DefinePK

Abstract

Brain tumor is a highly aggressive and life-threatening disease; this leads to a very short duration of life to patients when at higher stages. However, early detection of tumors can be cured and extend the life of patients. The early detection of tumor in the brain Magnetic Resonance Imaging (MRI) or Computed Tomography (CT) scan images are more challenging as it needs careful study. The proposed work used MRI images for diagnosing brain tumor through binary classification as tumor or non-tumor images through deep learning algorithm, Convolutional Neural Network (CNN) model is used for feature extraction from images. The pre-trained architecture VGG16 is used for feature extraction from the images. These features are used for training and validating Machine learning algorithms and comparing their performance. When the tumor results in the classification of image, the work also implements segmentation of tumor from the brain MRI images to detect the tumor in a given image using image processing techniques. The proposed work performs brain tumor classification, detection as well as segmentation. The application is developed as a web application, which is useful for users to get the results quickly and accurately through the proposed novel technique. Experimental results show that the proposed work has achieved the highest accuracy of 91.6% with a logistic regression model.