Advanced

Dosimetric and geometric evaluation of MRI as the only imaging modality for the radiotherapy treatment process of localized prostate cancer

Sjöberg, Johan (2010)
Medical Physics Programme
Abstract
Introduction MRI provides better soft tissue contrast than CT and is an invaluable tool in the radiotherapy treatment process. The gold standard today is image fusion of MRI and CT data where targets and organs-at-risk are delineated on the MRI data and the dose calculation is performed on the CT data. It is appealing to exclude CT from the process in order to reduce cost and time and to eliminate the uncertainty inherited from the image registration. Since MRI data does not contain any electron density information, this property must be assigned to the data. The purpose of this study is to evaluate the use of MRI as the only modality for the entire radiotherapy treatment process of localized prostate cancer. A study of the dosimetric... (More)
Introduction MRI provides better soft tissue contrast than CT and is an invaluable tool in the radiotherapy treatment process. The gold standard today is image fusion of MRI and CT data where targets and organs-at-risk are delineated on the MRI data and the dose calculation is performed on the CT data. It is appealing to exclude CT from the process in order to reduce cost and time and to eliminate the uncertainty inherited from the image registration. Since MRI data does not contain any electron density information, this property must be assigned to the data. The purpose of this study is to evaluate the use of MRI as the only modality for the entire radiotherapy treatment process of localized prostate cancer. A study of the dosimetric consequences of MRI based dose planning and the contrast properties of novel intra-prostatic marker materials in multimodality imaging was conducted. Materials and methods Ten patients were selected for the dosimetry study and were both MR- and CT-scanned using routine clinical imaging parameters. 3DCRT, IMRT and VMAT plans were generated for the CT data, unit density assigned CT-data, unit density assigned MRI data and bulk density assigned MRI data. The resulting dose distributions were compared by calculating the root-mean-square values (rms) for the volumes enclosed by isodose levels 60%, 70%, 80%, 90%, 95% and 100% normalized to the prescription dose and by generating topographic maps of the percentage dose- and the absolute dose difference. DVH"s were compared in order to study general tendencies in the dose distributions. The geometric distortion was studied with the help of a geometric distortion evaluation phantom provided by the vendor and the effect of using the vendors built in gradient distortion correction algo-rithm was studied utilizing both phantom measurements and scans of three healthy volunteers. The contrast properties of the marker implant materials was studied by inserting them in two pieces of ham followed by CT- and MRI scanning them using clinical routine protocols. Megavoltage and kilovoltage on-board imager systems capable of performing CBCT were also used to test the visibility of the markers at the linear accelerator. The best suitable marker was implanted into four volunteer patients in order to study the visibility in vivo with ana-tomical structures present. ResultsDose discrepancies > 2 % are found mainly in clinically insignificant areas in all patients regardless if bony areas is assigned a relevant CT-number or not. The dosimetric uncertainty is however significantly reduced (p < 0.05) inside the target volume for the 3DCRT plans (rms at V95% reduced from 1.4 ± 0.4 to 0.6 ± 0.2) but not for the VMAT plans (p ≈ 0.07) for the bulk density assigned MRI data compared to unit density assigned MRI data (rms at V95% reduced from 1.3 ± 0.6 to 1.1 ± 0.6). In contrary to the hypothesis, the dosimetric uncertainty is signifi-cantly increased (p < 0.05) inside the target volume for IMRT plans for bulk density assigned MRI data com-pared to unit density assigned MRI data (rms at V95% increased from 0.8 ± 0.3 to 1.2 ± 0.4). No apparent reason for this was found, but it is believed that the chosen CT-number of 403 HU is an overestimate of the true mean for this tissue and that the fluence optimized plans are more sensitive to the choice of CT-number. Large indi-vidual differences are found when studying local percentage dose differences and are mostly prominent for the VMAT treatment technique. The gradient distortion correction algorithm reduces the geometric distortion in the MR-images but residual distortion is present. The gold/nitinol material composition for marker implants showed acceptable visibility in all imaging modalities except the on-board MV-imaging system. ConclusionMRI-only based radiotherapy treatment planning and setup of prostate cancer patients is feasible. Incorporating a quality assurance program in order to reduce the geometric distortion in the MR-images even further as well as assigning relevant CT-numbers to bone tissues reduces the dosimetric uncertainty, however more work is re-quired in order to study these effects more thoroughly. The gold/nitinol material composition is recommended for use in multimodality imaging when CBCT or kV imaging is employed. (Less)
Abstract (Swedish)
Idag finns det många olika sätt att avbilda människokroppens inre. Två av de vanligaste metoderna som används idag är magnetresonans (MR) och datortomografi (CT). MR använder sig av starka magnetfält och radiovågor för att skapa snittbilder av kroppen, medan CT använder sig av röntgenstrålar. Båda kamerorna har sina fördelar och nackdelar. MR ger bättre vävnadskontrast, det vill säga det är lättare att urskilja olika närliggande vävnadstyper men är dålig på att ge geometriskt korrekta bilder och kan inte avbilda skelettet särskilt bra. CT ger sämre vävnadskontrast, men har utmärkt förmåga att ge geometriskt korrekta bilder och skelettet är lätt att urskilja från mjukvävnad. Inom strålterapin där strålning används för att ta död på... (More)
Idag finns det många olika sätt att avbilda människokroppens inre. Två av de vanligaste metoderna som används idag är magnetresonans (MR) och datortomografi (CT). MR använder sig av starka magnetfält och radiovågor för att skapa snittbilder av kroppen, medan CT använder sig av röntgenstrålar. Båda kamerorna har sina fördelar och nackdelar. MR ger bättre vävnadskontrast, det vill säga det är lättare att urskilja olika närliggande vävnadstyper men är dålig på att ge geometriskt korrekta bilder och kan inte avbilda skelettet särskilt bra. CT ger sämre vävnadskontrast, men har utmärkt förmåga att ge geometriskt korrekta bilder och skelettet är lätt att urskilja från mjukvävnad. Inom strålterapin där strålning används för att ta död på cancerceller används både MR och CT i förberedande syfte. MR ger bilder där tumören lätt syns medan CT ger information om hur strålningen absorberas i kroppen.

Att göra både MR och CT är dyrare och tar längre tid än att utesluta CT helt. Om bara MR används, kan sjukhuset spara pengar, patienten kommer fortare till behandling och man undviker problemet med bildmatchningen. Vissa problem uppstår dock. För det första så innehåller MR-bilderna ingen information om hur strålningen absorberas i de olika organen i kroppen. Denna information måste tillsättas i efterhand. Eftersom människokroppen till stor del består av vatten så är det en ganska bra uppskattning att tilldela kroppen vattendensitet. För det andra så är MR bilderna inte geometriskt korrekta, det vill säga en rund boll är inte helt rund i en MR bild utan kanske oval. När datorn räknar ut stråldosen så görs det på en oval boll istället för en rund, och då blir det fel när bollen sedan ska strålbehandlas. För det tredje så måste istället MR-bilderna användas för att lägga upp patienten rätt vid strålbehandlingsmaskinen. Min studie har som syfte att utreda om MR baserad strålbehandling fungerar. Hur stort blir felet på grund av att kroppen tilldelas vattendensitet? Och hur stort blir felet på grund av geometrisk distorsion? Dessutom undersöker jag om det är möjligt att sikta in strålningen enbart med hjälp av MR bilder.

Felet som uppstår är ca 2-3 % för MR-dosplanering jämfört med CT-dosplanering. Tar man hänsyn till alla fel blir det totala felet för behandlingen ca 5 %. Detta är acceptabelt eftersom utgången av behandlingen påverkas negativt först vid 7-10 % fel. Man måste dock vara försiktig och först se till så att den geometriska distorsionen i MR-bilderna är så liten som möjligt genom att använda sig av inbyggd mjukvara i MR-kameran och se till att den hålls i topptrim. Jag har också undersökt ett material för små stift med storleken av riskorn som implanteras i patientens prostata som syns både på MR, CT och vanliga röntgenbilder. Dessa kan användas för att sikta in strålfälten. (Less)
Please use this url to cite or link to this publication:
author
Sjöberg, Johan
supervisor
organization
year
type
H2 - Master's Degree (Two Years)
subject
keywords
Strålterapi
language
English
id
2157106
date added to LUP
2011-09-13 13:30:43
date last changed
2011-12-08 13:31:34
@misc{2157106,
  abstract     = {Introduction MRI provides better soft tissue contrast than CT and is an invaluable tool in the radiotherapy treatment process. The gold standard today is image fusion of MRI and CT data where targets and organs-at-risk are delineated on the MRI data and the dose calculation is performed on the CT data. It is appealing to exclude CT from the process in order to reduce cost and time and to eliminate the uncertainty inherited from the image registration. Since MRI data does not contain any electron density information, this property must be assigned to the data. The purpose of this study is to evaluate the use of MRI as the only modality for the entire radiotherapy treatment process of localized prostate cancer. A study of the dosimetric consequences of MRI based dose planning and the contrast properties of novel intra-prostatic marker materials in multimodality imaging was conducted. Materials and methods Ten patients were selected for the dosimetry study and were both MR- and CT-scanned using routine clinical imaging parameters. 3DCRT, IMRT and VMAT plans were generated for the CT data, unit density assigned CT-data, unit density assigned MRI data and bulk density assigned MRI data. The resulting dose distributions were compared by calculating the root-mean-square values (rms) for the volumes enclosed by isodose levels 60%, 70%, 80%, 90%, 95% and 100% normalized to the prescription dose and by generating topographic maps of the percentage dose- and the absolute dose difference. DVH"s were compared in order to study general tendencies in the dose distributions. The geometric distortion was studied with the help of a geometric distortion evaluation phantom provided by the vendor and the effect of using the vendors built in gradient distortion correction algo-rithm was studied utilizing both phantom measurements and scans of three healthy volunteers. The contrast properties of the marker implant materials was studied by inserting them in two pieces of ham followed by CT- and MRI scanning them using clinical routine protocols. Megavoltage and kilovoltage on-board imager systems capable of performing CBCT were also used to test the visibility of the markers at the linear accelerator. The best suitable marker was implanted into four volunteer patients in order to study the visibility in vivo with ana-tomical structures present. ResultsDose discrepancies > 2 % are found mainly in clinically insignificant areas in all patients regardless if bony areas is assigned a relevant CT-number or not. The dosimetric uncertainty is however significantly reduced (p < 0.05) inside the target volume for the 3DCRT plans (rms at V95% reduced from 1.4 ± 0.4 to 0.6 ± 0.2) but not for the VMAT plans (p ≈ 0.07) for the bulk density assigned MRI data compared to unit density assigned MRI data (rms at V95% reduced from 1.3 ± 0.6 to 1.1 ± 0.6). In contrary to the hypothesis, the dosimetric uncertainty is signifi-cantly increased (p < 0.05) inside the target volume for IMRT plans for bulk density assigned MRI data com-pared to unit density assigned MRI data (rms at V95% increased from 0.8 ± 0.3 to 1.2 ± 0.4). No apparent reason for this was found, but it is believed that the chosen CT-number of 403 HU is an overestimate of the true mean for this tissue and that the fluence optimized plans are more sensitive to the choice of CT-number. Large indi-vidual differences are found when studying local percentage dose differences and are mostly prominent for the VMAT treatment technique. The gradient distortion correction algorithm reduces the geometric distortion in the MR-images but residual distortion is present. The gold/nitinol material composition for marker implants showed acceptable visibility in all imaging modalities except the on-board MV-imaging system. ConclusionMRI-only based radiotherapy treatment planning and setup of prostate cancer patients is feasible. Incorporating a quality assurance program in order to reduce the geometric distortion in the MR-images even further as well as assigning relevant CT-numbers to bone tissues reduces the dosimetric uncertainty, however more work is re-quired in order to study these effects more thoroughly. The gold/nitinol material composition is recommended for use in multimodality imaging when CBCT or kV imaging is employed.},
  author       = {Sjöberg, Johan},
  keyword      = {Strålterapi},
  language     = {eng},
  note         = {Student Paper},
  title        = {Dosimetric and geometric evaluation of MRI as the only imaging modality for the radiotherapy treatment process of localized prostate cancer},
  year         = {2010},
}