A Correlative Analysis of SOM and FCM Classifier for Brain Tumour Detection
Suchita Goswami1, Archana Tiwari2, Vivek Pali3, Ankita Tripathi4
1Suchita Goswami, M. Tech, Nagpur, India.
2Archana Tiwari, Department of Electronics Engineering, Shri Ramdeobaba College of Engineering & Management, Nagpur, Maharashtra.
3Vivek Pali, Department of Computer Science Engineering, Chhattisgarh Institute of Technology, Rajnandgaon, Chhattisgarh.
4Ankita Tripathi, M. Tech Nagpur, India
Manuscript received on 10 April 2019 | Revised Manuscript received on 17 April 2019 | Manuscript Published on 24 May 2019 | PP: 718-723 | Volume-8 Issue-6S3 April 2019 | Retrieval Number: F11420486S319/19©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open-access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: The task of reading the MRI (Magnetic resonance Imaging) scans is difficult in case of brain abnormality detection because of variance and complexity of tumours. The diagnosis of brain tumour requires a detailed analysis of MRI scan, which includes the detail about area of tumour, blood clotting etc. The process of diagnosis involves painful invasive surgery that can cause discomfort to patients. This paper presents a separate unsupervised learning based NN (neural network) classifier to detect a tumour in the magnetic resonance human being brain images and a separate FL (Fuzzy logic) classifier for the said above. In this paper, the brain tumour diagnostic procedure is divided into the following stages. The first stage comprises of image pre-processing which includes image resizing, noise filtering and thresholding. At stage two, extraction of features from the images got from MR brain was done by the use of GLCM (Grey level co-occurrence matrix). In third stage, brain tumour was diagnosed by using NN (Self organizing map) based classifier and FL (Fuzzy C-means clustering) classifier. The obtained accuracy of NN classifier is 96% and sensitivity is 92% and specificity is 66% and that of FL classifier is 98% and sensitivity as 100% and specificity as 66.6%. The comparative analysis of specificity, accuracy and sensitivity with other techniques based on previous work is used to evaluate the classification technique performance.
Keywords: Magnetic Resonance Imaging (MRI), Fuzzy C-means Clustering (FCM), Grey Level co-Occurrence Matrix (GLCM), Accuracy, Sensitivity, Specificity.
Scope of the Article: Computer Science and Its Applications