Super-resolution in 4D positron emission tomography

Respiratory synchronized acquisitions lead to noisy images. Super-resolution techniques deal with the enhancement of several slightly different low-resolution images into a single high-resolution image. A maximum a-posteriori (MAP) super-resolution algorithm has been implemented and applied to respiratory gated PET images for motion compensation. The algorithm was tested on a GATE simulated datasets. It consisted of 8 frames of the NCAT phantom with lesions between 15–22mm placed throughout the lungs. Images were reconstructed using the OPLEM algorithm. Super-resolution was performed on the gated frames through a MAP algorithm, using a Huber prior as a regularization term to ensure convergence. The optimization of the function yielded by the MAP method was performed through a steepest descent algorithm. Motion fields were recovered using a previously presented elastic registration algorithm. Image enhancement was assessed by estimating signal to noise ratio (SNR) and contrast in regions of interest.