Comparing Image Quality in Phase Contrast sub$\mu$ X-Ray Tomography - A Round-Robin Study

How to evaluate and compare image quality from different submicrometer (sub$\mu$) CT scans? A simple yet clever test phantom is used for recording 13 scans in a number of commercial and some non-commercial scanners. From the resulting volume images, signal and noise power spectra are modeled for estimating spatial signal-to-noise ratio (SNR spectrum). Using the same data, a time- and object-independent transfer function (MTF) is computed for each scan, including phase contrast strength and spatial resolution (MTF). SNR and MTF are compared to transmission measurements of the same test phantom. Making SNR object-independent by normalization with respect to the object spectrum yields detection effectiveness (DE) a new measure which reveals how technical differences as well as operator-choices strongly influence scan quality for a given measurement time. Using DE both source-based and detector-based sub$\mu$ CT scanners can be studied and optimized with respect to signal detection effectiveness using the metrics which are presented in this study. Future application of this work requires a particular scan acquisition scheme which will allow for measuring 3D signal-to-noise ratios.