Using measurements of the spatial SNR to optimize phase contrast X-ray imaging

X-ray phase contrast imaging is a measurement task which is challenging to optimize, because many physical effects determine signal and noise. If we describe the detail visibility by the spatial signal to noise ratio, SNR(u), we can optimize an imaging setup by maximizing its SNR(u). We propose a measurement method for the spatial SNR which is suitable for this purpose. It consists of measuring a series of images from which the spatial SNR is calculated. This allows a convenient and exact optimization of the SNR that does not rely on theoretical simplifications and is not specific to X-ray imaging. We demonstrate the measurement method for the example of choosing the optimal geometrical magnification for cone-beam inline X-ray phase contrast. Additionally, we propose the use of a known signal reconstruction method - the Wiener Deconvolution - to improve the detail visibility by post-processing images within the limits given by the measured SNR(u). As the SNR(u) gives the degree of this improvement, we derive a measure for the effective spatial resolution from the SNR(u).