Optimization-Based Motion Estimation in HR-PQCT

The imaging modality high-resolution peripheral quantitative computed tomography (HR-pQCT) enables assessment of bone mineral density and three-dimensional microarchitecture of peripheral limbs. Due to its long scan time, this modality is especially susceptible towards motion of the patient. The architecture and scanning protocol of existing scanners, which acquire only halfscans, make many existing methods of motion compensation inapplicable. In this work, an iterative motion estimation and compensation approach is proposed that is able to significantly reduce the amount of motion artifacts. This is achieved by jointly optimizing the motion parameters and reference image to minimize the data consistency error by exploiting the quasi-convex behavior of the objective functions observed near the ground truth. The proposed method was evaluated on a large collection of simulated sinograms and was able to remove motion artifacts almost completely.