83 Filtering Images (to Improve SNR)
Filters also exist with no other purpose but to improve the overall signal-to-noise ratio (SNR) of an image. This class of filter is normally k-space based, with one approach being to remove data from the corners of k space where a large portion of the noise contribution to an image originates. Figure 83.1A illustrates the k-space data for a single-slice acquisition, prior to application of a filter. Figure 83.1B illustrates the data after the application of an “elliptical” filter (so designated by one manufacturer), designed to reduce noise with minimal compromise in edge definition. The simple application of this or similar filters can increase SNR by 20%. Depending on vendor, the operator may or may not have control over the application of any filters, whether they be k-space or image based. Most user selectable, k space-based filters must be specified prior to image acquisition (as they are applied to the raw data).