Evaluating Graph Signal Processing for Neuroimaging Through\n Classification and Dimensionality Reduction

2017/03/06 by Mathilde Ménoret, Ménoret, Mathilde, Nicolas Farrugia +5 · 1 citation
Neuroscience · Computer Science · Biochemistry, Genetics and Molecular Biology · #Functional Brain Connectivity Studies #Advanced Graph Neural Networks #Bioinformatics and Genomic Networks

paper · pdf · doi:10.48550/arxiv.1703.01842

Abstract

Graph Signal Processing (GSP) is a promising framework to analyze\nmulti-dimensional neuroimaging datasets, while taking into account both the\nspatial and functional dependencies between brain signals. In the present work,\nwe apply dimensionality reduction techniques based on graph representations of\nthe brain to decode brain activity from real and simulated fMRI datasets. We\nintroduce seven graphs obtained from a) geometric structure and/or b)\nfunctional connectivity between brain areas at rest, and compare them when\nperforming dimension reduction for classification. We show that mixed graphs\nusing both a) and b) offer the best performance. We also show that graph\nsampling methods perform better than classical dimension reduction including\nPrincipal Component Analysis (PCA) and Independent Component Analysis (ICA).\n

Cited by

• Comparing and Scaling fMRI Features for Brain-Behavior Prediction

Related

• A Graph Signal Processing View on Functional Brain Imaging
• A Tutorial on Graph Theory for Brain Signal Analysis
• Gradients of Connectivity as Graph Fourier Bases of Brain Activity
• Pruning Graph Convolutional Networks to select meaningful graph frequencies for fMRI decoding
• Deep neural networks on graph signals for brain imaging analysis