Respiratory Motion Correction in PET with gated PET/CT data

                        Juan Eugenio Iglesias

    The introduction of the combined PET/CT has been one of the major
    achievements in the last decade in cancer diagnose. The hybrid
    PET/CT, being two separate detectors connected in conjunction and
    sharing the same bed, can hence acquire CT images pre- or post
    acquisition of the PET data, but suffers from the fact that the
    acquisitions are not done fully simultaneously. Furthermore, while
    the acquisition of CT is done in a fraction of second, the PET
    acquisition is done over a period of approximately 20 minutes,
    during which the patient breathes, resulting in a certain
    diaphragmatic motion and hence image blurring. If the obtained
    images are then fused, they are not fully co-registered. Moreover,
    the blurred PET suffers from an increased lesion volume and a
    lower maximum SUV.  The present master's thesis evaluates a novel
    method for correcting this unwanted blurring by means of a
    software correction. CT images are acquired for several
    respiratory phases and co-registered using the deformable "demons"
    algorithm. The PET data is also gated, and the contribution from
    every phase to the image is modified with transforms calculated
    from the CT images so that they are all referred to the same phase
    (an "optimal" one, in which with most injury to the tumor and
    least to the healthy tissue might be achieved when applying
    radiotherapy).