaking down the MR pictures.	Original Fuzzy C-means algorithm fails to segment image corrupted by noise, outliers, and other imaging artifacts.
   [3]	2014	Halder et al.	"Brain tumor detection using segmentation based Object labeling algorithm	This strategy removes the tumor by utilizing K-means algorithm took after by Object labeling algorithm.	It results in bad image quality with presence of noise.
   [4]	2014	Zeljkovic et al.	"Automatic brain tumor detection and segmentation in MR images	It demonstrated that the MRI or CT examine pictures are essential follow up analytic devices when a neurologic exam shows a probability of an essential or metastatic mind tumor presence.	Although MRI scans showed the tumor area identification but an exact loacation cant be determined.
   [5]	2013	Salah et al.	"Fully Automated Brain Tumor Segmentation Using Two MRI Modalities."In Advances in Visual Computing	The technique is tried utilizing 19 hand- fragmented genuine tumors which indicates extremely precise outcomes in contrast with an exceptionally late strategy (STS) as far as the Dice coefficient.	It gave better results but still there was some noisy images.
   [6]	2013	Dvorak et al.	"Automatic detection of brain tumors in MR images." In Telecommunications and Signal Processing (TSP)	The proposed strategy works with T2- weighted attractive reverberation pictures, where the head is vertically adjusted.	Not recognized the exact symmetry of brain.
   [7]	2013	Ulku et al.	"Computer aided brain tumor detection with histogram equalization and morphological image processing techniques."	The CAD framework has been exhibited which depends on histogram equalization.	CAD methods leads to poor quality images.
   [8]	2013	Vijay and Subhashini	An efficient brain tumor detection methodology using K-means clustering algorithm.	It has examined that the Segmentation of pictures holds a vital position in the territory of picture handling.	It leads to noisy images with poor brightness.
   [9]	2012	Parisot et al	"Graph-based detection, segmentation & characterization of brain tumors."	It clarified that new system for discovery, division and portrayal of brain tumor.	This strategy misuses past data in the structure of an extra diagram.
   [10]	2012	Bhattacharjee and Chakraborty	"Brain tumor detection from MR images: Image processing, slicing and PCA based reconstruction."	It clarified that calculation is created to quality out tumor from undesirable mind Magnetic Resonance (MR) symbolism.	It calculation depends upon PCA only so there was lesser scope.
   [11]	2012	Ghanavati et al.	"Automatic brain tumor detection in magnetic resonance images."In Biomedical Imaging (ISBI).	It considered about the Automatic discovery of brain tumor is a mind boggling work because of varieties in sort, size, area and state of tumors.	Leads to poor brightness.
   [12]	2012	Natarajan et al.	"Tumor detection using threshold operation in MRI brain images." In Computational Intelligence & Computing Research (ICCIC).	The procedure comprises of the accompanying strides: preprocessing by utilizing honing and middle channels, improvement of picture is performed by histogram balance, division of the picture is performed by thresholding.	The noise factor again not considered.
   [13]	2012	Abdullah et al	"Implementation of an improved cellular neural network algorithm for brain tumor detection."	It has proposed a brain tumor discovery strategy in light of cell neural systems (CNNs).	Neural system with the combination of cells was not efficient.
   [14]	2012	Maiti and Chakraborty	"A new method for brain tumor segmentation based on watershed and edge detection algorithms in HSV colour model.	In this technique watershed method is used in combination with edge detection operation.	The algorithm used was not sufficient.
   [15]	2012	M. Monica et al.	"Brain tumour detection using Pulse coupled neural network (PCNN) and back propagation network."	In this work, the pictures acquired through MRI are portioned and afterward encouraged to a model known as Pulse coupled neural system for identifying the nearness of tumor in the cerebrum picture.	Irregularity of pulses showed the presence of noise.

   An imperative element of these systems is their versatile nature, where "learning by illustration" replaces "programming" in taking care of issues.

3. COMPARATIVE ANALYSIS

   Table 1 has clearly shown the comparison among the various brain tumor techniques. However the comparison has clearly shown that the integration of k-means clustering with fuzzy c-means algorithm has better results than the available methods.

4. CONCLUSION

This paper has presented a comparison between some well known brain tumor detection techniques. Recent method which utilized was K-means clustering technique integrated with fuzzy C-mean algorithm and it gave better results as comparison to others. The review has clearly shown that each technique has its own benefits and limitations as well. . In near future, brain tumor detection will be improved further using decision based alpha trimmed filter which can give more better results.

REFERRENCES

1. Abdel-Maksoud, Eman, Mohammed Elmogy, and Rashid Al-Awadi. "Brain tumor segmentation based on a hybrid clustering technique." Egyptian Informatics Journal 16, no. 1 (2015): 71-81.

2. Preetha, R., and G. R. Suresh. "Performance Analysis of Fuzzy C Means Algorithm in Automated Detection of Brain Tumor." In Computing and Communication Technologies (WCCCT), 2014 World Congress on, pp. 30-33. IEEE, 2014.

3. Halder, Amitava, Chandan Giri, and Amiya Halder. "Brain tumor detection using segmentation based Object labeling algorithm." In Electronics, Communication and Instrumentation (ICECI), 2014 International Conference on, pp. 1-4. IEEE, 2014.

4. Zeljkovic, V., C. Druzgalski, Y. Zhang, Z. Zhu, Z. Xu, D. Zhang, and P. Mayorga. "Automatic brain tumor detection and segmentation in MR images." In Health Care Exchanges (PAHCE), 2014 Pan American, pp. 1-1. IEEE, 2014.

5. Salah, Mohamed Ben, Idanis Diaz, Russell Greiner, Pierre Boulanger, Bret Hoehn, and Albert Murtha."Fully Automated Brain Tumor Segmentation Using Two MRI Modalities."In Advances in Visual Computing, pp. 30- 39.Springer Berlin Heidelberg, 2013.

6. Dvorak, Pavel, Walter Kropatsch, and Karel Bartusek. "Automatic detection of brain tumors in MR images." In Telecommunications and Signal Processing (TSP), 2013 36th International Conference on, pp. 577-580. IEEE, 2013.

7. Ulku, Eyup Emre, and Ali Yilmaz Camurcu. "Computer aided brain tumor detection with histogram equalization and morphological image processing techniques." In Electronics, Computer and Computation (ICECCO), 2013 International Conference on, pp. 48-51. IEEE, 2013.

8. Vijay, J., and J. Subhashini. "An efficient brain tumor detection methodology using K-means clustering algorithm." In Communications and Signal Processing (ICCSP), 2013 International Conference on, pp. 653-657. IEEE, 2013.

9. Parisot, Sarah, HuguesDuffau, StéphaneChemouny, and Nikos Paragios."Graph-based detection, segmentation & characterization of brain tumors." In Computer Vision and

   Pattern Recognition (CVPR), 2012 IEEE Conference on, pp. 988-995. IEEE, 2012

10. Bhattacharjee, Rupsa, and MonishaChakraborty. "Brain tumor detection from MR images: Image processing, slicing and PCA based reconstruction." In Emerging Applications of Information Technology (EAIT), 2012 Third International Conference on, pp. 97-101.IEEE, 2012.

11. Ghanavati, Sahar, Junning Li, Ting Liu, Paul S. Babyn, Wendy Doda, and G. Lampropoulos."Automatic brain tumor detection in magnetic resonance images."In Biomedical Imaging (ISBI), 2012 9th IEEE International Symposium on, pp. 574-577.IEEE, 2012.

12. Natarajan, P., N. Krishnan, Natasha SandeepKenkre, Shraiya Nancy, and BhuvaneshPratap Singh."Tumor detection using threshold operation in MRI brain images." In Computational Intelligence & Computing Research (ICCIC), 2012 IEEE International Conference on, pp. 1-4. IEEE, 2012.

13. Abdullah, Azian Azamimi, Bu Sze Chize, and Yoshifumi Nishio. "Implementation of an improved cellular neural network algorithm for brain tumor detection." In Biomedical Engineering (ICoBE), 2012 International Conference on, pp. 611-615. IEEE, 2012.

14. Maiti, I., and M. Chakraborty. "A new method for brain tumor segmentation based on watershed and edge detection algorithms in HSV colour model." In National Conference on Computing and Communication Systems. 2012.

15. M. Monica, Subashini, and Sarat Kumar Sahoo. "Brain tumour detection using Pulse coupled neural network (PCNN) and back propagation network." In Sustainable Energy and Intelligent Systems (SEISCON 2012), IET Chennai 3rd International on, pp. 1-6. IET, 2012.

16. Sheikh, Anjum, R. K. Krishna, and Subroto Dutt. "Energy Efficient Approach for Segmentation of Brain tumor Using Ant Colony Optimization." International Journal of Computer Technology and Electronics Engineering (IJCTEE) Volume1.