Empirically determined spatial properties of noise in prostate diffusion weighted imaging

Motivation: Development of quantitative QC methods for prostate DWI is important in clinical practice, necessitating reliable estimates of spatially heterogenous noise. Goal(s): To demonstrate a method for empirical noise measurement and examine suitability of several regions of interest for DW image-based noise estimation. Approach: In data obtained at 3T with an endo-rectal coil, we use raw noise measurements to simulate pure-noise k-space datasets that are post-processed to yield noise maps. Results: Good agreement is found between noise levels measured from noise maps in the prostate and the estimates of noise produced from the internal obturator muscles and bladder ROIs in high b-value DWI. Impact: In prostate DWI, the internal obturator muscle and bladder regions in the high b-value images can be used to estimate noise levels in the prostate tissue with low bias. This can facilitate development of quantitative QC for prostate DWI.