Room: Exhibit Hall | Forum 5
Purpose: To quantify the accuracy of positional correction vectors obtained from TOPAS-simulated orthogonal proton radiographs of a patient head.
Methods: A clinical x-ray CT image set of a patient head was misaligned by 11 random positional offsets. The offset CT images were resampled back into the DICOM space of the original CT. X-ray and proton digitally reconstructed orthogonal radiographs (X-DRRs, p-DRRs) were generated for each of the 11 offset scenarios. X-DRRâ€™s of the offset CTs were generated with a commercial treatment planning system. The p-DRRâ€™s were simulated using the TOPAS model of a pCT radiography system currently under construction. The simulated detector data was fed into an iterative most reconstruction algorithm to produce the p-DRRs. Four experienced radiological technicians used a clinically available alignment software to obtain the correction vectors from each of the 30 DRR image sets. Correction vectors for both X-DRRâ€™s and p-DRRâ€™s were averaged and compared to the known misalignments from each offset CT image sets.
Results: The average alignment errors observed using the X-DRRâ€™s in the Lt/Rt, Sup/Inf, Ant/Post directions were -0.002cm, -0.004cm, and 0.001cm, with standard deviations of 0.022cm, 0.019cm and 0.016cm, respectively. Translational errors observed with p-DRRâ€™s were -0.011cm, 0.010cm, and 0.016cm, with standard deviations of 0.033 cm, 0.038 cm and 0.037 cm, respectively.
Conclusion: Whereas the primary benefit of proton radiography is the pre-treatment verification of water-equivalent proton pathlength in beam direction, these result show that 2D alignment accuracy and precision of p-DRRs for head alignments is comparable to that of X-DRRs.