Skip to main content
. 2023 Sep 4;96(1150):20230292. doi: 10.1259/bjr.20230292

Figure 2.

Figure 2.

The kernel method for PET image reconstruction when using an MR image to provide the feature vectors for forming the spatial basis functions. This figure shows how this change of spatial basis can result in notable noise reduction even when just a small percentage of the original data are used. 34 The original KEM 32 was based on similarities of 3-point time–activity curves for each voxel when deriving the spatial basis functions. KEM, kernel method; MLEM, maximum likelihood–expectation maximisation; PET, positron emission tomography.