Vol. 28, No. 13, (2019), pp. 426-430
A Detailed Survey on Brain Tumor Detection using Classification and Optimization techniques
C.Moorthy Dr. K.R.Aravind Britto,
Dr. R.Vimala, S.Saravanan
Assistant Professor, Department of ECE, V.S.B Engineering College, Karur, India.
Associate Professor, Department of ECE,
PSNA College of Engineering & Technology, Dindigul, India.
Associate Professor, Department of EEE,
PSNA College of Engineering & Technology, Dindigul, India.
Assistant Professor, Department of ECE,
PSNA College of Engineering & Technology, Dindigul, India
Abstract
The death due to cancer is formed at the ending stage of the cancer severity. Hence, the detection of these cancers at a starting or primary stage is essential for preventing the sudden death in patients. The earlier detection of brain tumors is used to prevent death and this earlier detection is possible by implementing the image processing techniques on the captured or scanned brain MRI images. The implications of soft computing approaches with optimization techniques and their applications are discussed with their simulation results.
Keywords: Brain, tumor, primary stage, soft computing, optimization
1. Introduction
The abrupt of blood flow in cell creates the abnormal activity in cell. This abnormal activity of cell leads to form the cancer. In world, there are number of cancer types available. It starts from mouth cancer to end with foot cancer. Some cancer may not lead to death and some cancer may leads to immediate death. These categories are based on its severity, shape, size and position or location. In human, brain tumors and cervical tumors are the killing type tumor categories which lead to sudden death in patients. These kinds of tumors can be cured if they are predicted in starting or primary stages. Generally, the death due to cancer is formed at the ending stage of the cancer severity. Hence, the detection of these cancers at a starting or primary stage is essential for preventing the sudden death in patients.
Vol. 28, No. 13, (2019), pp. 426-430
Figure 1 (a) and (b) Abnormal brain MRI images (c) Normal brain MRI images Fig.1 (a) and (b) shows the abnormal brain MRI images and Fig.1 (c) shows the normal brain MRI images.
The earlier detection of brain tumors is used to prevent death and this earlier detection is possible by implementing the image processing techniques on the captured or scanned brain MRI images. In general method, the brain tumor detection using image processing methods are listed in the following points.
Preprocessing
o Noise reduction filter o Enhancement
o Artifact reduction
o Image pixel transformations
Feature Extractions o Texture features o Binary features o Pattern features o Energy features o Law’s texture features
Classifications
Fig.2 shows the brain tumor identification and detection methods using image processing techniques.
Vol. 28, No. 13, (2019), pp. 426-430
Figure 2 Brain tumor identification methods
Fig.2 shows the brain tumor detection methods which include noise reduction, image enhancement, artifact reduction and image domain transformations.
2. LITERATURE SURVEY
Survey based on classifications
Ji et al. (2011) developed an algorithm for detecting the abnormal tumor regions in brain images. The authors constructed Fuzzy C Means (FCM) classification algorithm for classifying each pixel in brain image into either category 1 or category 2. The category 1 represented normal image and category 2 represented abnormal tumor image. Both categories were trained and classified using FCM classification algorithm. The authors obtained 76.5% of classification rate for the classifications of brain images using their proposed method.
Louis et al. (2007) classified the brain images into normal or abnormal based on the extracted feature set in the brain images. The authors framed the behavioral model of the brain tumor detection and classification system with respect to its different behavioral responses from both normal and abnormal image categories. The authors obtained 81.2%
of classification rate for the classifications of brain images using their proposed method.
Moonis et al. (2002) proposed brain tumor detection technique with its primary point’s detection procedure on source brain MRI images. The authors estimated the volume content in each abnormal region in brain MRI images and then these volume contents were analyzed using its fuzzy connectedness algorithm. The authors achieved 67.8% of sensitivity rate and 71% of specificity rate by implementing their proposed algorithm on the images available in BRATS 2015 open access dataset.
Survey based on feature extractions and Optimization
Vol. 28, No. 13, (2019), pp. 426-430
Thirumurugan et al. (2016) identified tumor regions in brain images. The authors particularly detected Glioblastoma tumor regions in brain MRI images by proposing Adaptive Neuro Fuzzy Inference System (ANFIS) classifier. The authors achieved 78.1%
of sensitivity rate and 82.9% of specificity rate by implementing their proposed algorithm on the images available in BRATS 2015 open access dataset.
Semenov et al. (2016) developed an efficient methodology for differentiating the normal or healthy brain MRI images from tumor affected brain MRI images using optimization which was integrated with classification approaches. The author applied their proposed algorithm on the brain MRI images which were available on open access dataset
BrainWeb. Sharma et al. (2013) used optimization algorithm such as Genetic Algorithm (GA) for optimizing the extracted features from the source brain MRI images.
The authors used artificial neural network classification algorithm for detecting and segmenting the tumor regions in Glioma brain MRI images. The authors achieved 89.1%
of sensitivity rate and 91.3% of specificity rate by implementing their proposed algorithm on the images available in BRATS 2015 open access dataset.
In Dugas-Phocion et al. (2004), presented a method of adding partial volume effect to the Expectation Maximization (EM) algorithm and using the Mahalanobis distance directly within the EM. A new classification methodology based on neural networks was proposed by Zijdenbos et al. (2002). It is an automatic lesion detection system which incorporates long range contextual features. Their segmentation method used pixel-wise classifications to perform over T2w lesion image segmentations.
In Stefan Bauer et al. (2012), a tumour growth modelling combined with registration algorithms was employed. The tumour was grown in the atlas based on a new multi scale, multi physics model including tissue deformation. Large-scale deformations are handled with an Eulerian approach for finite element computations, which can operate directly on the image voxel mesh. Consequently, dense correspondence between the modified atlas and patient image was established using non-rigid registration. Their technique provides atlas-based segmentation of tumour bearing brain images as well as for improved patient specific simulation and diagnosis of tumour progression.
Datasets used in brain tumor detection
The detection and segmentation of tumors in brain tumors requires open access datasets. Most of the conventional methods use BRATS 2015 and BrainWeb dataset.
These dataset are open for all researchers for doing their brain tumor detection and segmentation method. These dataset are also contains manual detected and segmented tumor images.
BRATS 2015 utilizes multi-institutional pre-operative MRI scans and focuses on the segmentation of intrinsically heterogeneous (in appearance, shape, and histology) brain tumors, namely gliomas. Furthermore, to pinpoint the clinical relevance of this segmentation task, BraTS’18 also focuses on the prediction of patient overall survival, via integrative analyses of radiomic features and machine learning algorithms.
The Brain Web dataset contains simulated brain MRI data based on two anatomical models: normal and multiple sclerosis (MS). For both of these, full 3-dimensional data volumes have been simulated using three sequences (T1-, T2-, and proton-density- (PD-) weighted) and a variety of slice thicknesses, noise levels, and levels of intensity non- uniformity.
3. CONCLUSION
This paper states various and different conventional methods for the detection and segmentation process of brain tumors in brain MRI images. The implications of soft
Vol. 28, No. 13, (2019), pp. 426-430
computing approaches and their applications are discussed with their simulation results.
The optimization techniques which are used in the detection and segmentation of tumor regions in brain MRI images are also discussed in detail.
References
[1] Z. Ji, Q. Sun, D. Xia, A framework with modified fast FCM for brain MR images segmentation, Pattern Recognition, 44(2011), 999–1013.
[2] D.N. Louis, H. Ohgaki, O.D. Wiestler, W.K. Cavenee, P.C. Burger, A. Jouvet, B.W. Scheithauer, P. Kleihues, The 2007 WHO classification of tumours of the central nervous system, Acta Neuropathologica, 114(2007), 97-109.
[3] G. Moonis, J. Liu, J.K. Udupa, D.B. Hackney, Estimation of tumor volume with fuzzy-connectedness segmentation of MR images, American Journal of Neuroradiology, 23 (2002), 352–363.
[4] Zhu, L, Chen, Y, Lin, Y, Lin, C & Yuille, A 2012, ‘Recursive segmentation and recognition templates for image parsing’, IEEE transactions on pattern analysis and machine intelligence, vol. 34, no. 2 , pp. 359-371.
[5] Thirumurugan, P, Ramkumar, D, Batri, K & Siva Sundhara Raja, D 2016,
‘Automated detection of glioblastoma tumor in brain magnetic imaging using ANFIS classifier’, International Journal of Imaging Systems and Technology, vol. 26, no. 2, pp. 151-156.
[6] Sharma, M & Mukharjee, S 2013, Brain tumor segmentation using genetic algorithm and artificial neural network fuzzy inference system (ANFIS). In Advances in Computing and Information Technology Springer, Berlin, Heidelberg, pp. 329-339.
[7] Semenov, S, Hopfer, M, Planas, R, Hamidipour, A & Henriksson, T 2016,
‘Electromagnetic tomography for brain imaging: 3D reconstruction of stroke in a human head phantom’, in IEEE Conference on Antenna Measurements &
Applications (CAMA), pp. 1-4.
[8] G. Dugas-Phocion, M. Ballester, G.Malandain, N. Ayache, C. Lebrun, S.
Chanalet and C. Bensa, “Hierarchical segmentation of multiple sclerosis lesions in multi-sequence MRI,” in Proc. ISBI, pp. 157–160, 2004.
[9] Zijdenbos, R. Forghani and A. Evans, “Automatic pipeline analysis of 3D MRI data for clinical trials: Application to multiple sclerosis,” IEEE Trans. Med.
Imaging, vol. 21, pp. 1280–1291, 2002.
[10] Stefan Bauer, Christian May, Dimitra Dionysiou, Georgios Stamatakos, Philippe Buchler and Mauricio Reyes, “Multiscale Modeling for Image Analysis of Brain Tumor Studies,” IEEE Transactions on Biomedical Engineering, vol. 59, no. 1, pp. 25-29, 2012.
