Geometrical uncertainties, radiotherapy planning margins, and the ICRU-62 report

The purpose of this study was to determine interfractional variation of the centroid position and volume of internal target volume (ITV) during stereotactic body radiation therapy (SBRT) of lung cancer. From January 2014 to August 2014, a total of 32 patients with 37 primary or metastatic lung tumors were enrolled in our study. All patients received SBRT treatment in 4–5 fractions to a median dose of 48 Gy. Both 3D CT and 4D CT scans were used for radiotherapy treatment planning. 3D CBCT was acquired prior to treatment delivery to verify patient positioning. A total of 163 3D CBCT images were available for evaluation. 3D CBCT scans acquired for verification were registered with simulation CT scans. The ITVs were contoured on all verification 3D CBCT scans and compared to the initial gross target volume (GTV) or ITV in treatment planning system. GTV was based on 3D CT while ITV was based on both 3D CT and 4D CT. To assess the interfractional variation of ITV centroid position, we used vertebrae body adjacent to the tumor as reference point when performing the registration procedure. To eliminate the effect of time on tumor volume between simulation CT scan and the first fraction, the interfractional variation of ITV was evaluated from the first fraction to the last fraction. The overall 3D vector shift was 4.4±2.5 mm (range: 0.4−13.8 mm). The interfractional variation of ITV centroid position in superior‐inferior, anterior‐posterior, and left‐right directions were −0.7±2.7 mm, −1.4±3.4 mm, and −0.5±2.2 mm, respectively. No significant difference was observed between three directions (p=0.147). Large interfractional variations (≥5 mm) were observed in 12 fractions (9.3%) in superior‐inferior direction, 24 fractions (18.6%) in anterior‐posterior direction, and 5 fractions (3.9%) in left‐right direction. No time trend of tumor volume change measured in 3D CBCT was detected during four fractions (p=0.074). A significant (p=0.010) time trend was detected when evaluating the time trend of ITV change during 5 fractions and diameter was found to be significantly correlated with the ITV change (p=0.000). ITV did not show significant regression during SBRT treatment, but interfractional variation in the ITV centroid position was observed, especially in anterior‐posterior direction. An isotropic margin of 7 mm around ITV might be necessary for adequate coverage of interfractional variation of ITV centroid position, but only in case no soft tissue‐based setup is performed during SBRT treatment.PACS number(s): 87.55.dk

Section: Discussionmentioning

confidence: 99%

Evaluation of interfractional variation of the centroid position and volume of internal target volume during stereotactic body radiotherapy of lung cancer using cone‐beam computed tomography

Sun

Cheng

et al. 2016

“…Magnitude of rotational image uncertainty could also be factored into departmental PTV margin guidelines. ( 7 , 8 ) …”

Section: Discussionmentioning

confidence: 99%

Systematic offset of kV and MV localization systems as a function of gantry angle

Mullins

Herman

2010

Mechanical flex of the gantry and mounted imaging panels leads to systematic offsets in localization image isocenter as a function of gantry angle for linear accelerator‐mounted image guidance systems. Subsequently, object positions obtained from localization radiographs may be offset, resulting in greater target positioning uncertainty. While current QA procedures measure kV/MV image agreement, these measurements do not provide insight to apparent isocenter position for either single imaging system as a function of gantry rotation. This study measures offset as a function of gantry angle in kV and MV imaging systems on four treatment machines to investigate the magnitude of systematic offsets and their reproducibility between systems and machines, as well as over time. It is shown that each machine and energy has a reproducible pattern of offset as a function of gantry angle that is independent of kV/MV agreement, and it varies by machine. kV and MV offset ranges are on the order of 1.5 mm in the R/L and A/P directions, and 0.5 mm in the S/I direction. Variability of kV‐MV agreement is on the order of 0.7 mm. At certain angles, combinations of localization images could show a compounded offset of over 2 mm, exceeding the desired certainty threshold. Since these trends are persistent over time for each machine, online correction for image offsets as a function of gantry angle could improve the margin of positioning uncertainty.PACS number: 87.55.Qr

“…The statistical analysis approach employed in this work is similar to those reported in pervious works. ( 7 , 13 , 19 , 20 ) The random errors were calculated from the standard deviation (σ) of the normal distributions that provide best fit to the shifts histograms in three dimensions: A‐P, R‐L and S‐I. These shifts were calculated by registering the kV radiographic images to DRRs acquired directly after patient setup and represent daily variation in patient setup.…”

Section: Methodsmentioning

confidence: 99%

Evaluation of the setup accuracy of a stereotactic radiotherapy head immobilization mask system using kV on‐board imaging

Ali

Tubbs

Hibbitts

et al. 2010

The purpose of this study was to evaluate setup accuracy and quantify random and systematic errors of the BrainLAB stereotactic immobilization mask and localization system using kV on‐board imaging. Nine patients were simulated and set up with the BrainLAB stereotactic head immobilization mask and localizer to be treated for brain lesions using single and hypofractions. Orthogonal pairs of projections were acquired using a kV on‐board imager mounted on a Varian Trilogy machine. The kV projections were then registered with digitally‐reconstructed radiographs (DRR) obtained from treatment planning. Shifts between the kV images and reference DRRs were calculated in the different directions: anterior‐posterior (A‐P), medial‐lateral (R‐L) and superior‐inferior (S‐I). If the shifts were larger than 2 mm in any direction, the patient was reset within the immobilization mask until satisfying setup accuracy based on image guidance has been achieved. Shifts as large as 4.5 mm, 5.0 mm, 8.0 mm in the A‐P, R‐L and S‐I directions, respectively, were measured from image registration of kV projections and DRRs. These shifts represent offsets between the treatment and simulation setup using immobilization mask. The mean offsets of 0.1 mm, 0.7 mm, and −1.6 mm represent systematic errors of the BrainLAB localizer in the A‐P, R‐L and S‐I directions, respectively. The mean of the radial shifts is about 1.7 mm. The standard deviations of the shifts were 2.2 mm, 2.0 mm, and 2.6 mm in A‐P, R‐L and S‐I directions, respectively, which represent random patient setup errors with the BrainLAB mask. The BrainLAB mask provides a noninvasive, practical and flexible immobilization system that keeps the patients in place during treatment. Relying on this system for patient setup might be associated with significant setup errors. Image guidance with the kV on‐board imager provides an independent verification technique to ensure accuracy of patient setup. Since the patient may relax or move during treatment, uncontrolled and undetected setup errors may be produced with patients that are not well‐immobilized. Therefore, the combination of stereotactic immobilization and image guidance achieves more controlled and accurate patient setup within 2 mm in A‐P, R‐L and S‐I directions.PACS numbers: 87.56.‐v, 87.56.Da