Noise Model Selection for Multichannel Diffusion-Weighted MRI

We examine the use of various diagnostics for model choice for multichannel diffusion-weighted MRI, which is important for inferring the correct tractography, as noise properties can differ between reconstruction techniques and scanners. These are calculated for image data obtained under various different settings of a Philips Achieva 3T scanner. A simulation study was carried out which showed these to be reasonably effective at identifying the true model.