Generalized dynamic PET inter-frame and intra-frame motion correction - Phantom and human validation studies

Abstract

Patient motion can significantly hamper the high-resolution imaging capability of PET scanners. Frame-acquired (dynamic) PET images are degraded by inter-frame and intraframe motion artifacts that can degrade the quantitative and qualitative analysis of acquired PET data. This calls for appropriate motion-correction techniques that can considerably reduce (ideally eliminate) inter-frame and intra-frame motion artifacts in dynamic PET images. We present a novel approach called Generalized Inter-frame and Intra-frame Motion Correction (GIIMC) algorithm [1] that unifies in one framework the inter-frame motion correction capability of Multiple Acquisition Frames and the intra-frame motion correction feature of (MLEM)-type Deconvolution methods. Our method employs a fairly simple but new approach of using time-weighted average of attenuation sinograms to reconstruct individual (dynamic) frames. We also provide a mean-motion threshold for individual frames to construct a framing sequence.