Program Information
Analysis of Acquisition Parameters and Pharmacokinetic Models From DCE-CT with Application to Outcome Assessment of Radiation Therapy
J Niedzielski*, M Miften , B Jones , K Goodman , D Thomas , University of Colorado School of Medicine, Aurora, CO
Presentations
SU-K-601-15 (Sunday, July 30, 2017) 4:00 PM - 6:00 PM Room: 601
Purpose: Dynamic Contrast Enhanced CT (DCE-CT) is a viable technique for outcome assessment in Radiation Therapy. We analyzed the variability of perfusion parameters from different pharmacokinetic models and DCE-CT acquisition parameters, and then derived optimal parameters for pancreatic SBRT treatment outcome assessment.
Methods: DCE-CT acquisitions were obtained using a perfusion validation phantom (Shelley Medical Imaging) and patient data for SBRT treatment of pancreatic cancer. Patlak and Extended Tofts pharmacokinetic models were implemented in Matlab. Models were fit to the DCE-CT phantom data, and tested for goodness-of-fit and reproducibility. The validation phantom was imaged with several different tumor/vasculature flow ratios (1:1, 1:4), as well as different acquisition parameters (120/80kV, 150/70mA, 1.5/3.0s acquisition sampling time) for a flow ratio of 1:1. Patient data were acquired as part of a Phase I clinical protocol for pancreatic SBRT prior to and 6 weeks after completion of treatment of primary adenocarcinoma of the pancreas were used.
Results: Patlak analysis showed KTrans values of 171.3±21.8 and 136.6±11.9 (mL/100mL/min), for repeat acquisitions for flow ratios of 1:1 and 1:4, respectively. KTrans for 80kV, 70mA, and 3.0s sampling time were: 201.1, 197.8, and 113.3 (mL/100mL/min), respectively. Patient KTrans was 20.0 at baseline and 16.3 (mL/100mL/min) at follow up.Extended Tofts model analysis showed KTrans values of 132.9±34.8 and 99.5±7.4 (mL/100mL/min), for repeat acquisitions for flow ratios of 1:1 and 1:4, respectively. KTrans for 80kV, 70mA, and 3.0s sampling time were: 182.5, 165.6, and 169.3 (mL/100mL/min), respectively. Lowering kV had the largest model fit error than lowering mA or sampling time (RMSE=186.8 vs 155.3 vs 159.8, respectively).
Conclusion: Patient data was best represented by a lower tumor/vasculature flow ratio (1:4). The Patlak model had more reproducible perfusion parameters. Lower kV and sampling time negatively affects accuracy of perfusion parameters.
Contact Email: