Generalizable signal-to-equivalent thickness calibration enhances soft tissue contrast in diagnostic CBCT

Abstract Cone beam computed tomography (CBCT) X-ray imaging suffers from poor soft tissue contrast, HU instability, and beam hardening artifacts. The generalizability of the commonly used signal-to-equivalent thickness calibration (STC) method is limited by scattering in large field-of-view (FOV) settings. We therefore introduce scatter-reduced STC (SR-STC) to improve robustness across varying scatter conditions. For STC, multiple acrylic thicknesses were imaged to derive pixel-wise calibration curves relating measured beam intensity to material thickness. In the SR-STC approach, a scatter intensity image was subtracted from each thickness projection prior to calibration. Projection data was processed prior to reconstruction. A spectral CT phantom with tissue-equivalent targets was scanned with varying scattering conditions utilizing a clinical head and neck CBCT device. SR-STC performance was evaluated against FFC, STC, and the vendor reconstruction pipeline. In addition, a linear regression analysis was performed using clinical CT to study HU agreement with clinical benchmark. The introduced SR-STC method improved soft tissue contrast across almost all imaging parameters compared to FFC and vendor reconstructions with the standard STC providing a smaller improvement. For most high contrast targets, SR-STC provided the highest contrast. SR-STC demonstrated best HU value agreement with helical CT. The SR-STC provided high subjective image quality, reducing beam hardening-related artefacts. The conventional STC method did not perform as well across varying imaging settings. This highlights the need for a scatter-reducing act on the common STC method. The introduced SR-STC method was found to be an easy-to-implement calibration method with the potential to improve HU accuracy, soft tissue contrast, and to mitigate beam hardening artifacts in diagnostic CBCT devices.