Author: Gisell Ruiz Boiset, Paulo ROBERTO Costa, Luuk J Oostveen, Elsa Bifano Pimenta, Ioannis Sechopoulos, Alessandra Tomal 👨🔬
Affiliation: Radboud University Medical Center, University of São Paulo (USP), Institute of Physics, Universidade Estadual de Campinas. Instituto de Física Gleb Wataghin 🌍
Purpose: This study aimed to evaluate the precision and accuracy of volume measurements for solid nodules (SNs) and ground-glass opacities (GGOs) in lung images acquired using energy-integrating CT (EICT) and CdZnTe-photon counting CT (PCCT) systems, reconstructed with deep learning-based methods
Methods: Four 3D-printed SNs and four handmade GGOs were embedded in a patient-specific lung phantom. Ground-truth nodule volumes were established via ultra-high-resolution mCT imaging. The phantom was scanned with CTDIvol = 1.4 mGy using EICT and PCCT systems. Ten scans per protocol were performed, repositioning the phantom to mimic clinical variability. Images were reconstructed using DLR at normal resolution (NR) and high resolution (HR). Comparisons focused on EICT-NR and PCCT-HR. Accuracy was determined by the relative error (RE) compared to ground truth, and precision by interquartile range (IQR).
Results: The actual volumes of the SNs were 142, 428, 808, and 993 mm³ (SN1-SN4), while for the GGOs, the volumes were 1161, 1169, 1205, and 1848 mm³ (GGO1-GGO4). The PCCT-HR reconstructions demonstrated better accuracy compared to the EICT-NR reconstructions, with medians RE% ranged, respectively, from -0.5% to 2.7% and -2.8% to 2.4% for EICT-NR. Both systems tended to underestimate volumes, although EICT-NR showed slight overestimations for SN4 (993 mm³) and GGO4 with RE% values of -1.5% and -2.8%, respectively, whereas PCCT showed smaller errors (-0.5% and -0.4%, respectively). In terms of precision, PCCT-HR also outperformed EICT-NR, showing smaller IQRs that varied from 0.3 to 3.0, while EICT-NR presented IQRs ranging from 0.2 to 5.3.
Conclusion: The PCCT system with DLR demonstrated significantly higher accuracy and precision compared to EICT. These results highlight the potential of PCCT for improving pulmonary nodule volume assessment in clinical practice.