A Combination of K-Means and Fuzzy C-Means for Brain Tumor Identification

Christy Atika Sari,Wellia Shinta Sari,Hidayah Rahmalan

doi:10.15294/sji.v8i1.29357

Abstract

Keywords are the labels of your manuscript and critical to correct indexing and searching. MRI or Magnetic Resonance Imaging is one of the health technologies used to scan the human body in order to get an image of an orgasm in the body. MRI imagery has a lot of noise that blends with the tumor object, so the tumor is quite difficult to detect automatically. In addition, it will be difficult to distinguish tumors from brain texture. Various methods have been carried out in previous studies. The method often used in the previous method is segmentation, but the process is quite heavy and the results that are less accurate are still the main obstacles. This study combines the K-Means method and Fuzzy C-Means (FCM) to detect tumors on MRI. The purpose of the combination is to get the advantages of each algorithm and minimize weaknesses. The method used is Contrast Adjustment using Fast Local Laplacian, K-Means FCM, Canny edge detection, Median Filter, and Morphological Area Selection. The dataset is taken from www.radiopedia.org. Data taken were 73 MRI of the brain, of which 57 MRIs with brain tumors and 16 MRIs of normal brain Evaluation of research results will be calculated using Confusion Matrix. The accuracy obtained is 91.78%.

Highlights

MRI or Magnetic Resonance Imaging is one of the technologies in the health sector that is used to scan the human body in order to obtain an image of an organ in the body
[2] Baid [3], Benson [4], and Padlina [5] it can be concluded that the use of K-means and FCM (Fuzzy C-Means) clustering methods is very good because in terms of program travel time is faster than those using the SVM, PSO or GA method and the results of image segmentation are accurate in detecting brain tumors with accuracy. an average of more than 88%, but must be accompanied by a noise removal step to reduce noise in the image
The difficulty of determining an accurate method with a fast process is an obstacle to detecting brain tumors on MRI images

Summary

Introduction

MRI or Magnetic Resonance Imaging is one of the technologies in the health sector that is used to scan the human body in order to obtain an image of an organ in the body This technology uses a magnetic field and radio waves that act like sensors to get the structure of internal organs. In previous paper by [2] Baid [3], Benson [4], and Padlina [5] it can be concluded that the use of K-means and FCM (Fuzzy C-Means) clustering methods is very good because in terms of program travel time is faster than those using the SVM, PSO or GA method and the results of image segmentation are accurate in detecting brain tumors with accuracy. K-Means are more susceptible to local optima and outliers, which means they are more sensitive to color differences

Objectives

Methods

Findings

Conclusion