Abstract
Deep learning, a subset of artificial intelligence in computer science, has become crucial in understanding the structural and functional connectivity of the human brain connectome. It offers novel insights into the comprehensive probabilistic modeling of the brain. This study aims to provide an overview of various deep learning techniques applied to the human brain connectome and to review significant structural and functional connectivity findings using MRI images for different brain diseases. A detailed literature search was conducted using the PRISMA model across databases such as Scopus, web of Science, and PubMed. The primary search terms included “Brain,” “Connectome,” “Deep Learning,” and “Neuroimaging or MRI.” This search identified 113 relevant studies out of a total of 882. The systematic review found that deep learning algorithms are rapidly widely used in neuroscience. Traditional neural network approaches, such as convolutional neural networks (CNN), graph neural networks (GNN), and artificial neural networks (ANN), remain prevalent. These algorithms are often tailored to address specific tasks, with MRI images serving as the primary data source for brain imaging. Deep learning has significant potential to enhance the understanding of structural and functional brain models in neuroscience applications. However, several challenges must be addressed to utilize deep learning more effectively in brain mapping. Accumulating detailed data is crucial for developing intelligible DL algorithms to achieve this goal.
Keywords: Connectome, Deep Learning, Human Brain, MRI, Neuroscience, PRISMA.