Verification of MLC based real‐time tumor tracking using an electronic portal imaging device

Abstract

Purpose:The authors have developed a novel technique using an electronic portal imaging device ͑EPID͒ to verify the geometrical accuracy of delivery of dose-rate-regulated tracking ͑DRRT͒. This technique, called verification of real-time tracking with EPID ͑VORTE͒, can potentially be used for both on-line and off-line quality assurance ͑QA͒ of MLC-based dynamic tumor tracking. Methods: The shape and position of target as a function of time, which is assumed to be known, is projected onto the EPID plane. This projected sequence of apertures as a function of time ͑target motion͒ is then used as the reference. The accuracy of dynamic MLC tracking can then be assessed by how well the delivered beam follows this projected target motion without the use of a physical moving phantom. The beam apertures controlled by DRRT ͑aperture motion͒ is detected by the EPID as a function of time. The aperture motion is compared to the target motion to evaluate tracking error introduced by DRRT. The accuracy of VORTE was measured using film measurements of ten static fields. The VORTE for dynamic tumor tracking was tested with several target motions, including ͑1͒ rigid-body two-dimensional ͑2-D͒ cyclic motion in the superior-inferior direction with various period and amplitude; ͑2͒ the above 2-D cyclic motion plus cyclic deformation; and ͑3͒ 2-D cyclic motion with both deformation and rotation. For each target motion, the controlled aperture motion resulting from DRRT was acquired at ϳ8 Hz using EPID in the continuous-acquisition mode. Leaf positions in all captured frames were measured from the EPID and compared to their expected positions. The passing rate of 2 mm criteria for all leaves from all frames was calculated for each of the four patterns of tumor motion. Additionally, the root-meansquare ͑RMS͒ deviations of the centroid of the apertures between the designed and delivered beams were calculated for all three cases. Results: The accuracy of MLC-leaf position determination by VORTE is 0.5 mm ͑1 standard deviation͒ by comparison to film measurements. With DRRT, the passing rates using the 2 mm criteria for all acquired frames are 100% for the 2-D displacement, 99% for the 2-D displacement with deformation, and 88% for the 2-D displacement combined with both deformation and rotation. The RMS deviations are 0.6 mm for the 2-D displacement, 1.0 mm for the 2-D displacement with deformation, and 1.1 mm for the 2-D displacement combined with both deformation and rotation. Conclusions: The VORTE can measure the accuracy of MLC-based tumor tracking without the necessity of employing a moving phantom. Moreover, it can be used for complex target motion ͑i.e., 2-D displacement combined with deformation and rotation͒ that is difficult to create with physical moving phantoms. Therefore, the VORTE and the novel QA process illustrated by this study have a great potential for verifying real-time tumor tracking.

Feasibility of predicting tumor motion using online data acquired during treatment and a generalized neural network optimized with offline patient tumor trajectories

“…In image-guided radiotherapy systems (IGRT), either the aperture of the radiation beam [1][2][3][4][5][6][7][8][9][10] or the patient couch 11 can be adjusted to match the detected tumor motion. The Cyber-Knife 12 radiosurgery system (Fig.…”

“…Tracking with gantry-mounted linear accelerators has been limited to research using phantoms. [2][3][4][5][6][7][8][9][10] In most of these studies, system latency due to image acquisition, image saving and reading, tumor detection and aperture adjustment has been identified as one of the challenges for real-time adaptive treatment. 6 Latencies ranging from 50 to 1400 ms have been reported.…”

“…However, it suffers from prolonged treatment time due to repeated verifications before each beam delivery and can only treat small tumor volumes. Tracking with gantry‐mounted linear accelerators has been limited to research using phantoms . In most of these studies, system latency due to image acquisition, image saving and reading, tumor detection and aperture adjustment has been identified as one of the challenges for real‐time adaptive treatment .…”

A comprehensive study of the mechanical performance of gantry, EPID and the MLC assembly in Elekta linacs during gantry rotation

“…[7][8][9] Moreover, linear accelerator (linac) rotation can affect gantry-mounted accessories such as the electronic portal imaging device (EPID), since the EPIDsupporting arm is not mechanically perfect and rigidly attached. With the growing application of EPIDs in preand post-treatment dosimetry verification, [10][11][12][13][14][15] realtime dosimetry verification 16,17 and real-time tumour tracking for intrafraction motion management in modern radiotherapy, [18][19][20] it is essential to characterize and account for the mechanical system imperfections of linacs.…”

Feasibility study on inverse four‐dimensional dose reconstruction using the continuous dose‐image of EPID

“…In support of this idea, the continuous dose imaging of exit radiation by EPID was recently demonstrated with sufficient dosimetric accuracy by McCurdy and Greer. 23 Han-Oh et al 24 demonstrated the feasibility of using EPID images to determine the geometrical accuracy of tumor tracking treatment with a moving MLC system.…”