Calibration of a robotic cone-beam CT scanner to Hounsfield Units

Cone-Beam Computed Tomography (CBCT) has become an essential imaging tool in medical applications, particularly in radiation therapy, due to its ability to generate high-resolution volumetric images with relatively low radiation exposure. However, a persistent challenge with CBCT is its inability to accurately represent Hounsfield Units (HU), primarily due to artifacts such as beam hardening, scatter radiation, and noise. These limitations hinder its use for precise tissue density measurements, a critical aspect of treatment planning. This study addresses this challenge by developing a robust calibration methodology that aligns CBCT intensity values with those of conventional Computed Tomography (CT) scans. The calibration process involves scanning a cylindrical phantom embedded with inserts of known electron densities, segmenting these inserts, and deriving a calibration curve to convert CBCT pixel intensities into standardized HU values. Additionally, registration techniques are applied to spatially align CBCT and CT images, enabling direct comparison and validation. Image subtraction and data matching methods further assess calibration accuracy, ensuring that discrepancies between CBCT and CT datasets are minimized, and it is tested using a thorax phantom, which simulate complex anatomical structures. The results demonstrate that calibration significantly enhances the accuracy of CBCT imaging, particularly in high-density regions where HU deviations are most pronounced. By applying the calibration curve, tissue differentiation is improved, leading to more precise dose calculations and treatment planning in radiotherapy. These findings highlight the potential of CBCT calibration to bridge the gap between CBCT and CT imaging, making CBCT a more viable tool for quantitative medical imaging. Ultimately, this study contributes to the ongoing development of CBCT technology, enhancing its clinical utility in radiation therapy and diagnostic imaging.