Investigating Dose Sensitivity to Landmark Error and Clinical Workflow Practices Following Lung SBRT.
Abstract
Purpose
Stereotactic Body Radiation Therapy (SBRT) of the lung can cause large deformations which make accurate dose accumulation for retreatment challenging. This study investigates the effect of different deformable image registration (DIR) methods on dose error.
Methods
Ten patients treated with lung SBRT (45-60Gy/4-5Fx) had up to 30 landmark points defined within 5cm of the planning target volume (PTV) by physician consensus on planning and 12-month follow-up CT. Two commercially available DIR software systems were used to rigidly register the images by two starting locations: spine and carina. DIR was then performed with and without user revisions for each rigid image. Linear mixed effects modeling in R was conducted to determine significant interactions with relative dose error among software system, rigid starting location, registration method, landmark error, and binned 1cm distances 0 to 5 cm from PTV.
Results
The absolute average dose error for system 1 and system 2 across all points was 4.28±10.23Gy and 4.2±9.94Gy for the rigid, 2.23±9.77Gy and 2.78±9.76Gy for the default, and 2.23±7.46Gy and 2.43±9.64Gy for the user adjusted registrations. Main effects for the system and rigid starting location all had p-values > 0.093, while distance from PTV and landmark error were significant with p-values <0.001. Interactions among the distance from PTV, landmark error and registration type were significant with p-values < 0.05 to < 0.01.
Conclusion
No significant difference was seen between software systems or rigid registration starting locations. Relative dose errors decreased as distance from PTV increased in accordance with dose fall off. For distance bins further from the PTV, greater landmark error led to greater dose error which was mitigated through user revision. A clinical workflow utilizing user adjusted DIR could reduce dose error particularly at increased distances from the PTV for challenging retreatment dose accumulation.