Purpose: 4DCBCT phase images acquired during treatment (DT) can have poorer image quality due to undersampling of kV-projections (55-90 projections/phase). The suitability of using during treatment average images (DT-4DAvg) for IGRT setup correction between SBRT lung treatment beams needs investigation.
Methods: DT-4DCBCTs from four SBRT lung patients with a range in MUs (900-1618), IGTV volume (0.5-7.6cc) and tumor excursion (6.4-15.1mm) were evaluated. Lung motion phantom DT-4DCBCTs were acquired to isolate the influence of kV-projection number on DT-4DAvg images. DT-4DAvg images were compared to DT-3DCBCTs reconstructed using the same DT-4DCBCT kV-projections. Since kV-projections acquired in DT-4DCBCT have greater sampling than pre-treatment 3DCBCTs the reconstructed DT-3DCBCT can serve as a reference. This same comparison was performed using pre-treatment phantom 4DCBCT kV-projections to evaluate the impact using different reconstruction algorithms (4D vs. 3D) had on comparison results. For patient images, target and lung HUs and IGTV shifts between the two CT datasets were compared. IGTV shifts were determined by (1) couch correction values from XVI registration and (2) deformable image registration. For phantom images, HUs, image profiles and IGTV diameter were compared.
Results: DT-4DCBCTs with a range of kV-projections were generated (549-911). DT-4DAvg images with less kV-projections (549) had the largest differences (-105 and -50 for patient and phantom images) in target HU compared to DT-3DCBCT. Lung HU differences were fairly consistent, independent of kV-projection number (-20 to 60) for phantom, patient, and pre-treatment CBCTs. Between datasets, phantom image profiles and IGTV diameters (max difference 1.1 mm) were similar. The largest IGTV shift difference observed in the lateral, longitudinal, or vertical direction was only 1.4mm.
Conclusion: Image profile, target diameter, and IGTV shift agreement demonstrates DT-4DAvg images can be safely used for image guidance of SBRT lung patients. Target HU differences are significant for small MU beams with less kV-projections.