| 主权项 |
1. A computer-implemented method for penalized-likelihood reconstruction of a Positron Emission Tomography (PET) image, the PET image comprising one or more image voxel values, each of which is indexed by an x-coordinate and y-coordinate index, the computer including a processor, the method comprising:
generating, by the processor, a regularization function, the regularization function comprising a penalty function as a function of voxels neighboring the one or more image voxel values, a smoothing parameter resulting in non-uniform spatial resolution over a reconstructed PET image due to spatially varying sensitivities, the smoothing parameter controlling strength of the regularization function, and one or more modulation factors modulating the smoothing parameter, the modulation factors independent of the one or more image voxel values and one or more PET emission datasets, and spatially variable to compensate for the spatially varying sensitivities in the x-coordinate and y-coordinate index; reconstructing, by the processor, the PET image from the PET emission dataset using the regularization function; and calculating, by the processor, each of the modulation factors as a mean of a first sensitivity factor associated with a first voxel in the PET voxel dataset and a second sensitivity factor associated with a second voxel in the PET voxel dataset. |
