An Efficient Brain Tumor Detection Using Ensemble Learning

Abstract

Brain tumors are serious health condition which need to be timely diagnosed and treated to avoid undesirable happenings. Most cancer-related deaths worldwide are caused by brain tumors. For effective treatment and better patient outcome, brain tumors must be detected correctly at their early stage in magnetic resonance imaging (MRI). MRI images are widely used for the diagnosis of brain tumors. The literature shows some detection approaches using deep learning methods to detect brain tumors but they lag in performance. To overcome this issue, we have proposed an ensemble learning based model that integrates transfer learning-based models with convolutional neural network (CNN). The proposed model trained and evaluated using a standard brain tumor dataset consisting of 3000 brain MRI images and achieved high detection accuracy of 98% with a very low false positive rate. Thus the proposed methodology is quite useful in the detection of brain tumors. The proposed method can help radiologists to identify brain tumors precisely which in turn may result an earlier diagnosis and better patient treatment.