Automated Tumor Detection using Convolution Neural Network

Brain tumors are quite formidable primarily because they can occur at any place within the brain and may also exhibit dissimilar behavior patterning. The disease in its initial stage is not very dangerous, if it is not treated then the result is a lot of damage to the brain. To overcome these challenges, further research and development is needed and more specifically the implementation of new technologies. This paper seeks to contribute to these goals by developing a deep learning convolutional neural network with transfer learning approach, that can help in detection of brain tumors with high accuracy and efficiency. In the long run, it should help in the early diagnosis and timely appropriate treatment of patients which should somehow contribute to the development of neurooncology. This study proposes a model that involves MRI for detection of brain tumors through the application of deep learning. The high performance of this model is attributed to the transfer learning technique that has been applied on VGG16 trained on 3000 medical MRI images of the human brain. The proposed model provided quite satisfactory accuracy of 99%. 33% on this dataset. Using this automated detection model, it is possible for the healthcare professionals to be able to predict the presence of brain tumors quite well using this system.