Generative Adversarial Networks in Neuroimaging

Abstract

A brain tumor is a growth of abnormal cells in the brain or near it. There are different types of brain tumors which can affect the brain and it can result in frequent strong headaches, blurred vision, and seizures. Recent studies show that cases of brain tumors are becoming more common in both genders, with a predicted 1.5% increase by 2030. To prevent this, deep learning has been introduced in the field of neuroimaging to predict the brain tumor in early stage. The deep learning model uses magnetic resonance imaging (MRI) images of the brain to detect any kind of tumor. But the challenging part is getting a suitable and sufficient dataset. If the dataset is not sufficient, it leads to overfitting issue. To tackle this, the authors have proposed a technique using deep convolutional generative adversarial network (DCGAN) to enhance the dataset which will improve tumor detection accuracy.