Automatic Detection Brain Segmentation to Detect Brain Tumor Using MRI

Abstract

Brain tumors are a type of disease in the form of lumps of meat that grow in the brain. In differentiating brain tumor tissue from normal tissue become a difficulty caused by the same colors are an obstacle in seeing brain tumors using MRI images. Accuracy is needed in analyzing brain tumors. However, currently, radiographers (radiologists) still analyze the results of manual MRI images of brain tumors. Therefore we need a method that is able to segment MRI images precisely and automatically, with the aim of obtaining faster and more accurate image segmentation of brain tumors so that we can know the percentage of brain tumors found in the brain. To overcome difficulties when segmenting brain tumors in separating brain tumor tissue from other tissues such as normal brain tissue, cerebrospinal fluid, fat, and edema, a learning-based system method that will carry out the training process uses Haar training to narrow the MRI image so that it is more focused on the part of the head object. Then median filtering is performed to maintain the edge of the image on the MRI image. Then the segmentation process using the thresholding method is run, then repeated to take the largest area. Segmentation of brain is carried out by marking the brain area and the area outside the brain using the DAS method and then cleaning the skull using the cropping method. In this research, 12 images of MRI brain tumors were used. The results of segmentation compared to area of the brain tumor and area of the brain tissue. The system obtains a calculation of the tumor area having an average error of 10,5%.