Non-stationary noise estimation in accelerated parallel MRI data (MISS Workshops 2016)

Poster presented at the Medical Imaging Summer School'2016, Favignana, Italy

Abstract:

The aim of this study is to retrieve spatially variant noise patterns from accelerated parallel MRI data using only a single image. Variance-stabilizing transformations (VSTs) for noncentral Chi data are derived: (1) an analytic model, and (2) a numerical model to improve the performance for low signal-to-noise ratios (SNRs). The VSTs generate Gaussian-like distributed variates from noncentral Chi data. The noise patterns are estimated then using Gaussian homomorphic filter.