Intrafractional tracking accuracy in infrared marker-based hybrid dynamic tumour-tracking irradiation with a gimballed linac.

Abstract

[Purpose]To verify the intrafractional tracking accuracy in infrared (IR) marker-based hybrid dynamic tumour tracking irradiation ("IR Tracking") with the Vero4DRT. [Materials and methods]The gimballed X-ray head tracks a moving target by predicting its future position from displacements of IR markers in real-time. Ten lung cancer patients who underwent IR Tracking were enrolled. The 95th percentiles of intrafractional mechanical (iE^[95]_M), prediction (iE^[95]_P), and overall targeting errors (iE^[95]_T ) were calculated from orthogonal fluoroscopy images acquired during tracking irradiation and from the synchronously acquired log files. [Results]Averaged intrafractional errors were (left–right, cranio-caudal [CC], anterior–posterior [AP]) = (0.1 mm, 0.4 mm, 0.1 mm) for iE^[95]_M, (1.2 mm, 2.7 mm, 2.1 mm) for iE^[95]_P , and (1.3 mm, 2.4 mm, 1.4 mm) for iE^[95]_T. By correcting systematic prediction errors in the previous field, the iE^[95]_P was reduced significantly, by an average of 0.4 mm in the CC (p < 0.05) and by 0.3 mm in the AP (p < 0.01) directions. [Conclusions]Prediction errors were the primary cause of overall targeting errors, whereas mechanical errors were negligible. Furthermore, improvement of the prediction accuracy could be achieved by correcting systematic prediction errors in the previous field.