the same tissues. For plans A and B, the dose of a specific point in the CTV shell received in the treatment would be the sumof the phase-weighted doses of the corresponding points in the three different phases. The minimal dose, the average dose, andthe DVH of the CTV shell were then calculated from the doses of the 18 points.

Results: The minimal dose points were at the inferior and superior portions of the CTV shell. For the CTV shell 5 mm fromthe GTV surface, the minimal doses of the 5 patients were 98%�2%, 92%�17%, and 72%�25% of the prescription dose forplans A, B and C, respectively. For the CTV shell 10 mm from the GTV surface, the minimal doses were 67%�5%, 72%�15%,and 13%�10% of the prescription dose for plans A, B and C, respectively. Plan A showed relative uniform doses within theCTV shells, and minimal variation among patients.

Conclusions: Using the ITV planning technique (plan A) or a 10 mm expansion around the GTV (plan B), the CTV shell 10mm from the GTV surface would generally receive at least 60% of the prescription dose (biological equivalent dose of about50 Gy), which is sufficient to control microscopic diseases. However, using gating treatment based on the plan C could yieldsubstantial underdosing to the CTV. Caution should be taken to implement gating techniques for stereotactic lung radiotherapy.

Author Disclosure: J. Jin, None; Q. Chen, None; M. Ajlouni, None; S. Li, None; B. Movsas, None.

2720 Real-Time Respiration Monitoring via IMRT Treatment Beam and 4DCT

W. Lu, K. J. Ruchala, M. Chen, Q. Chen, G. H. Olivera

TomoTherapy Inc., Madison, WI

Purpose/Objective(s): Real-time knowledge of intra-fraction motion is essential for four-dimensional (4D) radiotherapy.Surrogate-based and internal-fiducial-based methods may have drawbacks such as false correlations, being invasive, deliveringextra patient radiation, or require complicated hardware and software. We developed a real-time, non-surrogate, non-invasivemethod to monitor respiratory motion during radiotherapy treatments. This method directly utilizes the treatment beam and thusimposes no extra radiation to the patient.

Materials/Methods: This method requires a real time detector system and a 4DCT image. The basic idea is to correlate thereal-time measured detector signal from the treatment beam with the pre-calculated signals assuming that the beam passesthrough the different phases of the 4DCT image. The on-line processes only involve detector signal readout, and 1D correlationof the measured signal with the pre-calculated signals. The respiration phase is determined by the peak correlation. Real-time3D internal motion can be reconstructed by combining the determined respiration phase with the 4DCT.

The method was tested with extensive simulations based on a 4DCT of a lung cancer patient. Three different TomoTherapydelivery sinograms, either optimized (intensity modulated) or not, were used. Each sinogram contains around 1000 activeprojections. Three arbitrary breathing patterns, either regular or irregular, and two dose levels, 2Gy/fraction and 2cGy/fraction,were used to study the robustness of this method against detector quantum noise and patient scatter.

Results: Table 1 summarizes the results of our simulations. For the 2Gy/fraction simulations, the respiration phases wereaccurately determined in real-time for most projections of the treatment, except for a few projections at begin or end of thesinogram where beam intensities were extremely low. At 2cGy/fraction dose level, the method can still determine the respirationphase very well with only about 5% of projections having errors greater than 1 phase (0.5 second).

Conclusions: It is demonstrated that our method can monitor the respiratory motion within �1 phase in real time. This methodcan be easily implemented in any radiotherapy machine with high speed detector system. The motion information obtained canbe used to either verify or correct the treatment delivery in real-time. This technique can also be applied in other systems suchas orthogonal X-ray systems, although in those cases it would cause additional non-treatment radiation.

Author Disclosure: W. Lu, None; K.J. Ruchala, None; M. Chen, None; Q. Chen, None; G.H. Olivera, None.

2721 Intra- and Inter-fraction Variations in Free Breathing During Curative Radiotherapy for Lung Cancer

T. Juhler-Nøttrup1, S. Korreman1, L. Aarup1, H. Nystrom1, M. Olsen1, A. Pedersen2, L. Specht2

1Rigshospitalet Section 3994, Copenhagen, Denmark, 2Rigshospitalet Section 5074, Copenhagen, Denmark

Purpose/Objective(s): Quantification of intra- and interfraction variations in uncoached respiration in fractionated radiotherapyof lung cancer, thus enabling estimates of possible margin reduction when using respiratory gating.

Materials/Methods: Eleven lung cancer patients referred for curative radiotherapy had their free uncoached respirationmonitored over 30 treatment fractions. No selection on basis of respiration was done. The in-house custom-made monitoringsystem was based on tracking an external marker box, and a fixed reference marker on the couch, enabling temporalmeasurement of absolute changes in respiration. The respiratory baseline was defined as the 5% fractile of the marker box

DoseMean Phase Detection Errors of 3

Breathing Patterns Sinogram 1 Sinogram 2 Sinogram 3

2Gy/fraction % projections with 0 phase error 100 98.2 97.5% projections with 1 phase error 0 0.5 1.2% projections with 2 or more phases error 0 1.3 1.3

2cGy/fraction % projections with 0 phase error 92.6 81.0 71.5% projections with 1 phase error 6.8 15 21.8% projections with 2 or more phases error 0.6 4.0 6.7

S610 I. J. Radiation Oncology ● Biology ● Physics Volume 66, Number 3, Supplement, 2006

positions in a breathing cycle. The respiratory amplitude was defined as encompassing the 95% and the 5% fractiles of markerbox positions in a breathing cycle.

The total respiratory motion span was defined as encompassing the 95% and the 5% fractiles of the marker box positionsthroughout the 30 fractions.

Results: The median intrafraction variation in respiratory baseline for all patients was 1.6 mm (range 0.2–7.4 mm).The median respiratory amplitude, within one fraction, for all patients was 4.5 mm (range 1.6–11.2 mm). The median

intrafractional variation factor in amplitude (maximal amplitude in one fraction/minimal amplitude in one fraction) was 2.0(range 1.7–3.4).

The median interfraction variation in respiratory baseline for all patients was 14.8 mm (range 5.5–31.0 mm). The interfractionvariation in baseline was systematically correlated to fraction number with R2�0.6 for 4 of 11 patients. For these four patients,the baseline decreased over time.

For two patients, a decreasing baseline was related to a decreasing body diameter, measured 3 times over the treatment time.Overall, the intrafraction variations in baseline accounted for around 10% of intrafraction variations in baseline.

The median interfractional variation factor in respiratory amplitude for all patients (over all maximal amplitude for onepatient/ over all minimal amplitude for one patient) was 5.8 (range 3.2–25). Thus the interfractional changes in amplitude werearound three times larger than the intrafractional changes in amplitude.

For all patients, the median anterior-posterior respiratory motion span over 30 fractions was 21.8 mm (range 8.8–41.7 mm).The respiratory motion span includes interfraction variations in both baseline and amplitude. The interfraction variation inbaseline accounts on average for 70% of the respiratory motion span, thus being more pronounced than the variations inamplitude.

Although the median respiratory amplitude measured in a single fraction was only 4.5 mm, the median anterior-posteriorrespiratory motion span over 30 fractions was 21.8 mm.

Conclusions: Uncoached respiration in lung cancer patients is characterized by intrafractional variation in amplitude andrespiratory baseline that makes coaching mandatory if margin reduction on basis of respiratory gating is considered.Interfractional variations in respiration are up to ten times larger than interfraction variations. If interfraction variations inrespiratory baseline are not addressed, the interfraction variations in respiration are underestimated. Safe reduction of marginsfor gated radiotherapy requires handling of interfractional changes in respiratory baseline.

Author Disclosure: T. Juhler-Nøttrup, The Danish Cancer Society, B. Research Grant; Varian, C. Other Research Support; S.Korreman, Varian, C. Other Research Support; L. Aarup, Varian, C. Other Research Support; H. Nystrom, Varian, C. OtherResearch Support; M. Olsen, None; A. Pedersen, Varian, C. Other Research Support; L. Specht, Varian, C. Other ResearchSupport.

2722 An Analysis of the Accuracy of the 6D Tracking With CyberKnife

M. Inoue1, K. Sato1, I. Koike2

1Yokohama CyberKnife Center, Yokohama, Japan, 2Yokohama City University, Yokohama, Japan

Purpose/Objective(s): Radiosurgery utilizes stereotactic principles of localization and multiple cross-fired beams to deliver alarge radiation dose to a well-defined target with little or no fractionation. The CyberKnife (CK) provides framelessimage-guided stereotactic irradiation which can robotically track, detect and correct for patient movement. The CK therapybeams are aimed at the lesion by a six-axis robotic arm which allows the CK to deliver arbitrarily shaped treatments whiletracking patient motion. Real-time image guidance is provided by two orthogonal x-ray imaging systems. Inverse treatmentplanning algorithm takes advantage of the flexibility afforded by robotic targeting providing for a high degree of doseconformality within non-spherically shaped lesions. The use of stereotactic radiosurgical systems to treat lesions requires a highdegree of positional accuracy. In this study we evaluate the positional accuracy of irradiation with 6D tracking and correctionas compared to the many 3D tracking-correction studies.

Materials/Methods: The positional accuracy of irradiation with 6D tracking and correction depends on a) target trackingprecision and b) beam pointing accuracy.Target precision was measured with an anthropomorphic skull phantom mountedon a robotic arm capable of positioning the phantom anywhere within the camera’s field-of-view with the precision ofmicrons. The phantom was imaged in a CT study from which the DRRs, used in the image correlation process, werecomputed. The phantom position was stepped in various combinations of three directions ranging from -10 to �10mm along each axis, and rotated in various combinations on all three axes ranging from 0 to 1.0 degree. After movingthe robotically mounted phantom, the imaging system acquired a pair of images and computed the net movement valueof the phantom for comparison with that obtained with the robotic arm. The beam pointing accuracy was measured by a2 mm diameter He-Ne laser positioned along the axis of the radiation of the linac and a 2 mm diameter light-sensitivecrystal sphere, fiber optically coupled to a light-sensing diode and amplifier. When the laser light strikes the center ofthe crystal sphere it produces a large signal. We created a plan aiming the center of crystal sphere and moving the linacby targeting the center of crystal sphere with laser light. Then the phantom was rotated 1 degree about each axis.Beam pointing error was measured by rescanning the position of the linac until the maximum signal from the crystal wasfound.

Results: The beam pointing error had a root-mean-square radial error of 0.48 mm from 28 different points. The tracking errorhad a root-mean-square radial displacement error of 0.22 mm and 0.2 degree rotational error.

Conclusions: The positional accuracy of irradiation with 6D tracking and correction of CK system was satisfactory.

Author Disclosure: M. Inoue, None; K. Sato, None; I. Koike, None.

S611Proceedings of the 48th Annual ASTRO Meeting