Quantitative Quality Assurance of 4DCT Volumetric and Motion Accuracy Using the Quasar™ Respiratory Motion Platform and a Custom Phantom
Abstract
Purpose
Implement a Four-Dimensional CT (4DCT) procedure using a QUASAR™ Respiratory Motion Platform and a custom sphere phantom to quantitatively evaluate the volumetric reconstruction accuracy and spatial location tracking precision of 4DCT acquisitions.
Methods
A Siemens SOMATOM go.Open Pro with Varian RGSC was utilized for 4DCT acquisitions. One static CT scan was acquired as a baseline followed by a continuous (helical) and intermittent (axial) 4DCT. A sphere phantom of known volume was securely attached to a QUASAR™ Respiratory Motion Platform to simulate tumor motion during respiration. During 4DCT acquisitions, the QUASAR moves the sphere phantom 2 cm superiorly and 2 cm inferiorly (4 cm total). Multiple respiratory patterns and breathing rates (4-18 BPM) were tested to evaluate the system’s sensitivity. A workflow using imaging software (MiM Software Inc.) automatically contours the sphere on each of the 10 respiratory phases per 4DCT acquisition and the static baseline CT scan. These contours provide the volume and centroid coordinates of the sphere phantom. A Python script was developed to further automate analysis. Measurements were repeated to ensure reproducibility.
Results
The 4DCT QA workflow was successfully implemented and is time efficient while providing clinically relevant data. Analysis of the scans demonstrated that volumetric reconstruction accuracy was strongly dependent on breathing rate, the type of 4DCT scan, and the phases within the scan (some phases were repeatedly outliers in the data). The scans demonstrated poor performance on extreme breathing rate settings (4 BPM or 18 BPM). The range of motion showcased high accuracy producing less than 1 mm of error from the known value.
Conclusion
The developed 4DCT QA procedure provides an effective framework to evaluate 4DCT reconstruction accuracy and highlights system limitations. Additionally, the QA is adoptable using vendor-based motion phantoms, simple custom phantoms, and open-source analysis tools.