- Email: [email protected]
IMAGE REGISTRATION, MOTION MODELING AND TISSUE DEFORMATION
S 46
S YMPOSIUM
This figure compares the central-slice dose distributions of a 9 field IMRT head-and-neck treatment plan. The calculated dose distribution is labelled Eclipse, and two independent measurements are labelled EBT (2D film), and Presage (3D dosimetry utilizing optical-CT scanning). All measurements were made in a dosimetry insert compatible with the IMRT credentialing phantom through a collaboration with the Radiologic Physics Center (RPC). 122 speaker CHEMICAL VAPOUR DEPOSITED DIAMONDS M. Rebisz-Pomorska1 , D. Tromson1 , P. Bergonzo1 , A. Isambert2 1 C OMMISSARIAT À L’E NERGIE ATOMIQUE , LIST, Diamond Sensors Laboratory, Gif-sur-Yvette, France 2 I NSTITUT G USTAVE R OUSSY, Physics Department, Villejuif, France
One of the main aims for the improvement of the metrology during cancer treatment is the development of new tools for the precise measurement of the delivered dose. For stereotactic radiotherapy, identified problems such as the lack of lateral electronic equilibrium and the step fall-off in dose in the penumbra of small fields requires the use of small detectors (smaller than beam width), with fast response to be able to follow the fluctuations of the beam. For this reason we have developed single-crystal diamond detector (SCDD) with high spatial resolution of 1 mm. The optimisation of the Chemical Vapour Deposition (CVD) growth conditions to obtain fast detector response and the tests in clinical environment are the two main points to be filled to validate the use of these devices for the dosimetry of novel, IMRT and mini-beam, radiotherapy techniques. In the frame of The MAESTRO project (Methods and Advanced Equipment for Simulation and Treatment in Radio-Oncology, 6th FP) the clinical tests towards the use of such Single Crystal Diamond Detector (SCDD) for dose measurements have been performed. Our device was characterized under various radiation conditions. Results were successfully evaluated with respect to the most drastic requirements of the Code of practice as defined by IAEA (International Atomic Energy Agency) and namely the TRS-398 and the MAESTRO project specifications. The evaluation of the dose measured with SCDD for various IMRT fields and the comparison to the PTW semiflex air ionization chamber as well as to the Konrad (Siemens OCS) treatment planning system (TPS) dose calculations is reported. The obtained very good agreement between measured with SCDD and calculated with TPS dose values demonstrate the quality of such single crystal diamond detectors for IMRT applications. Very promising results show the potentiality of use small synthetic diamond detector for monitoring of beam with width below 2 cm. Diamond detectors are also excellent candidates for particle- therapy applications as 2-D beam monitoring, single particle counting and beam energy spectroscopy.
T UESDAY, S EPTEMBER 1, 2009
provides sub-millimetre spatial resolution. The basic beam data were measured for a stereotactic radiotherapy collimator system (BrainLAB) using 6 MV photons and compared with the corresponding data acquired with a small diamond detector, a PinPoint ionisation chamber and polymer gels. Furthermore, comparisons were made between beam profiles of various radiosurgery accelerators, as well as penumbra studies. Results: The measurements of BrainLAB’s stereotactic collimator system showed an excellent agreement between DOSI and the diamond and gels dosimeters with percentage depth dose agreement to better than 1% for all depths. The agreement on off-axis ratios was better than 3% for all collimators and the agreement on relative output factors were at the 1% level. However, an increasing discrepancy with decreasing field size was observed between the relative output factors measured with DOSI and those measured the PinPoint ionisation chamber, especially for collimators with a diameter smaller than 15 mm. The beam profiles between the different stereotactic delivery systems agreed within 2% and 5% for the smallest fields. Conclusions: The results of this work support previous findings that: a) air ion chambers are not suitable for small field dosimetry since they result in penumbra broadening and require significant corrections due to severe charged particle transport alterations, b) gels and diamond detectors provide high resolution and accurate small field profile measurements, as long as the necessary dose rate corrections are correctly applied for the latter and c) the novel silicon-diode array (DOSI) used in this study has been shown to be suitable for small field profile measurements. This work emphasizes the need to perform profile measurements of small fields (for acceptance testing, commissioning and/or periodic quality assurance purposes) using more than one dosimetric method. We believe this to be an effective step towards reducing the overall uncertainty related to SRS/SRT treatments. Finally, DOSI’s fast read-out electronics made it possible for all measurements to be recorded within 45 minutes including time to change collimators. This should reduce the overall time for commissioning and QA measurements, an important factor especially for busy radiotherapy departments.
Autosegmentation and deformable registration algorithms 124 speaker IMAGE REGISTRATION, MOTION MODELING AND TISSUE DEFORMATION D. Hawkes1 , J. McClelland1 , M. Modat1 , D. Barratt1 , E. Rijkhorst1 , Y. Hu1 , J. Hipwell1 , S. Ourselin1 1 U NIVERSITY C OLLEGE L ONDON, Centre for Medical Image Computing, London, United Kingdom
The quality of radiotherapy delivery is largely determined by the spatial conformance of radiation dose and the target tumour. Imaging and registration are the means to determine this spatial conformance. When the rigid body assumption holds, image registration is a well validated and accurate technology and is widely used for lesions which have a fixed relationship to bony anatomy. Extension to tissues that deform or move is a significant challenge. A number of algorithms have been proposed to align datasets using non-rigid transformations. Unfortunately finding a non-rigid transformation that aligns two (or more) images is an ill-posed problem with many potential solutions and while alignment of the transformed datasets can appear very good, precise spatial correspondence is often poor. For registration to be useful it has to infer location, i.e. to interpolate or extrapolate spatial data, where no direct check is possible and so validation is difficult. We review non-rigid registration technologies, validation strategies and fast implementations suitable for clinical application. We present recent results in the breast, lung, liver and prostate and show how a combination of learnt models of motion derived by image registration, or constraints determined using biomechanical knowledge, can both improve imaging for treatment planning and can better direct therapy to moving or deforming targets.
123 speaker SILICON DIODE ARRAYS S. Manolopoulos1 1
U NIVERSITY H OSPITAL B IRMINGHAM, Birmingham, United Kingdom
Background and purpose: Small photon fields are increasingly used in modern radiotherapy and especially in IMRT and SRS/SRT treatments. The uncertainties related to small field profile measurements can introduce significant systematic errors to the overall treatment process. These measurements are challenging mainly due to the absence of lateral charged particle equilibrium conditions, detector size/composition effects as well as positioning problems. Material and Methods: A new kind of dosimeter (DOSI) was developed on technology commonly encountered in particle physics experiments and was used to measure the basic beam characteristics of a clinical radiosurgery system. The device is capable of dynamic measurements in real time and
125 speaker ATLAS-BASED SEGMENTATION FOR RADIOTHERAPY PLANNING G. Malandain1 , O. Commowick2 , L. Ramus3 1 INRIA, Sophia Antipolis, France 2 C HILDREN ’ S H OSPITAL, Department of Radiology, Boston, USA 3 DOSI SOFT S.A., Cachan, France
Radiotherapy planning requires accurate delineations of the tumour and of the critical structures, or organs at risks (OARs). Manual segmentation suffers from inter- and intra-expert variability and requires a considerable amount of time. Computer-assisted methods address these two issues. An atlasbased segmentation is proposed to delineate all organs at risk at once. A digital anatomical atlas is made of a (3D) medical image together with the delineations of the OARs. Atlas-based segmentation consists in warping the medical image onto the actual image to be segmented, applying the computed
S YMPOSIUM
This figure compares the central-slice dose distributions of a 9 field IMRT head-and-neck treatment plan. The calculated dose distribution is labelled Eclipse, and two independent measurements are labelled EBT (2D film), and Presage (3D dosimetry utilizing optical-CT scanning). All measurements were made in a dosimetry insert compatible with the IMRT credentialing phantom through a collaboration with the Radiologic Physics Center (RPC). 122 speaker CHEMICAL VAPOUR DEPOSITED DIAMONDS M. Rebisz-Pomorska1 , D. Tromson1 , P. Bergonzo1 , A. Isambert2 1 C OMMISSARIAT À L’E NERGIE ATOMIQUE , LIST, Diamond Sensors Laboratory, Gif-sur-Yvette, France 2 I NSTITUT G USTAVE R OUSSY, Physics Department, Villejuif, France
One of the main aims for the improvement of the metrology during cancer treatment is the development of new tools for the precise measurement of the delivered dose. For stereotactic radiotherapy, identified problems such as the lack of lateral electronic equilibrium and the step fall-off in dose in the penumbra of small fields requires the use of small detectors (smaller than beam width), with fast response to be able to follow the fluctuations of the beam. For this reason we have developed single-crystal diamond detector (SCDD) with high spatial resolution of 1 mm. The optimisation of the Chemical Vapour Deposition (CVD) growth conditions to obtain fast detector response and the tests in clinical environment are the two main points to be filled to validate the use of these devices for the dosimetry of novel, IMRT and mini-beam, radiotherapy techniques. In the frame of The MAESTRO project (Methods and Advanced Equipment for Simulation and Treatment in Radio-Oncology, 6th FP) the clinical tests towards the use of such Single Crystal Diamond Detector (SCDD) for dose measurements have been performed. Our device was characterized under various radiation conditions. Results were successfully evaluated with respect to the most drastic requirements of the Code of practice as defined by IAEA (International Atomic Energy Agency) and namely the TRS-398 and the MAESTRO project specifications. The evaluation of the dose measured with SCDD for various IMRT fields and the comparison to the PTW semiflex air ionization chamber as well as to the Konrad (Siemens OCS) treatment planning system (TPS) dose calculations is reported. The obtained very good agreement between measured with SCDD and calculated with TPS dose values demonstrate the quality of such single crystal diamond detectors for IMRT applications. Very promising results show the potentiality of use small synthetic diamond detector for monitoring of beam with width below 2 cm. Diamond detectors are also excellent candidates for particle- therapy applications as 2-D beam monitoring, single particle counting and beam energy spectroscopy.
T UESDAY, S EPTEMBER 1, 2009
provides sub-millimetre spatial resolution. The basic beam data were measured for a stereotactic radiotherapy collimator system (BrainLAB) using 6 MV photons and compared with the corresponding data acquired with a small diamond detector, a PinPoint ionisation chamber and polymer gels. Furthermore, comparisons were made between beam profiles of various radiosurgery accelerators, as well as penumbra studies. Results: The measurements of BrainLAB’s stereotactic collimator system showed an excellent agreement between DOSI and the diamond and gels dosimeters with percentage depth dose agreement to better than 1% for all depths. The agreement on off-axis ratios was better than 3% for all collimators and the agreement on relative output factors were at the 1% level. However, an increasing discrepancy with decreasing field size was observed between the relative output factors measured with DOSI and those measured the PinPoint ionisation chamber, especially for collimators with a diameter smaller than 15 mm. The beam profiles between the different stereotactic delivery systems agreed within 2% and 5% for the smallest fields. Conclusions: The results of this work support previous findings that: a) air ion chambers are not suitable for small field dosimetry since they result in penumbra broadening and require significant corrections due to severe charged particle transport alterations, b) gels and diamond detectors provide high resolution and accurate small field profile measurements, as long as the necessary dose rate corrections are correctly applied for the latter and c) the novel silicon-diode array (DOSI) used in this study has been shown to be suitable for small field profile measurements. This work emphasizes the need to perform profile measurements of small fields (for acceptance testing, commissioning and/or periodic quality assurance purposes) using more than one dosimetric method. We believe this to be an effective step towards reducing the overall uncertainty related to SRS/SRT treatments. Finally, DOSI’s fast read-out electronics made it possible for all measurements to be recorded within 45 minutes including time to change collimators. This should reduce the overall time for commissioning and QA measurements, an important factor especially for busy radiotherapy departments.
Autosegmentation and deformable registration algorithms 124 speaker IMAGE REGISTRATION, MOTION MODELING AND TISSUE DEFORMATION D. Hawkes1 , J. McClelland1 , M. Modat1 , D. Barratt1 , E. Rijkhorst1 , Y. Hu1 , J. Hipwell1 , S. Ourselin1 1 U NIVERSITY C OLLEGE L ONDON, Centre for Medical Image Computing, London, United Kingdom
The quality of radiotherapy delivery is largely determined by the spatial conformance of radiation dose and the target tumour. Imaging and registration are the means to determine this spatial conformance. When the rigid body assumption holds, image registration is a well validated and accurate technology and is widely used for lesions which have a fixed relationship to bony anatomy. Extension to tissues that deform or move is a significant challenge. A number of algorithms have been proposed to align datasets using non-rigid transformations. Unfortunately finding a non-rigid transformation that aligns two (or more) images is an ill-posed problem with many potential solutions and while alignment of the transformed datasets can appear very good, precise spatial correspondence is often poor. For registration to be useful it has to infer location, i.e. to interpolate or extrapolate spatial data, where no direct check is possible and so validation is difficult. We review non-rigid registration technologies, validation strategies and fast implementations suitable for clinical application. We present recent results in the breast, lung, liver and prostate and show how a combination of learnt models of motion derived by image registration, or constraints determined using biomechanical knowledge, can both improve imaging for treatment planning and can better direct therapy to moving or deforming targets.
123 speaker SILICON DIODE ARRAYS S. Manolopoulos1 1
U NIVERSITY H OSPITAL B IRMINGHAM, Birmingham, United Kingdom
Background and purpose: Small photon fields are increasingly used in modern radiotherapy and especially in IMRT and SRS/SRT treatments. The uncertainties related to small field profile measurements can introduce significant systematic errors to the overall treatment process. These measurements are challenging mainly due to the absence of lateral charged particle equilibrium conditions, detector size/composition effects as well as positioning problems. Material and Methods: A new kind of dosimeter (DOSI) was developed on technology commonly encountered in particle physics experiments and was used to measure the basic beam characteristics of a clinical radiosurgery system. The device is capable of dynamic measurements in real time and
125 speaker ATLAS-BASED SEGMENTATION FOR RADIOTHERAPY PLANNING G. Malandain1 , O. Commowick2 , L. Ramus3 1 INRIA, Sophia Antipolis, France 2 C HILDREN ’ S H OSPITAL, Department of Radiology, Boston, USA 3 DOSI SOFT S.A., Cachan, France
Radiotherapy planning requires accurate delineations of the tumour and of the critical structures, or organs at risks (OARs). Manual segmentation suffers from inter- and intra-expert variability and requires a considerable amount of time. Computer-assisted methods address these two issues. An atlasbased segmentation is proposed to delineate all organs at risk at once. A digital anatomical atlas is made of a (3D) medical image together with the delineations of the OARs. Atlas-based segmentation consists in warping the medical image onto the actual image to be segmented, applying the computed