Click here to

Session: Cone-Beam CT [Return to Session]

Influence of Threshold Calibration On CT Image Quality of a Photon-Counting Detector

P Rodesch*, D Richtsmeier, L Zalavari, M Bazalova-Carter, University of Victoria, Victoria, BCCA,


SU-H300-IePD-F8-5 (Sunday, 7/10/2022) 3:00 PM - 3:30 PM [Eastern Time (GMT-4)]

Exhibit Hall | Forum 8

Purpose: To assess how updating the energy threshold calibration of a photon-counting detector (PCD) impacts conventional and K-edge CT image quality.

Methods: Two different energy threshold calibrations were implemented in the digital-to-analog converter of a Redlen Technology PCD composed of 16 submodules with 24x32 pixels each. Both calibrations (denoted as A provided by the manufacturer and B developed here) were measured per pixel with various K-edge techniques using different K-edge materials and data processing. The calibrations were evaluated with monoenergetic sources to quantify the spectral accuracy and pixel homogeneity. The ratio mean over variance (RVM) of counts in an air projection was acquired to evaluate the submodule homogeneity. CT acquisitions with 5 bins were performed to evaluate image quality with a 25/40/50/65/80/120 keV threshold configuration. The noise power spectrum (NPS) was measured from two consecutive water phantom CT reconstructions. K-edge subtraction images were computed from CT acquisitions with gadolinium and gold inserts. The K-edges were detected by thresholding the subtraction images at +3 standard deviation (measured in the water background). The signal-to-noise ratio (SNR) was then measured.

Results: The monoenergetic measurements showed a difference of +0.7 keV and -1.0 keV from theoretical values for calibrations A and B, respectively. From these measurements and the RVM, it was shown that calibration B had better pixel and submodule homogeneity. This had the benefit in that the NPS presented less variation in low spatial frequencies for calibration B. In K-edge images, calibration B had a greater SNR (+2.6%) and corrected for ring artifacts.

Conclusion: We presented an easy-to-implement threshold calibration with a similar spectral accuracy to the manufacturer calibration. However, our calibration resulted in better pixel homogeneity improving image quality metrics in CT images. Calibration B is easy to implement and enables the monitoring of the long-term behavior of a PCD.


Calibration, Semiconductor Detectors, Image Analysis


IM- Cone Beam CT: Dual-energy and spectral

Contact Email