Detection of brain tumors on MRI images using active contour segmentation and convolutional neural network

The brain tumor is a dangerous disease because it attacks vital organs and can affect anyone. Therefore, magnetic resonance imaging (MRI) is usually used for the early diagnosis of the disease. This paper discusses the application of image processing for the detection of brain tumors based on MRI images. The initial stage is preprocessing, which includes gray scaling techniques, median filters, intensity settings, and histogram equalization. The next stage is segmentation to obtain objects from brain tumors in the image using the active contour method. The segmentation results are then trained using the Convolution Neural Network (CNN) algorithm to classify images with brain tumors and images without brain tumors. In the testing process, the accuracy value is 0.8456, and the reliability value is 1.1467. This accuracy and reliability value shows that the combination of active contour segmentation techniques and CNN classification techniques can detect brain tumors well on MRI images.