Brain Tumor Segmentation of T1w MRI Images Based on Integrated Clustering Algorithm

Early detection of brain tumors helps the specialists to take care of the affected persons, also reduce the threat, and enhance the possibility of existence. Brain tumor detection was once performed manually, which required a great deal of expertise and was time-consuming. Magnetic resonance imaging (MRI) modality used to identify tumors at early. The aim this research paper is to develop an automated segmentation method toward efficiently segmenting and extracting brain tissues from T1w MRI images. The proposed medical image segmentation method involves three main phases in the presented work: pre-processing, clustering, and validation. The presented KIFCM integrates the K-means with fuzzy c-means (FCM) techniques to takes advantage of them to overcome its limitations. To verify the output of each method, all the experiments are run on three brain tumor datasets: BRATS 2015, Brain Web simulated, and Harvard datasets. The computational time, segmentation accuracy, and the number of iterations were used to determine the algorithms results. From all the responses, the K-means clustering technique faster to detect the tumor from large datasets than FCM clustering but suffers from inadequate tumor detection. FCM algorithm detects the tumor location accurately and retains more details from the abnormal image than the K-means algorithm, but it takes more processing time. As per all responses, the KIFCM is an efficient approach to detect tumors with less time and high segmentation accuracy of 98.78%than other clustering methods. In comparison to existing approaches, the presented brain tumor segmentation system outer performs well in detecting the brain tumors with high segmentation accuracy and reduced processing time.