Evaluation value of single-phase images from photon-counting CT-based low-dose pancreatic dynamic volume perfusion scanning for pancreatic cancer imaging

Objective This study aims to evaluate the image quality of single-phase images in low-dose pancreatic dynamic volume perfusion computed tomography (VPCT) scans based on photon-counting computed tomography (PCCT) and assess their feasibility for morphological evaluation of pancreatic ductal adenocarcinoma (PDAC). Methods From April to August 2024, participants suspected of PDAC were prospectively enrolled and underwent preoperative pancreatic VPCT scans. A total of 55 patients with pathologically confirmed PDAC were included, and radiation doses were recorded and calculated. Perfusion time-attenuation curves were analyzed, and optimal single-phase images of pancreatic parenchymal phase and portal venous phase were extracted from T3D, 55 keV, and 70 keV virtual monoenergetic images (VMIs) for clinical diagnosis and morphological evaluation of PDAC. Subjective scoring of images was performed, and CT values of lesions, normal pancreatic parenchyma, peripancreatic vessels, and abdominal wall fat were measured. The standard deviation of the CT value of abdominal wall fat was used as the image noise value. Signal-to-noise ratios (SNRs) of lesions, normal pancreatic parenchyma, and peripancreatic vessels, as well as contrast-to-noise ratios (CNRs) of lesions and peripancreatic vessels were calculated. Inter-observer agreement of subjective evaluations was evaluated using Cohen's kappa coefficient. One-way repeated measures ANOVA and Friedman H tests were used to compare the quantitative evaluation indicators and subjective scores among groups. Results The VPCT scanning achieved an average effective radiation dose of (18.0±4.9) mSv. The single-phase images of pancreatic parenchymal phase and portal venous phase extracted from reconstructed images met clinical diagnostic quality standards, with all subjective image quality scores above 3 points. In the single-phase images of pancreatic parenchymal phase reconstructed at 55 keV VMI, the CNR of PDAC lesions was 11.1±4.4, while for T3D it was 9.1±3.7 and for 70 keV VMI it was 8.0±3.2. In images of portal venous phase reconstructed at 55 keV VMI, the CNR of PDAC lesions was 6.3±3.0, while for T3D it was 5.7±3.0 and for 70 keV VMI, it was 4.9±2.7. Single-phase images reconstructed from 55 keV VMI showed the highest CNR of PDAC lesions (P<0.001), along with the highest SNR of normal pancreatic parenchyma and peripancreatic vessels, the highest CNR of peripancreatic vessels, and optimal subjective contrast scores of PDAC lesions. There was no significant difference in image noise between 55 keV VMI and T3D reconstructed single-phase images [parenchymal phase: (8.3±2.1) vs. (8.2±2.1), portal venous phase: (8.4±2.1) vs. (8.3±2.2)] (P=0.599, 0.683). Both had higher image noise than that of 70 keV VMI (parenchymal phase: 7.1±1.9, portal venous phase: 7.3±1.8) (P<0.001). Conclusion Single-phase images obtained from perfusion scanning PCCT-based low-dose pancreatic dynamic VPCT imaging protocols can be used for PDAC imaging and morphological evaluation. The 55 keV VMI reconstruction can further optimize the image quality of PDAC.