A Review of Deep Learning Models for Brain Tumor Prediction Using MRI Images

Tumor is the leading cause of death in today’s time. Many people lost their lives due to this disease. It can be developed at any part of the body. The brain tumor is one such type of tumor which is the deadliest. It develops when cells start growing abnormally in our brain. It hampers the overall functioning of our brain and can be life-threatening also. So, it becomes necessary to detect it at an early stage. The evolution of deep learning made it possible to detect tumor in the human body very early. With the help of MRI images, it becomes quite easy to detect the tumor on time. Several methods based on deep learning have been proposed to analyze these MRI images. Every method has its advantages and disadvantages. One of the most adopted methods for tumor detection is CNN, as it works tremendously exceptionally in the field of image recognition and tumor classification. In this paper, we have studied how CNN is one of the best methods to effectively analyze and predict brain tumor using MRI images. The CNN model is reviewed and implemented for brain tumor prediction. To increase the accuracy of the model a feature extraction technique is performed using VGG16. Further, the performance of the prediction model is measured using different metrics. The proposed model achieved 93% accuracy, 0.92 recall, 0.94 precision, and 0.91 F1-score respectively. When compared with existing models such as densenet, alexnet, and resnet, the proposed model achieved better results.