An effective transfer learning model for multiclass brain tumor classification using MRI images

The brain tumor is a life-threatening disease which is caused by the uncontrolled spread of cancerous cells. The diagnosis of this disease is a challenging task due to the variability and size of the lesion. Traditionally, tumors are detected using invasive techniques which are time-consuming and may cause manual errors. This may have a severe impact on patient’s survival rate. The diagnosis of tumors can be improved by adopting automated computer-aided approaches. In this article, a deep learning model is adopted to classify brain MRI images. Three different types of brain tumours, including gliomas, meningiomas, and pituitary tumours, are taken into account in the experimental study together with normal images devoid of malignancies. An effective Mobilenet with Adam Optimizer is proposed for classifying the brain MRI images. The performance of Mobilenet is compared with EfficientNet, AlexNet and VGG16. The classification accuracy of Mobilenet is 91% which is better than the other compared algorithms.