Evaluation of a New Motion-correction Algorithm Using On-rigid Registration in Respiratory-gated PET/CT Images of Liver Tumors

Abstract

OBJECTIVE: The present study aimed to determine the qualitative and quantitative accuracy of the Q.Freeze algorithm in PET/CT images of liver tumors. METHODS: A body phantom and hot spheres representing liver tumors contained 5.3 and 21.2 kBq/mL of a solution containing 18F radioactivity, respectively. The phantoms were moved in the superior-inferior direction at a motion displacement of 20 mm. Conventional respiratory-gated (RG) and Q.Freeze images were sorted into 6, 10, and 13 phase-groups. The SUVave was calculated from the background of the body phantom, and the SUVmax was determined from the hot spheres of the liver tumors. Three patients with four liver tumors were also clinically assessed by whole-body and RG PET. The RG and Q.Freeze images derived from the clinical study were also sorted into 6, 10 and 13 phase-groups. Liver signal-to-noise ratio (SNR) and SUVmax were determined from the RG and Q.Freeze clinical images. RESULTS: The SUVave of Q.Freeze images was the same as those derived from the body phantom using RG. The liver SNR improved with Q.Freeze, and the SUVsmax was not overestimated when Q.Freeze was applied in both the phantom and clinical studies. Q.Freeze did not degrade the liver SNR and SUVmax even though the phase number was larger. CONCLUSIONS: Q.Freeze delivered qualitative and quantitative motion correction than conventional RG imaging even in 10-phase groups.