Least-error projection sorting to optimize retrospectively gated cardiac micro-CT of free-breathing mice

To develop and characterize a technique for optimizing image quality by eliminating streaking artifacts in retrospectively gated microcomputed tomography (micro-CT) images of mice caused by insufficient and irregular angular sampling.

A least-error sorting technique was developed to minimize streak artifacts in retrospectively gated cardiac micro-CT images. To ensure complete filling of projection space, for each angular position, the projection acquired closest to the desired cardiac phase is used to reconstruct a volumetric image. An acrylic slanted-edge phantom undergoing cyclic motion was used to characterize the system's resolution. The phantom was scanned using a volumetric micro-CT scanner equipped with a flat-panel detector mounted on a slip-ring gantry. Projection images of the moving phantom were collected over a period of 60 s using a variety of acquisition protocols with the rotation period of the gantry ranging from 1 to 5 s. The modulation transfer function (MTF) of the reconstructed images was measured for many combinations of acquisition and reconstruction parameters. The use of the least-error technique was also demonstrated in vivo.

The motion blurring introduced into the images at physiologically significant velocities of 6 cm∕s agreed well with predicted values; limiting resolution (frequency at 10% MTF) degraded from 2.5 to 1.0 mm(-1) for a velocity of 6 cm∕s and 5 s∕rotation gantry speed. Faster gantry rotation speeds led to improved temporal resolution but the scanner's data storage and transfer rates and field of view limitations made scanning at gantry speeds faster than 2 s∕rotation impractical.

The least-error technique effectively eliminates streaking artifact caused by missing views and allows for optimization of image quality in retrospectively gated micro-CT.