An adaptive genetic algorithm for misalignment estimation (AGAME) in spiral, sequential and circular cone-beam micro-CT

The reconstruction of volumetric datasets based on micro-CT scans is a common task in every small animal imaging lab. The used reconstruction algorithms thereby rely on the exact knowledge of the scanner geometry. If this geometry is misaligned and hence not known accurately severe artifacts in terms of edge blurring and a loss in spatial resolution appear in the reconstructed images as long as no geometry calibration is performed. We propose a novel method for misalignment estimation of micro-CT scanners using an adaptive genetic algorithm. Furthermore not only the misaligned scanner geometry is estimated but also the direction vector of table movement as well as the displacement between different imaging chains within a scanner. This not only allows for the superposition of reconstructed images originating from different imaging chains or a spiral scan trajectory but also for a stitching of volumes from sequence scans. The calibration procedure does not rely on dedicated calibration phantoms, a sequence scan of a single metal bead is sufficient to calibrate the whole imaging system. To analyze the performance of the algorithm a micro-CT system with known misalignment is simulated and the accuracy of the estimated parameters is evaluated as a function of the segmentation quality of the metal beads. To assess the capabilities of the algorithm in a real world scenario the calibration of a micro-CT scanner is performed and several reconstructions using the nominal geometry and the estimated geometry are carried out. The results indicate that the proposed algorithm successfully estimates all geometry parameters as misalignment artifacts in the reconstructed volumes vanish and the spatial resolution is increased.