Fuzzy Clustering Approach for Brain Tumor Detection

Abstract

This paper focuses on using a simple algorithm to detect the range and shape of tumors in brain MR images and to determine the stage of the tumor from the given tumor region. Tumors are uncontrolled tissue growth that can occur in any section of the body. Tumors come in a number of varieties, each tumor type has its own set of characteristics and treatment options. Because of its nature in confined space of the cerebral cavity, brain tumors are inherently dangerous and life-threatening (space formed inside the skull).According to the studies carried out in developed countries, an increasing number of people with brain tumors have perished as a result of faulty detection. In most cases, a CT scan or MRI focusing on the cerebral cavity yields a comprehensive view of the brain. Some general risk factors and symptoms was established after extensive research and statistical analysis based on those people who are affected by brain tumors. The growth of technology in science day and night aims to provide new therapeutic approaches. The clinician examines this imaging method for the purpose of detecting and diagnosing a brain tumor. However, this method helps to accurately predict the size of the tumor, as well as the stage of the tumor from the tumor’s location is detected during the process. The k-means and fuzzy c-means are the algorithms that are used in this study to segment brain tumors from the given brain tumor image. This method enables the accurate and repeatable segmentation of tumor tissue, comparable to manual segmentation. Furthermore, it cuts down on analysis time and detects tumor stage from a particular tumor region. Finally, create a system that uses Java to identify the tumor stage that is easier to use, less expensive, and saves time.

Introduction

I. INTRODUCTION

MRI or CT scans are usually used to image the brain’s anatomy. The entire procedure is recorded in this paper using an MRI scan. For diagnosis, an MRI scan is more convenient than a CT scan. It does not have any adverse effect on the human body. It works by the use of magnetic fields and radio waves. Numerous algorithms have been created for the diagnosis of brain tumors. However, they may have some disadvantages in terms of identification and extraction.

Two algorithms are used in this work to perform segmentation. Algorithms for clustering K-means and Fuzzy C-means. As a result, it produces a correct tumor segmentation result. Tumors are caused by excessive tissue growth in every portion of the body. A brain tumor is a mass of cells in your brain that are not normal. There are two general groups of brain tumors: Primary brain tumors start in brain tissue and tend to stay there. Secondary brain tumors are more common. Normally, a brain tumor has an effect on the CSF (Cerebral Spinal Fluid). It contributes to strokes. The surgeon treats the strokes rather than the tumor. Therefore, early identification of tumors is crucial for successful treatment. When a brain tumor is diagnosed at an early stage, the patient’s life expectancy may be extended. This will add about one or two years to the lifespan. Typically, tumor cells are classified into two groups. They are classified as Mass and Malignant. It is somewhat difficult to detect a malignant tumor in a bulk tumor. In this post, we will explore how to diagnose brain tumors using brain MRI images and how to assess the stage of the tumor depending on the given region of tumor. Treatment for a brain tumor is determined by the tumor’s form and level as well as according to its size and location and the general health and medical records. In the majority of cases, the main aim of this therapy is to totally remove or kill the tumor. The majority of brain tumors are curable if detected and treated early.

An individual who has been diagnosed with some kind of brain tumor is at an elevated risk of having another form of brain tumor. An individual who has two or more close relatives(father, mother,brother, sister or child) who have developed brain tumors has an increased chance of developing his or her own brain tumor. Occasionally, family members will inherit a mutation that makes the brain more vulnerable and raises the chance of developing a brain tumor. Around 5% of brain tumors are thought to be caused by inherited (genetic) causes or disorders. The aim of this work is to develop a method that can inform people about their estimated risk of developing a brain tumor, whether they are at risk or not, and by how much. Java is used to build the detection platform. Finally, we have systems that detect the tumor and its form, as well as the stage of the tumor, from a given region of the tumor.

II. LITERATURE SURVEY

For magnetic resonance imaging, the method of image registration and data fusion theory presented here has been modified for the segmentation of the magnetic resonance images. This research aims to propose an image registration and data fusion method that is optimal for segmentation of magnetic resonance images. This method enables the accurate and rapid detection of brain tumors. This device utilises the K-means algorithm to provide an effective and rapid tool for diagnosing brain tumors [1].

Meena and Raja demonstrated how to use the Spatial Fuzzy C Means (PET-SFCM) is used for PET scan image datasets to cluster data in 3D. To better assist the clustering process, a proposed algorithm helps to incorporate spatial neighbourhood information into the FCM, and this helps in the modification of the objective function of each cluster simultaneously. This algorithm has been applied and validated on a large data set of patients with neurodegenerative disorders of the brain, such as Alzheimer’s disease. It has been shown to be efficient in real-world patient data sets [2].

In this case, the presented framework   features   three image segmentation algorithms, namely K-Means clustering, Expectation Maximization, and Normalized Cuts. The main objective of this project is to find a solution to the issue of cutting up a picture into separate regions. When we compare the two unsupervised learning algorithms, K-means and EM, to a graph-based algorithm, the Normalized Cut, we equate them to each other. Clustering algorithms that divide a data set into clusters based on a given distance measure, such as K-means and EM, are two common methods for clustering [3].

Funmilola et al. introduced the Combining the properties of Fuzzy C-means and K-means, the fuzzy K-C-means form can be used. While most of this work has focused on clustering algorithms like k-means and fuzzy c-means, this portion has done much of the work on both approaches. The k-c-means clustering algorithm was constructed by utilising these separate algorithms. This allowed the algorithm to provide a more effective result in terms of computational time. the algorithms have been put to the test and proven correct by comparing them to human brain MRI images Results have been thoroughly investigated and reported [4].

Wilson and Dhas, respectively, The SWI technique was used to detect iron in the brain by first using K-means and then using Fuzzy C-means. accurate assessment of iron stores is needed in a number of neurodegenerative diseases because of the relationship between iron accumulation and the aetiology of these diseases a susceptibility the weighted imaging SWI (SWI weighted for sensitivity) assists in getting an accurate depiction of a tissue’s properties which are different from the structures under which it is located [5].

The overview of various brain tumor diagnostic techniques provided here contains a full examination of their various types. The main purpose of this article is to provide a comprehensive Type-II fuzzy expert method for diagnosing human brain tumors (Astrocytoma tumors) with T1-weighted Magnetic Resonance Imaging (MRI) with contrast. four distinct modules make up the proposed Type-II fuzzy image processing method: pre-processing, segmentation, feature extraction, and approximate reasoning [6].

Human intuition plays a key role in pattern recognition, and mainstream mathematics cannot accurately accommodate this complicated and ambiguously defined system. This, coupled with the inadequacy of fuzzy mathematics to fully embrace these ideas, has led to the implementation of various fuzzy approaches [7].

This paper presents a technique that offers an efficient and synergistic algorithm for brain tumor diagnosis, utilising median filtering, K Means segmentation, FCM segmentation, and a final threshold segmentation. in this method, we aim to improve the accuracy of tumor images obtained via MRI and to use that increased accuracy to guide estimations on the size of the tumors [8].

The author of the presented work is conducting an examination of various algorithms that can be used to color images, text, and gray scale images. When image segmentation is done, the resultant collection of segments or contours will take up the whole image, or a collection of segments and contours will be derived from the image. A pixel in a particular region can be identical in some way, such as color, strength, or texture [9].

This thesis showed how to isolate features from brain images and detect tumors using k-means and C-mean clustering techniques [10].

A. Objectives

1. Identify the brain tumor on given image.
2. Detect the tumor part in given image.
3. Calculate the area of brain tumor.
4. Identify the stage of tumor.
5. Predict the disease from the given area of tumor.

III. PROPOSED SYSTEM

The clustering techniques used in the proposed work are a mixture of two others. Preprocessing, segmentation, feature extraction, and approximation reasoning are the four primary modules of the system under consideration. The median filter is used to filter the images in the pre-processing stage of the procedure. Following that, image segmentation is carried out using the K-means and Fuzzy C-means algorithms to separate objects. The image thresholding method is mainly used for feature extraction, and then approximation reasoning is used to detect the area of the tumor and position of the tumor in the given MRI image. It also determines the stage of the tumor from the resulting area of a brain tumor. To put it another way, ultimately design a system to detect the stage of a tumor that is simpler, less expensive, and saves time.

A. Explanation

1. Pre-processing: In pre-processing step, removal of the noise from the input image is carried out. It also performs filtering of noise and other artifacts in the image and sharpens the edges of the image.The conversion of the RGB to grey and reshaping is also performed in this step. It includes median filter for noise removal. Noise may arrive due to the thermal effect but the possibility is very less in the modern MRI scan method.

2. Segmentation using K-means

Steps

1. We need to specify the no of cluster value as k.
2. Randomly choose the k cluster centers
3. Mean or center of the cluster will be calculated
4. The distance between each pixel to each cluster cente
5. If the distance is near to the center then move to that cluster
6. Otherwise move to next cluster
7. Re-estimate the center.
8. Repeat the process until the center doesn’t move.

3. Segmentation using Fuzzy C means: In fuzzy-c mean clustering data is processed by giving the partial membership value to each pixel in the image.The algorithm has the membership value set ranges from 0 to 1. It is basically a multi valued logic that allows intermediate values i.e.,The pixel which is member of one fuzzy cluster can also be member of other fuzzy cluster in the same image. There is no sudden transition between full membership and non-membership.

4. Approximate Reasoning: In the approximate reasoning step the tumor area is calculated using the binarization method. That is the image having only two values either black or white (0 or 1). And then classify the stage of tumor from the given area of tumor.

B. Algorithms

1. K-means Algorithm: K-Means clustering is one of the unsupervised learning algorithm for clusters. Clustering is the technique where pixels are grouped according to the same characteristics. In the k-means algorithm, we have to define the number of clusters k. Then k-cluster center are chosen randomly. The distance between the each pixel to each cluster centers is computed. There is a comparison between a single pixel and all cluster centers using the distance formula. The pixel is assigned to the particular cluster which has shortest distance among all. At last the centroids for the clusters are computed by taking the average of all pixels of that cluster. After that each pixel is compared to all centroids. The process continuous until the center converges.
2. Fuzzy C-Means Algorithm: The Fuzzy C-means clustering is an unsupervised clustering algorithm which can be applied to several problems involving feature analysis, clustering, medical diagnosis and image segmentation. In FCM algorithm each data point belongs to a cluster to a degree specified by a membership grade. The FCM algorithm minimizes the objective function for the partition of data set, x = [x1, x2,  xd]

C. Mathematical Model

Mathematical equation for K-means Clustering

IV. RESULT AND DISCUSSION

Let us consider the brain tumor image procured from MRI, containing the tumor in Fig.2.

Median filtering is applied on the input image to get rid of the unwanted noises. The output image is displayed in the Fig.3.

K means algorithm is implemented on noise filtered image obtain as a output of median filtering. In FIG. 4. a white spot is seen, which is outcome of K-means clustering on the input image. K-Means clustering is an unsupervised learning algorithm. There is no labelled data for this clustering, unlike in supervised learning.This algorithm performs the division of objects into clusters that has similar properties and are dissimilar to the objects belonging to another cluster

Fuzzy C-Means segmentation is eventually implemented on the resulted image obtained from K-Means segmentation. The region affected by ulcer is highlighted in this process. The outcome of the Fuzzy-c mean segmentation is shown in Fig. 5.

Thresholding segmentation is implemented on the image once the FCM segmentation is completed. The outcomes are spectacular and intended approach is efficacious in nature to an extent. Fig.6. shows the resultant image procured after the implementation of thresholding segmentation.

Conclusion

Segmentation of brain tumor is done in this presented work. To start, the image is pre-processed using the median filter technique. If noise is detected in the given input MR image, it is extracted before K-means phase then the output image is given as a input to K-means clustering algorithm. The noise- free image is fed into the k-means algorithm and then into fuzzy C-mean algorithm, which extracts the tumor from the MRI image. Eventually, estimated logic is used to calculate tumor area and location, and finally, the resulting area of the tumor is used to identify the stage of the tumor, which is simpler, less expensive, and faster.

References

[1] Samir Kumar Bandhyopadhyay and Tuhin Utsab Paul, “Automatic Segmentation of Brain Tumor from Multiple Images of Brain MRI” International Journal of Application or Innovation in Engineering Man- agement (IJAIEM), Volume 2, Issue 1, January 2013. [2] A. Meena, “Spatial Fuzzy C-Means PET Image Segmentation of Neu- rodegenerative Disorder” , A. Meena et.al / Indian Journal of Computer Science and Engineering (IJCSE). [3] Suman Tatirajua and Avi Mehta, “Image Segmentation using k-means clustering EM and Normalized Cuts”. IEEE Trans. Parallel Diatribe. Syst., vol. 19, no. 5, pp. 710–720, May 2008. [4] Ajala Funmilola A*, Oke O.A, Adedeji T.O and Alade O.M, Adewusi E.A, “Fuzzy k-c-means Clustering Algorithm for Medical Image Seg- mentation”, Journal of Information Engineering and Applications ISSN 2224-5782 (print) ISSN 2225-0506 (online) Vol 2, No.6, 2012. [5] Beshiba Wilson and Julia Punitha Malar Dhas, “ An Experimental Analysis of Fuzzy C-Means and K-Means Segmentation Algorithm for Iron Detection in Brain SWI using Matlab”, International Journal of Computer Applications (0975 – 8887) Volume 104 – No 15, October 2014. [6] M.H. Fazel Zarandia, M. Zarinbal and M. Izadi, “Systematic image processing for diagnosing brain tumors”, Department of Industrial En- gineering, Amirkabir University of Technology, P.O. Box 15875-4413, Tehran, Iran , journal homepage: www.elsevier.com/locate/asoc [7] Samarjit Das, “Pattern Recognition using the Fuzzy c-means Technique”. International Journal of Energy, Information and Communications Vol. 4, Issue 1, February, 2013. [8] Varnish Rajesh, Bharathan Venkat, Vikesh Karan and M. Poonkodi, “Brain Tumor Segmentation and its Area Calculation in Brain MR Images Using K-Mean Clustering and Fuzzy C-Mean Algorithm”, Department of Computer Science and Engineering, SRM University [9] Krishna Kant Singh1 and Akansha Singh, “A Study of Image Segmen- tation Algorithms for Different Types of Images”, IJCSI International Journal of Computer Science Issues, Vol. 7, Issue 5, September 2010. [10] Payal mistry, shagun akhauri, sayali patil, s.p.tondare,”Segmentation of brain tumor and its area calculation in brain MRI images using k-mean clustering and fuzzy c- mean algorithm

Copyright

Copyright © 2022 Miss. Sonali Bharat Dhumal, Prof. M. S. Tamboli. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.