Magnetic Resonance Brain Imaging
Modeling and Data Analysis Using R

Appendix A: Smoothing techniques for imaging problems

This appendix intends to present issues in non-parametric regression and image denoising that are relevant for the analyses of neuroimaging experiments presented in Chapters 4, 5 and~6 of the book. The focus is on a class of edge-preserving, i.e., structural adaptive, methods that are based on the Propagation-Separation approach. We detail how these relate to non-parametric kernel smoothing (or filtering), provide algorithmic details and discuss their properties.