- Email: [email protected]
EP-1732: Estimation of total MRI-CT image registration errors for the pelvis
S261
ESTRO 33, 2014 Conclusions: ARIA 13, TrueBeam 2.0 and the PerfectPitch couch together allow online 6 DOF setup corrections and 4D volumetric verification at a lower dose than on standard OBI. Multi-scan CBCT and end to end integration enables replanning without requiring an additional patient visit. 4D adaptive radiotherapy is now realistic in a busy centre.
computerised tomography (4DCT) in which abdominal motion is used as a surrogate for respiratory-induced motion can be used to account for intra- fraction motion [2]. The aim of this study was to determine the accuracy of 4DCT scanning techniques employed for generating an internal target volume (ITV), which takes account of intra-fraction motion.
EP-1732 Estimation of total MRI-CT image registration errors for the pelvis J. Sjöberg1, A. Carlberg1, J. Nilsson2 1 Karolinska University Hospital, MRI Physics Department, Stockholm, Sweden 2 Karolinska University Hospital, Radiotherapy Physics Department, Stockholm, Sweden
Materials and Methods: Images of the Quasar Programmable Respiratory Motion Phantom (Quasar, Modus Medical Devices) fitted with a 4DCT imaging insert (P/N:500-3317) were acquired on a Philips Brilliance Big Bore (85 cm aperture) CT scanner. The phantom insert was programmed to move with a range (R=15 and R=30 mm pk-pk) of sinusoidal and cosine patterns with different breaths per minute (BPM)/period (s) (20/3 and 10/6). Retrospective spiral gating was used to produce volumetric data sets representing a percentage of the breathing cycle over one period (CT0 to CT90). Maximum (MAX) and Average (AVE) Intensity Projections (IP) were computed from these data. The Eclipse™ treatment planning system (Varian Medical Systems, Palo Alto, CA) was used to create ITVs on: CT0 to CT90; MAX IP; and AVG IP and the positional and volumetric imaging accuracy established.
Purpose/Objective: Computation of proper treatment margins requires intricate knowledge of the geometric uncertainty contribution from the imaging modalities and the subsequent image registration used for target delineation. Literature on the subject indicates a non-zero uncertainty contribution. The objective of this study is to investigate the MRI-CT image registration errors encountered for pelvic examinations. Materials and Methods: Standard pelvic CT scans was acquired of 26 patients. The patients were thereafter within one hour scanned in treatment position at the newly installed 3T MRI scanner using a T1 weighted gradient echo pulse sequence and a T2 weighted turbo spin echo pulse sequence, both employing a 3D geometric distortion correction algorithm. The images were transferred to the PACS- and ARIA systems. The registration module of ARIA v. 11 with standard registration parameters employing rigid registration using the mutual information metric was used for image registration and the subsequent registration quality was visually evaluated. Images of patients with implanted fiducial markers were manually registered using the positions of the markers. Discrepancies were measured using the internal distance measuring tool in the lateral, anterior-posterior (AP) and the craniocaudal (CC) direction for the bladder when using the mutual information metric whereas the largest discrepancy irrespective of direction was quantified for the femoral heads, body contour and fat/muscle boundaries. Results: Registration errors of 0±1 mm, 7±6 mm and 13±8 mm were found for the femoral heads, fat/muscle boundaries and the body contour, respectively and the bladder was misregistered by 13±10 mm, 10±7 mm and 14±10 mm in the AP, lateral and CC direction respectively when using the mutual information metric. Registration errors of 6±4 mm, 10±5 mm and 15±8 mm for the femoral heads, fat/muscle boundaries and the body contour, respectively were found when performing manual registration based on the marker positions. The large discrepancies and standard deviations for the soft tissue organs can be attributed to organ motion between the image acquisitions and reflects normal inter- and intrafractional organ motion. The statistically significant (p<0.05) worsening of the registration quality of the femoral heads is also a consequence of normal prostate movement between the scans. The negligible discrepancy for the relatively rigid femoral heads when using the mutual information metric is consistent with previously reported values. Conclusions: The excellent registration quality for the femoral heads when using the mutual information metric allows for routine preplanning MRI-CT image registration of the pelvis when daily matching on skeletal structures are performed online at the linear accelerator, without the need to revise treatment margins. The image quality of the implanted fiducial markers was deemed not satisfactory for clinical use and in need of optimization before routine implementation. EP-1733 Quality assurance of four-dimensional computed tomography for stereotactic ablative radiotherapy W.H. Nailon1, J. Puxeu Vaque1, D.B. McLaren2, S.C. Erridge2 1 Western General Hospital Edinburgh Cancer Centre, Department of Oncology Physics, Edinburgh, United Kingdom 2 Western General Hospital Edinburgh Cancer Centre, Department of Clinical Oncology, Edinburgh, United Kingdom Purpose/Objective: In the treatment of non-small cell lung cancer (NSCLC), which results in over 1 million deaths worldwide every year, surgical resection remains the gold standard treatment for fit patients. However, a significant proportion of these patients are medically unfit or decline surgery. Stereotactic ablative radiotherapy (SABR) is widely regarded as the treatment of choice for these patients where a high biological dose of radiation is delivered in 3 to 8 fractions over a 2 to 3 week period [1]. However, intra-fraction motion must be taken into account when delivering such high doses per fraction. Four-dimensional
Results: The overall variability(μ±σ) in outlining the 27 cm3 cube within the 4DCT insert over all phases (CT0 to CT90) when driven by a sinusoidal and cosine input with BPM=(10 and 20) and R=(15 and 30 mm pk-pk) was 29.25±0.523 cm3. Using Eclipse™ the ITVs created were: Merge=(44.35±0.68 and 61.74±1.41 cm ); MAX IP=(45.25±2.75 and 62.76±2.49 cm ); AVG IP=(45.70 ± 1.06 and 58.15 ± 0.25 cm ) for actual volumes of 40.5 and 54 cm. Table 1 shows the percentage errors for the merge, AVE IP and MAX IP data sets where the sine(20/30) refers to 20 BPM at 30 mm pk-pk.
Table
1:
Percentage
error
for
Merge,
AVE
IP
and
MAX
IP.
Conclusions: The time required for contouring in SABR is significantly longer than normal and methods for creating ITVs must be carefully assessed before clinical use. These results demonstrate the sensitivity of the ITV to variations in BPM and amplitude of the breathing signal and that it is important to take account of them in SABR planning. References [1]. Lagerwaard FJ et al. Int J Radiat Oncol Biol Phys 2012; 83(1):348-353. [2]. Hurkmans CW et al. Int J Radiat Oncol Biol Phys 2011; 80(3):918-927. EP-1734 Validation of a deformable registration image of a commercial treated planning system R. García-Mollá1, N. Marco-Blancas1, J. Bonaque1, L. Viduera1 1 Consorcio Hosptialario Provincial de Castellon, Radiofisica y Protección Radiológica, Castellon, Spain Purpose/Objective: This study is based on evaluating the accuracy of registration deformable image (RDI) in distances from points of interest (POIs) in the computed tomography (CT) and in the cone beam computed tomography (CBCT). To determine the implications of these distances in doses, the lack of Inverse consistency (IC) of the RDI algorithm were used.
ESTRO 33, 2014 Conclusions: ARIA 13, TrueBeam 2.0 and the PerfectPitch couch together allow online 6 DOF setup corrections and 4D volumetric verification at a lower dose than on standard OBI. Multi-scan CBCT and end to end integration enables replanning without requiring an additional patient visit. 4D adaptive radiotherapy is now realistic in a busy centre.
computerised tomography (4DCT) in which abdominal motion is used as a surrogate for respiratory-induced motion can be used to account for intra- fraction motion [2]. The aim of this study was to determine the accuracy of 4DCT scanning techniques employed for generating an internal target volume (ITV), which takes account of intra-fraction motion.
EP-1732 Estimation of total MRI-CT image registration errors for the pelvis J. Sjöberg1, A. Carlberg1, J. Nilsson2 1 Karolinska University Hospital, MRI Physics Department, Stockholm, Sweden 2 Karolinska University Hospital, Radiotherapy Physics Department, Stockholm, Sweden
Materials and Methods: Images of the Quasar Programmable Respiratory Motion Phantom (Quasar, Modus Medical Devices) fitted with a 4DCT imaging insert (P/N:500-3317) were acquired on a Philips Brilliance Big Bore (85 cm aperture) CT scanner. The phantom insert was programmed to move with a range (R=15 and R=30 mm pk-pk) of sinusoidal and cosine patterns with different breaths per minute (BPM)/period (s) (20/3 and 10/6). Retrospective spiral gating was used to produce volumetric data sets representing a percentage of the breathing cycle over one period (CT0 to CT90). Maximum (MAX) and Average (AVE) Intensity Projections (IP) were computed from these data. The Eclipse™ treatment planning system (Varian Medical Systems, Palo Alto, CA) was used to create ITVs on: CT0 to CT90; MAX IP; and AVG IP and the positional and volumetric imaging accuracy established.
Purpose/Objective: Computation of proper treatment margins requires intricate knowledge of the geometric uncertainty contribution from the imaging modalities and the subsequent image registration used for target delineation. Literature on the subject indicates a non-zero uncertainty contribution. The objective of this study is to investigate the MRI-CT image registration errors encountered for pelvic examinations. Materials and Methods: Standard pelvic CT scans was acquired of 26 patients. The patients were thereafter within one hour scanned in treatment position at the newly installed 3T MRI scanner using a T1 weighted gradient echo pulse sequence and a T2 weighted turbo spin echo pulse sequence, both employing a 3D geometric distortion correction algorithm. The images were transferred to the PACS- and ARIA systems. The registration module of ARIA v. 11 with standard registration parameters employing rigid registration using the mutual information metric was used for image registration and the subsequent registration quality was visually evaluated. Images of patients with implanted fiducial markers were manually registered using the positions of the markers. Discrepancies were measured using the internal distance measuring tool in the lateral, anterior-posterior (AP) and the craniocaudal (CC) direction for the bladder when using the mutual information metric whereas the largest discrepancy irrespective of direction was quantified for the femoral heads, body contour and fat/muscle boundaries. Results: Registration errors of 0±1 mm, 7±6 mm and 13±8 mm were found for the femoral heads, fat/muscle boundaries and the body contour, respectively and the bladder was misregistered by 13±10 mm, 10±7 mm and 14±10 mm in the AP, lateral and CC direction respectively when using the mutual information metric. Registration errors of 6±4 mm, 10±5 mm and 15±8 mm for the femoral heads, fat/muscle boundaries and the body contour, respectively were found when performing manual registration based on the marker positions. The large discrepancies and standard deviations for the soft tissue organs can be attributed to organ motion between the image acquisitions and reflects normal inter- and intrafractional organ motion. The statistically significant (p<0.05) worsening of the registration quality of the femoral heads is also a consequence of normal prostate movement between the scans. The negligible discrepancy for the relatively rigid femoral heads when using the mutual information metric is consistent with previously reported values. Conclusions: The excellent registration quality for the femoral heads when using the mutual information metric allows for routine preplanning MRI-CT image registration of the pelvis when daily matching on skeletal structures are performed online at the linear accelerator, without the need to revise treatment margins. The image quality of the implanted fiducial markers was deemed not satisfactory for clinical use and in need of optimization before routine implementation. EP-1733 Quality assurance of four-dimensional computed tomography for stereotactic ablative radiotherapy W.H. Nailon1, J. Puxeu Vaque1, D.B. McLaren2, S.C. Erridge2 1 Western General Hospital Edinburgh Cancer Centre, Department of Oncology Physics, Edinburgh, United Kingdom 2 Western General Hospital Edinburgh Cancer Centre, Department of Clinical Oncology, Edinburgh, United Kingdom Purpose/Objective: In the treatment of non-small cell lung cancer (NSCLC), which results in over 1 million deaths worldwide every year, surgical resection remains the gold standard treatment for fit patients. However, a significant proportion of these patients are medically unfit or decline surgery. Stereotactic ablative radiotherapy (SABR) is widely regarded as the treatment of choice for these patients where a high biological dose of radiation is delivered in 3 to 8 fractions over a 2 to 3 week period [1]. However, intra-fraction motion must be taken into account when delivering such high doses per fraction. Four-dimensional
Results: The overall variability(μ±σ) in outlining the 27 cm3 cube within the 4DCT insert over all phases (CT0 to CT90) when driven by a sinusoidal and cosine input with BPM=(10 and 20) and R=(15 and 30 mm pk-pk) was 29.25±0.523 cm3. Using Eclipse™ the ITVs created were: Merge=(44.35±0.68 and 61.74±1.41 cm ); MAX IP=(45.25±2.75 and 62.76±2.49 cm ); AVG IP=(45.70 ± 1.06 and 58.15 ± 0.25 cm ) for actual volumes of 40.5 and 54 cm. Table 1 shows the percentage errors for the merge, AVE IP and MAX IP data sets where the sine(20/30) refers to 20 BPM at 30 mm pk-pk.
Table
1:
Percentage
error
for
Merge,
AVE
IP
and
MAX
IP.
Conclusions: The time required for contouring in SABR is significantly longer than normal and methods for creating ITVs must be carefully assessed before clinical use. These results demonstrate the sensitivity of the ITV to variations in BPM and amplitude of the breathing signal and that it is important to take account of them in SABR planning. References [1]. Lagerwaard FJ et al. Int J Radiat Oncol Biol Phys 2012; 83(1):348-353. [2]. Hurkmans CW et al. Int J Radiat Oncol Biol Phys 2011; 80(3):918-927. EP-1734 Validation of a deformable registration image of a commercial treated planning system R. García-Mollá1, N. Marco-Blancas1, J. Bonaque1, L. Viduera1 1 Consorcio Hosptialario Provincial de Castellon, Radiofisica y Protección Radiológica, Castellon, Spain Purpose/Objective: This study is based on evaluating the accuracy of registration deformable image (RDI) in distances from points of interest (POIs) in the computed tomography (CT) and in the cone beam computed tomography (CBCT). To determine the implications of these distances in doses, the lack of Inverse consistency (IC) of the RDI algorithm were used.