Brain Tumour Detection Using Stacked Deep Learning Approaches

A brain tumour is a mass or uncontrolled growth of abnormal cells in the brain. tumours have different types and characteristics and have different treatments. Some tumours are noncancerous (benign tumours), some tumours are cancerous (malignant), and brain tumours can be of any type or size. Given the seriousness of this disease, it must be detected early in order to increase the chance of recovery for the patient. Early detection of the tumour plays a major role in the treatment of the patient, as early detection of the tumour would increase the chances of treatment for the patient. In this paper, we used a stacked scratch-built and transfer learning convolutional neural network model to detect brain tumour with high accuracy. The model was designed and implemented on two datasets. These datasets contain a set of images for brain tumour detection. For testing, we used our models, VGG16, VGG19, AlexNet, MobileNetV2 and InceptionResNetV2 individually and further improved the accuracy of the results by integrating all these models into one model and ultimately got better results than the previous models. Experimental results are excellent when compared with contemporary brain tumour detectors. Interestingly, we achieve between 99.5% to 100% success in detecting brain tumour based on recall, precision, and F1-Score,