TU-E-BRB-06: Best in Physics (Therapy) - Development and Experimental Validation of EPID-Based 4D Dose Reconstruction

To develop and validate an EPID-based 4D patient dose reconstruction framework accounting for linac delivery uncertainties, interfractional and intrafractional motions, and interplay effect.

Patients with fiducial markers were scanned with 4D-CT for SBRT planning. Before treatment, in-room 4D-CT was performed. Both the MLC and the tumor movements were tracked by continuously acquiring EPID images during treatment. Instead of directly using the heterogeneous transit photon fluence measured by the EPID, this method reconstructed the incident beam fluence based on the MLC apertures measured by the EPID and the delivered MU recorded by the linac. To account for the time-dependent-geometry, the incident fluence distributions were sorted into their corresponding phases based on the tumor motion pattern detected by the EPID and accumulated as the incident fluence map for each phase. Together with 4D-CT, it was then used for Monte Carlo dose calculation. Deformable registration was performed to sum up the phase doses for treatment assessment. The feasibility of using the transit EPID images for incident fluence reconstruction was evaluated against EPID in-air measurements. The accuracy of 3D- and 4D-dose reconstruction was validated by a motordriven cylindrical diode array for six clinical SBRT plans.

The average difference between the measured and reconstructed fluence maps is within 0.16%. The reconstructed 3D-dose shows 1.4% agreement in the CAX-dose and >98.5% gamma-passing-rate (2%/2mm) in the peripheral-dose. A distorted dose distribution is observed in the measurement for the moving ArcCheck-phantom. The comparison between the measured and the reconstructed 4D-dose without considering interplay fails the gammaevaluation (59%-88.9% gamma-passing-rate). In contrast, when the interplay is considered, the dose distortion phenomena is successfully represented in the reconstructed dose (>97.6% gamma-passing-rate).

The experimental validation demonstrates that the proposed method provides a practical way to reconstruct the fractional 4D-doses received by the patient and enables adaptive SBRT strategy.