Prostate Gland Motion in Prone and Supine Positions Assessed in Real Time by Implanted...
2
2.0 mm) in SI, and 0.5 mm (-0.1; -0.1 to 1.0 mm) in the RL direction. This was greater for the CT scans repeated during treatment: the 1SD (mean; range) were 1.9 mm (0.1: -3.9 mm to 3.9 mm) in AP, 1.9 mm (-0.7: -4.2 to 5.0 mm) in SI, and 0.7 mm (-0.1: -2.9 to 1.0 mm) in RL directions. The mean (range) VOI were 0.93 (0.72 - 0.98) for the prostate and 0.87 (0.51 - 0.98) for the rectum for the same day CTs. The repeat CTs during the course of treatment showed a larger shape variation with VOIs of 0.86 for prostate and 0.74 for rectum. The prostate-rectum interface agreed well even at low VOI values. Repeat dosimetry of the worst case scenarios still showed 100% of the prostate receiving the prescription dose. Conclusions: A setup procedure using a rectal balloon combined with daily KV X-ray imaging is effective. The setup error is less than 3 mm with only minor internal prostate movements and deformations. Author Disclosure: A.K. Lee, None; R. Kudchadker, None; B. Choi, None; S. Choi, None; J. Bluett, None; R. Zhu, None; L. Dong, None. 2860 Automated Registration of Large Deformations for Adaptive Radiation Therapy of Prostate Cancer A. R. Godley, E. Ahunbay, C. Peng, J. Christensen, X. Li Medical College of Wisconsin, Milwaukee, WI Purpose/Objective(s): This work develops a novel approach to accurately and efficiently register large deformations and tests the approach for adaptive radiation therapy planning of prostate cancer. Materials/Methods: A software tool using masks to guide a fast symmetric Demons algorithm was developed using ITK to reg- ister planning to treatment fractions CT images. The tool was tested for deformations of prostate, bladder and rectum and used to calculate accumulated dose for multiple treatment fractions for 5 prostate cancer patients. An initial registration is achieved by aligning the prostate center of mass in each image, akin to the patient alignment by radiation therapists in IGRT. A rigid alignment of the skeleton can also be performed. The region of the CT images deformably registered is defined by the volume containing a given minimum percent (4% used here) of the maximum dose, thereby increasing the speed of registration without reducing the accuracy of the dose calculation. A multi-resolution deformation further improves speed. For large organ discrepancies ( \70% Dice’s coefficient (DC), defined as overlap divided by average volume), masks are used to guide the deformation and con- centrate the registration on the organs of interest. The bladder and rectum volume are masked with a uniform intensity of -1000 and 1000 Hounsfield number, respectively in both planning and treatment images. Calculation of the overlap of these organs is facil- itated by the masks. The masks can be manually or automatically generated. The registration tool requires no user interaction. Results: The deformable registration tool accurately registered multiple fractions for the 5 patients with initial organ overlaps as low as 36% DC. The overlap of the plan and deformed organ contours became 96-99% DC. Registration of 512 x 512 x 100 pixel images took 5-8 minutes depending on the dose field volume. The prostate and seminal vesicles were correctly placed even though they are not masked. The dose delivered to the prostate was compared between rigid and prostate center of mass alignment, with the latter being much closer to the plan, as expected. The cumulative (actual) doses for multiple (e.g., 25) fractions with large defor- mation were computed and verified. Fractional doses to the prostate, rectum and bladder had standard deviations of 3.2, 9.9, and 5.2 cGy, respectively. The actual doses showed a prostate underdose of 0.5%, and 5-10% changes for the rectum and bladder. Conclusions: Five prostate patients were successfully registered, despite large organ discrepancies, more will be presented. The approach developed can be effectively used for adaptive planning by providing details of interfractional changes during prostate radiation therapy. Author Disclosure: A.R. Godley, None; E. Ahunbay, None; C. Peng, None; J. Christensen, None; X. Li, None. 2861 Prostate Gland Motion in Prone and Supine Positions Assessed in Real Time by Implanted Electromagnetic Transponders W. M. Butler, B. S. Kurko, B. C. Murray, G. S. Merrick Schiffler Cancer Center, Wheeling, WV Purpose/Objective(s): To compare prostate gland motion in prone and supine positions to determine the frequency, duration, and magnitude of out-of-tolerance events. Materials/Methods: Twenty-four patients implanted with Calypso Beacon electromagnetic transponders were planned in the prone position with chest pillow support and hip immobilization using custom-formed thermoplastic. Transponder coordinates from computed tomography (CT) were entered into the monitoring system. Both prior to and post-treatment, the coordinates of the centroid of the transponders were monitored for five minutes at 10 times per second with the patient supine. Patient setup in the prone position was confirmed by cone-beam CT, and transponder locations were monitored throughout treatment. Because the transponder centroid was set within 0.2 mm of the reference, all motion was considered relative to zero. Tolerance was set at 4 mm in the anterior/posterior direction and 5 mm in the lateral and superior/inferior directions. To ensure that the single session comparing prone versus supine positioning was representative for each patient, mean deviations of transponder coordinates were determined for previous treatments - a total of over 800 fractions. Results: Using repeated-measures analysis, there was no significant intra-patient difference in the variation of transponder coor- dinates between the patient positioning test and previous treatment sessions. The standard deviation from the setup zero for prone versus supine positioning was 1.6 mm vs. 2.2 mm AP, 1.5 mm vs. 1.8 mm for lateral, and 1.9 mm vs. 1.8 mm sup/inf. No patient was out of tolerance during the positioning comparison, and only two patients had more than one instance of out-of-tolerance dis- placement during all 800 monitored fractions. As a percentage of monitoring time, displacements .4 mm for prone versus supine were 0% vs. 1.5% for AP, 1.6% vs. 2.1% for lateral, and 3.4% vs. 2.0% for sup/inf. The percentage of time occupied by displace- ments between 3 mm and 4 mm for prone versus supine was 7.8% vs. 9.4% for AP, 7.2% vs. 8.6% for lateral, and 11.5% vs. 9.2% for sup/inf. In the sup/inf direction in the prone position, patient breathing created a sinusoidally oscillating motion that was more pronounced than for patients in the supine position. S554 I. J. Radiation Oncology d Biology d Physics Volume 72, Number 1, Supplement, 2008
Transcript of Prostate Gland Motion in Prone and Supine Positions Assessed in Real Time by Implanted...
S554 I. J. Radiation Oncology d Biology d Physics Volume 72, Number 1, Supplement, 2008
2.0 mm) in SI, and 0.5 mm (-0.1; -0.1 to 1.0 mm) in the RL direction. This was greater for the CT scans repeated during treatment:the 1SD (mean; range) were 1.9 mm (0.1: -3.9 mm to 3.9 mm) in AP, 1.9 mm (-0.7: -4.2 to 5.0 mm) in SI, and 0.7 mm (-0.1: -2.9 to1.0 mm) in RL directions. The mean (range) VOI were 0.93 (0.72 - 0.98) for the prostate and 0.87 (0.51 - 0.98) for the rectum for thesame day CTs. The repeat CTs during the course of treatment showed a larger shape variation with VOIs of 0.86 for prostate and0.74 for rectum. The prostate-rectum interface agreed well even at low VOI values. Repeat dosimetry of the worst case scenariosstill showed 100% of the prostate receiving the prescription dose.
Conclusions: A setup procedure using a rectal balloon combined with daily KV X-ray imaging is effective. The setup error is lessthan 3 mm with only minor internal prostate movements and deformations.
Author Disclosure: A.K. Lee, None; R. Kudchadker, None; B. Choi, None; S. Choi, None; J. Bluett, None; R. Zhu, None; L. Dong,None.
2860 Automated Registration of Large Deformations for Adaptive Radiation Therapy of Prostate Cancer
A. R. Godley, E. Ahunbay, C. Peng, J. Christensen, X. Li
Medical College of Wisconsin, Milwaukee, WI
Purpose/Objective(s): This work develops a novel approach to accurately and efficiently register large deformations and tests theapproach for adaptive radiation therapy planning of prostate cancer.
Materials/Methods: A software tool using masks to guide a fast symmetric Demons algorithm was developed using ITK to reg-ister planning to treatment fractions CT images. The tool was tested for deformations of prostate, bladder and rectum and used tocalculate accumulated dose for multiple treatment fractions for 5 prostate cancer patients. An initial registration is achieved byaligning the prostate center of mass in each image, akin to the patient alignment by radiation therapists in IGRT. A rigid alignmentof the skeleton can also be performed. The region of the CT images deformably registered is defined by the volume containinga given minimum percent (4% used here) of the maximum dose, thereby increasing the speed of registration without reducingthe accuracy of the dose calculation. A multi-resolution deformation further improves speed. For large organ discrepancies(\70% Dice’s coefficient (DC), defined as overlap divided by average volume), masks are used to guide the deformation and con-centrate the registration on the organs of interest. The bladder and rectum volume are masked with a uniform intensity of -1000 and1000 Hounsfield number, respectively in both planning and treatment images. Calculation of the overlap of these organs is facil-itated by the masks. The masks can be manually or automatically generated. The registration tool requires no user interaction.
Results: The deformable registration tool accurately registered multiple fractions for the 5 patients with initial organ overlaps aslow as 36% DC. The overlap of the plan and deformed organ contours became 96-99% DC. Registration of 512 x 512 x 100 pixelimages took 5-8 minutes depending on the dose field volume. The prostate and seminal vesicles were correctly placed even thoughthey are not masked. The dose delivered to the prostate was compared between rigid and prostate center of mass alignment, with thelatter being much closer to the plan, as expected. The cumulative (actual) doses for multiple (e.g., 25) fractions with large defor-mation were computed and verified. Fractional doses to the prostate, rectum and bladder had standard deviations of 3.2, 9.9, and 5.2cGy, respectively. The actual doses showed a prostate underdose of 0.5%, and 5-10% changes for the rectum and bladder.
Conclusions: Five prostate patients were successfully registered, despite large organ discrepancies, more will be presented. Theapproach developed can be effectively used for adaptive planning by providing details of interfractional changes during prostateradiation therapy.
Author Disclosure: A.R. Godley, None; E. Ahunbay, None; C. Peng, None; J. Christensen, None; X. Li, None.
2861 Prostate Gland Motion in Prone and Supine Positions Assessed in Real Time by Implanted Electromagnetic
TranspondersW. M. Butler, B. S. Kurko, B. C. Murray, G. S. Merrick
Schiffler Cancer Center, Wheeling, WV
Purpose/Objective(s): To compare prostate gland motion in prone and supine positions to determine the frequency, duration, andmagnitude of out-of-tolerance events.
Materials/Methods: Twenty-four patients implanted with Calypso Beacon electromagnetic transponders were planned in theprone position with chest pillow support and hip immobilization using custom-formed thermoplastic. Transponder coordinatesfrom computed tomography (CT) were entered into the monitoring system. Both prior to and post-treatment, the coordinates ofthe centroid of the transponders were monitored for five minutes at 10 times per second with the patient supine. Patient setupin the prone position was confirmed by cone-beam CT, and transponder locations were monitored throughout treatment. Becausethe transponder centroid was set within 0.2 mm of the reference, all motion was considered relative to zero. Tolerance was set at 4mm in the anterior/posterior direction and 5 mm in the lateral and superior/inferior directions. To ensure that the single sessioncomparing prone versus supine positioning was representative for each patient, mean deviations of transponder coordinateswere determined for previous treatments - a total of over 800 fractions.
Results: Using repeated-measures analysis, there was no significant intra-patient difference in the variation of transponder coor-dinates between the patient positioning test and previous treatment sessions. The standard deviation from the setup zero for proneversus supine positioning was 1.6 mm vs. 2.2 mm AP, 1.5 mm vs. 1.8 mm for lateral, and 1.9 mm vs. 1.8 mm sup/inf. No patientwas out of tolerance during the positioning comparison, and only two patients had more than one instance of out-of-tolerance dis-placement during all 800 monitored fractions. As a percentage of monitoring time, displacements .4 mm for prone versus supinewere 0% vs. 1.5% for AP, 1.6% vs. 2.1% for lateral, and 3.4% vs. 2.0% for sup/inf. The percentage of time occupied by displace-ments between 3 mm and 4 mm for prone versus supine was 7.8% vs. 9.4% for AP, 7.2% vs. 8.6% for lateral, and 11.5% vs. 9.2%for sup/inf. In the sup/inf direction in the prone position, patient breathing created a sinusoidally oscillating motion that was morepronounced than for patients in the supine position.
Proceedings of the 50th Annual ASTRO Meeting S555
Conclusions: The magnitude of deviations from the setup position was less in the prone position, but the mean supine deviationsagreed with prone to within 1 mm. The frequency of deviations greater than 3 or 4 mm was less for prone setups in the AP andlateral directions. Only in the sup/inf direction was the percent of time occupied by excursions greater than 3 mm greater for pronepositioning, and this arose from ventilatory oscillations.
Author Disclosure: W.M. Butler, None; B.S. Kurko, None; B.C. Murray, None; G.S. Merrick, None.
2862 Evaluation of a 3D Ultrasound System for Image Guided Radiation Therapy for Prostate Cancer
S. Wan, L. Stillwaugh, H. Prichard, J. Bowen, D. Provost
Sudbury Regional Hospital, Sudbury, ON, Canada
Purpose/Objective(s): Conventional portal imaging using megavoltage X-rays can only be used to align treatment fields with re-spect to bony anatomy. However, prostate moves daily relative to bony structure. Image-guided radiation therapy (IGRT) usingultrasound (US) is capable of visualizing soft tissue. We recently started using a 3D US system Restitu (Resonant Medical, Mon-treal, Canada) for prostate IGRT. The objective of this study is to assess the accuracy and precision of the Restitu system and todetermine an appropriate margin associated with this technology, which may result in reduced normal tissue toxicities and/or es-calated prescription dose.
Materials/Methods: Using CT imaging as the gold standard, this study investigates the correlation between prostate shifts deter-mined from CT and those from US images. After the planning CT and initial US scans, each patient undergoes 7 repeat weekly CTand US scans during the course of treatment. All scans are performed at the CT-Sim suite. Slice thickness in CT is 3 mm. Theprostate shifts between the planning and weekly CT scans are determined, so are the shifts between the initial and weekly US scans,on anterior-posterior (AP), right-left (RL) and superior-inferior (SI) directions. A strong correlation between the CT and US shiftswould show that US is as reliable as CT for prostate localization. Treatment margin may then be reduced.
Results: One hundred and forty pairs of CT-US shifts from 21 patients have been analyzed. Shifts of prostate were (mean ± SD, inmm): (1) from CT:�1.6 ± 5.3 (AP), 0.1± 2.3 (RL) and�0.4 ± 4.5 (SI); (2) from US:�1.3 ± 6.2 (AP), 0.3± 2.6 (RL) and�0.5 ± 4.9(SI); (3) difference between US and CT: 0.3 ± 2.7 (AP), 0.3± 2.2 (RL) and -0.1 ± 2.9 (SI). Correlation coefficient between CT andUS shifts were 0.94 (AP), 0.58 (RL) and 0.79 (SI). Using Stroom et al., 2002, based on CT data, PTV margin - without imageguidance- was calculated to be 11.5 mm (AP), 4.1 mm (RL) and 8.0 mm (SI); based on US data, it was 13.6 mm (AP), 4.7 mm(RL) and 8.6 mm (SI). With US-IGRT, if one only considers the residual error of US imaging with respect to CT, PTV marginwould be 5.2 mm (AP), 4.1 mm (RL) and 5.7 mm (SI).
Conclusions: Restitu 3D US system is a reliable alternative to CT for prostate localization, with the best correlation observed onAP direction - the most relevant to spare surrounding organs such as the rectum. To derive a conclusive PTV margin, we are cur-rently investigating other uncertainties, such as inter-user variability and systematic error between the localization laser systems inthe treatment rooms and those in the CT-Sim suite.
Author Disclosure: S. Wan, None; L. Stillwaugh, None; H. Prichard, None; J. Bowen, None; D. Provost, None.
2863 What is the Feasibility of Atlas-based IMRT Planning - What is the Accuracy if we used the Best Matched
IMRT Plan that we Apply Directly?A. Harrison1, J. Piper2, G. Kubicek1, R. Valicenti1, D. Adam1, J. Galvin1, Y. Xiao1
1Department of Radiation Oncology, Thomas Jefferson University Hospital, Philadelphia, PA, 2Department of ComputerScience, Wake Forest University, Winston-Salem, NC
Purpose/Objective(s): The advent of kV cone beam based IGRT has facilitated isocenter and positioning accuracy to sub milli-meter levels and has opened the door to the possibility of adaptive targeting of planned fields. The adaptive radiation therapy pro-cess, however, should include real-time implementation of modified plans. This is not being practiced because of the systematictime constraints of structure definition and complex/IMRT case re-planning. This study tests the feasibility of applying template orAtlas IMRT prostate plans to multiple patients for possible IMRT planning within minutes.
Materials/Methods: An atlas of treatment contours and anatomical CT images for 99 prostate patients was generated using com-mercially available software (MIMvista Corp., Cleveland, OH). The test cases are matched to the Atlas using three different tech-niques: small field of view (FOV), large FOV, and large FOV with constraints. The match process takes less than one minute. Whenthe test case CT scans are matched to an ‘‘Atlas’’ CT, the corresponding Atlas IMRT plan is applied to the physician contouredvolumes on the test case. Recalculation of each plan takes 3-4 minutes. All plans are scaled for PTV coverage of 95% to the pre-scribed dose of 7,560 cGy. Plans are also rescaled for the same coverage for CTV.
Results: Nine test patients are matched with the three different techniques defined above. For a particular patient, the best matchedof the three techniques are used for comparison analysis. For most of the cases, the small FOV match is the one chosen. For templateplans scaled to cover PTV adequately, dose to 20% of the rectum volume averages 75 Gy (standard deviation (STD) 8 Gy) andpercent rectum volume receiving 65 Gy averages 31% (STD 11%). A total of 30% bladder volumes receive doses that average57 Gy (STD 18 Gy). Percent bladder volume receiving 65 Gy or more averages 26% (STD 14%). With template plans scaledto cover CTV, the above numbers for rectum are 68 Gy (STD 7 Gy) and 26% (STD 11%), for bladders are 52 Gy (STD18 Gy) and 22% (STD 15%).
Conclusions: The template plans scaled to deliver adequate dose to CTV have better sparing of critical structures as expected. WithIGRT readily available, it is conceivable to reduce margins of PTVs to within millimeters of the CTV. With reduced PTV marginsto CTV, there could be a higher percentage of template IMRT prostate plans based on Atlas contour matching deemed clinicallyacceptable. With more cases added to the Alas, it is reasonable to expect to find better matches as well. With only minutes to arriveat a usable IMRT plan, it could be considered as a possible solution to the conundrum of wanting adaptive real-time re-planning andthe reality of time and workflow issues.
Author Disclosure: A. Harrison, None; J. Piper, Yes, E. Ownership Interest; G. Kubicek, None; R. Valicenti, None; D. Adam,None; J. Galvin, None; Y. Xiao, None.
