Purpose: To evaluate the quality of image sets within the Ethos adaptive treatment workflow (EATW) using a vendor-provided phantom.
Methods: The QUART phantom (Varian GmbH, Switzerland) supplied with the Ethos linear accelerator is made of acrylic with a central module containing material inserts (air, Teflon, and polystyrene). The central module is sandwiched between two uniform background modules. Three image sets were used in the evaluation: planning CT (pCT), kV cone-beam-computed-tomography (CBCT) Head-iCBCT protocol scan taken during EATW, and synthetic CT (sCT) created using deformable registration of pCT and CBCT during EATW. The image quality tests of focus were CT Hounsfield Unit (HU) constancy, uniformity, and noise. HU constancy was tested by comparing average HU values in inserts on the center slice to pCT in corresponding regions of interest (ROIs). Uniformity was computed using maximum absolute HU difference of the mean HU in ROIs drawn in each cardinal direction and the center. Noise was taken as the maximum standard deviation from the uniformity ROIs, and SNR and CNR values were computed following the methodology described by the vendor.
Results: In comparison with the pCT, the maximum HU difference was 46.9±44.4 HU for the CBCT and 0.48±17.4 HU for the sCT. The uniformity value was 4.3±18.4 HU for the pCT, 33.4±34.7 HU for the CBCT, and 4.5±18.6 HU for the sCT. The noise was 14.0 HU for the pCT, 31.2 HU for the CBCT, and 14.4 HU for the sCT.
Conclusion: Across the tests performed, a consistent trend was observed that showed a recovery of the pCT image quality in the sCT. This comparability to pCT is important because the sCT is used for contouring, plan optimization, and dose calculation within the EATW. Future work includes an automated analysis quality assurance solution to incorporate a more comprehensive set of image quality tests.
Cone-beam CT, Image Analysis, SNR