Automated Diagnosis of Brain Tumour in MRI Images Using Deep Learning Techniques

Abstract

Brain is one of the most vital organ in human nervous system. Brain contains most lacks of nerves connected and communicating each other. An unusual growth or increase of nerves in the brain, which may affect the actual working of the brain, is called a Brain tumour. In the modern day the brain tumour was detected by traditional way of laborious and traditional approach. However, this lab analysis was too time consuming, which may lead to the increase the severity of the tumour cells. To address this issue, automated brain tumour detection techniques was introduced, which will help the early detection of tumour cells and helps for the timely medicating. In this study, The Cancer Imaging Archive was utilized as a dataset. The pre-processing and data augmentation techniques have been applied for training purposes using various deep learning (DL) models. The optimizers were used to determine stronger dataset to train the model. The pre-processed dataset were passed to the model to train on it. The deep learning (DL) model Convolutional Neural Network is used with the various pre-trained models like ResNet50, Xception, VGG19, MobileNetV2, InceptionV3. The accuracy of the different models was validated and tested to obtain the model which gives the good accuracy.  This research primarily focusing to help the medical professionals in their efforts to detect brain tumour through the use of imaging techniques.