Development of a Virtual Photon-Counting Micro-CT Imaging Platform for Preclinical Cancer Studies
Micro-CT imaging studies in mouse cancer models are vital for developing therapeutics. Multi-energy CT imaging with a photon-counting detector (PCD) improves material separation in cancer studies involving nanoparticle-based contrast agents and combination therapy studies involving radiation therapy and/or chemotherapy. However, achieving high quality imaging with photon-counting CT (PCCT) requires scan parameter optimization, which may not always be possible during in vivo cancer studies due to the need to limit radiation dose in live mice. In silico simulation of CT imaging allows extensive scan parameter tuning to improve image quality, but this method has not yet been adapted for mouse cancer studies. This work details our efforts towards an in silico PCCT pipeline for preclinical cancer studies that includes a digital phantom of a mouse with head and neck squamous cell carcinoma (HNSCC). We enhanced the mouse whole body (MOBY) phantom by transferring vasculature from a high-resolution mouse scan and adding a tumor model from CompuCell3D. Our PCCT simulation software models the whole imaging chain and includes a model for spectral distortion. A polynomial-based correction of the distorted spectral response was calibrated using real PCCT scans of known materials. PCCT simulations of the enhanced MOBY phantom with and without polynomial correction were compared to the ground truth using tumor metrics from material maps. Polynomial correction only improved root mean square error for 3 out of 4 known materials, suggesting a better correction is needed. Our simulations of MOBY with a tumor containing iodine and barium nanoparticles reproduced noise and material cross-contamination seen in real PCCT scans of mice with HNSCC. The polynomial correction improved the accuracy of 8 out of 10 tumor metrics across both the iodine and barium maps. Future work will focus on data-driven methods to improve the simulated spectral response, using tumor metrics for imaging parameter optimization, and confirming that image quality improvements in simulation translate to in vivo imaging.
