Respiratory motion correction with an improved demons algorithm for PET images

Respiratory motion is a major factor that affects the quality of PET images of the thoracic area. The diaphragm moves about 15-20 mm due to respiratory motion, which substantially degrades the effective spatial resolution of PET. In this paper, a gated acquisition method is used to correct the motion effects. In this method, an improved demons algorithm is proposed to align the gated images. The experimental results show that the quality of PET images is significantly improved when using our improved method and the proposed method has a faster convergence rate than the original demons algorithm.