Deep Convolutional Neural Networks for Brain Tumour Detection and Analysis

Analysing brain tumour with no human intervention is considered as a vital area of research. However, this has been successfully achieved using deep convolutional neural networks (DCNNs). They have performed exceptionally well in solving computer vision problems and many others such as visual object recognition, detection and segmentation. It is used in detecting the brain tumour by optimising the brain images using segmentation algorithms which are highly resilient towards noise and cluster size sensitivity problems with automatic region of Interest (ROI) detection. One of the main reasons choosing DCNNs is due to its high accuracy and it is not necessary to perform manual feature extraction in these networks. In this research paper we present an extensive review on how convolutional neural networks (CNNs) techniques are applied in brain magnetic resonance imaging (MRI) analysis, their architectures, pre-processing, data-preparation and post-processing strategies. Analysing how different CNNs architectures have evolved, discuss their strategies and examine their outputs is the primary aim of this paper. Finally, we present the future of CNNs in which we mention few research directions in the coming years.