Optimized Signal-to-Thickness Correction for Photon-Counting X-ray Detectors in Computed Tomography

Photon-counting X-ray detectors offer high detection efficiency, electronic noise suppression, and energy discrimination, enabling advanced spectral imaging in X-ray computed tomography (XCT). However, their practical use is challenged by non-linear pixel responses and temporal instabilities such as polarization effects, which degrade image quality and introduce severe ring artifacts to a reconstructed volume dataset. While conventional flat-field correction assumes a static detector response and fails to compensate for temporal changes, this work proposes an optimized approach that combines pre- and post-scan reference data using different weighting strategies to mitigate temporal effects without requiring additional reference measurements during an XCT scan. At the sam time, by utilizing a signal-to-thickness calibration (STC) over a conventional background equalization, non-linear pixel responses are accounted for as well. First, the temporal stability in the open beam image over roughly four hours was evaluated, and weighting functions were optimized to minimize the standard deviation in projection images. Subsequently, experimental XCT scan data of a material phantom was corrected using these weighting functions and the reconstructed data was assessed in terms of ring artifacts and contrast-to-noise ratio (CNR). Results show that weighted interpolation between pre- and post-scan references can significantly improve ring artifacts and CNR compared to single-reference corrections, with an exponential weighting function achieving the highest increase in CNR. These findings demonstrate that a temporally weighted STC correction is an effective approach to improve image quality in photon-counting XCT applications without the necessity of additional reference images and repositioning of the specimen during an XCT scan.