Final Report Summary - MAESTRO (Methods and advanced equipment for simulation and treatment in radio-oncology)
The major challenge in the field of radiotherapy (RT) is to improve the conformation of dose delivered to the target (tumour and nearby tissues) whatever its shape, in order to treat the tumour more efficiently while sparing the surrounding healthy tissues. In this framework, the aim of MAESTRO was to optimise the application of conventional radiation therapy and allow emerging technologies like intensity-modulated radiation therapy (IMRT), image-guided radiotherapy (IGRT), and proton therapy to grow in providing them with more precise tools, quicker calculation means and appropriate quality assurance devices and procedures. At the same time partners expected to increase European competitiveness in this field.
A new model predictive control system was combined with a video tracker and a Kalman filter motion predictor to detect surrogate motion and move the PSS to compensate for the detected motion taking into account the software delays as well as the PSS dynamics. The Control Theory Applications Centre (CTAC) and UHCW (Coventry, UK) developed a "breathing" thorax phantom to facilitate the clinical validation of the "breathing couch" developed. The phantoms as well as other simpler electromechanical devices were positioned onto the PSS. Their motion was computer controlled to replicate realistic external as well as internal organ motion. The motion detection was performed using external markers monitored by video cameras or infrared tracking systems (e.g. Polaris). The control algorithms exploited the measured and predicted position to calculate the control action required to move the PSS in real time to enable it to compensate for target movements in such a way that the markers remain as motionless with respect to the room coordinate system and the treatment beam (i.e. the target did not move with respect to the beam).
Kalman filter-based predictors were compared to neural networks, polynomial and bilinear filters. The effect of motion prediction was assessed and a typical prediction horizon of 0.3s selected as the most appropriate to perform motion compensation with a standard Elekta PSS. The effect of motion prediction was experimentally demonstrated and it was shown to clearly improve the overall motion compensation.
Various image tracking algorithms were developed by CTAC, the University of East Anglia (UEA) (Norwich, UK), Universidad de Castilla-La Mancha (UCLM) (Ciudad Real , Spain). The UEA and UCLM focused on shape and volume tracking whereas CTAC focused on surrogate tracking (i.e. points in space). The method developed by UCLM was based on geometrical active contours and prediction. Several clinical tests with different image data including video showed good results in responding to different changes of the topology and contrast regions. ROC analysis was carried out for the different tests, obtaining an average of 97 % sensibility and 99 % specificity.
Irrespective of the topics dealt with in the course of MAESTRO project, at least the following fields were considered to be subject to specific appeal from the medical world and hence to future research programmes:
1. control of RT equipment: improvement in the robotic devices used to deliver radiation treatment were becoming increasingly topical with several manufacturers outside the EU preparing new commercial products for image-guided and adaptive radiation therapy. On the other hand, the increasing complexity of RT devices required new control/management software, in particular for patient and treatment follow-up, fault detection, preventive maintenance. 2.
imaging: improvement of imaging equipment and software remained a topical question, not only for diagnostic and treatment planning, but also for image-guided RT (IGRT), including imaging for intra-fraction motion compensation. In the future, efficiency of treatments may benefit more from improvement of imaging technologies than from linac.
3. small beams: whether delivered thanks to conventional linacs equipped with micro-MLC or devices like "Cyberknife", beams featuring a few cm, and even as narrow as less than 1 cm, were spreading for treatment of cancerous and non-cancerous disorders. One major concern was the relevant dosimetry: primary references, adapted dosimeters featuring small active volumes, dose calculations.
4. "in vivo" dosimetry: although not always applicable, the assessment of the actual dose delivered to the tumour (and possibly to organ at risk) thanks to dose measurements during one or several fractions, was a quality assurance task of interest, which had become compulsory in some countries, including France. There was demand for convenient devices.
5. hadrontherapy: the implementation of proton and carbon beams were considered to remain the source of intensive research. Hadron beams were clearly more conformational and advantageous than conventional treatments (as well as IMRT), at least for certain tumour locations. Prof. J. Bourhis, head of the RT department at IGR, and scientific coordinator of the hadrontherapy project "Archade" in Caen, reported that from published and accepted clinical studies, it is believed that 10 % of all RT treatments would benefit from proton treatments, and 10 % of these (hence 1 % of the total RT) would give even better outcomes with carbon ion beams.
6. dose modulation: while most of present expected deliveries are homogeneous doses to the cancerous volume (even with modulated beams), RT was expected to adapt in the future to more and more numerous prescriptions of modulated doses to the tumour.
7. biology: taking into account individual patient biological data and assessing individual organ sensitivity to the type of radiation delivered, would help to optimize the RT treatment parameters. Then the target of a radiation therapy would have to shift from a "physical" dose delivery (Gy) to an actual biological effect delivery. In addition, targeted RT, based on injected medicine, has already been subject to studies.
8. long-term adverse effects of RT: current RT devices aimed at minimising the short- and medium-term adverse side effects of RT treatments. The risk of secondary radiation-induced cancer or long-term non-cancerous diseases (like heart ischaemia for instance) had been already addressed by several research teams, but this had not yet been taken into account in routine to possibly influence treatment choice. A lot of work remained to be performed.
Furthermore the follow-up of all doses received by patient during their life, either by treatment or diagnostic devices, would certainly be of interest.
A new model predictive control system was combined with a video tracker and a Kalman filter motion predictor to detect surrogate motion and move the PSS to compensate for the detected motion taking into account the software delays as well as the PSS dynamics. The Control Theory Applications Centre (CTAC) and UHCW (Coventry, UK) developed a "breathing" thorax phantom to facilitate the clinical validation of the "breathing couch" developed. The phantoms as well as other simpler electromechanical devices were positioned onto the PSS. Their motion was computer controlled to replicate realistic external as well as internal organ motion. The motion detection was performed using external markers monitored by video cameras or infrared tracking systems (e.g. Polaris). The control algorithms exploited the measured and predicted position to calculate the control action required to move the PSS in real time to enable it to compensate for target movements in such a way that the markers remain as motionless with respect to the room coordinate system and the treatment beam (i.e. the target did not move with respect to the beam).
Kalman filter-based predictors were compared to neural networks, polynomial and bilinear filters. The effect of motion prediction was assessed and a typical prediction horizon of 0.3s selected as the most appropriate to perform motion compensation with a standard Elekta PSS. The effect of motion prediction was experimentally demonstrated and it was shown to clearly improve the overall motion compensation.
Various image tracking algorithms were developed by CTAC, the University of East Anglia (UEA) (Norwich, UK), Universidad de Castilla-La Mancha (UCLM) (Ciudad Real , Spain). The UEA and UCLM focused on shape and volume tracking whereas CTAC focused on surrogate tracking (i.e. points in space). The method developed by UCLM was based on geometrical active contours and prediction. Several clinical tests with different image data including video showed good results in responding to different changes of the topology and contrast regions. ROC analysis was carried out for the different tests, obtaining an average of 97 % sensibility and 99 % specificity.
Irrespective of the topics dealt with in the course of MAESTRO project, at least the following fields were considered to be subject to specific appeal from the medical world and hence to future research programmes:
1. control of RT equipment: improvement in the robotic devices used to deliver radiation treatment were becoming increasingly topical with several manufacturers outside the EU preparing new commercial products for image-guided and adaptive radiation therapy. On the other hand, the increasing complexity of RT devices required new control/management software, in particular for patient and treatment follow-up, fault detection, preventive maintenance. 2.
imaging: improvement of imaging equipment and software remained a topical question, not only for diagnostic and treatment planning, but also for image-guided RT (IGRT), including imaging for intra-fraction motion compensation. In the future, efficiency of treatments may benefit more from improvement of imaging technologies than from linac.
3. small beams: whether delivered thanks to conventional linacs equipped with micro-MLC or devices like "Cyberknife", beams featuring a few cm, and even as narrow as less than 1 cm, were spreading for treatment of cancerous and non-cancerous disorders. One major concern was the relevant dosimetry: primary references, adapted dosimeters featuring small active volumes, dose calculations.
4. "in vivo" dosimetry: although not always applicable, the assessment of the actual dose delivered to the tumour (and possibly to organ at risk) thanks to dose measurements during one or several fractions, was a quality assurance task of interest, which had become compulsory in some countries, including France. There was demand for convenient devices.
5. hadrontherapy: the implementation of proton and carbon beams were considered to remain the source of intensive research. Hadron beams were clearly more conformational and advantageous than conventional treatments (as well as IMRT), at least for certain tumour locations. Prof. J. Bourhis, head of the RT department at IGR, and scientific coordinator of the hadrontherapy project "Archade" in Caen, reported that from published and accepted clinical studies, it is believed that 10 % of all RT treatments would benefit from proton treatments, and 10 % of these (hence 1 % of the total RT) would give even better outcomes with carbon ion beams.
6. dose modulation: while most of present expected deliveries are homogeneous doses to the cancerous volume (even with modulated beams), RT was expected to adapt in the future to more and more numerous prescriptions of modulated doses to the tumour.
7. biology: taking into account individual patient biological data and assessing individual organ sensitivity to the type of radiation delivered, would help to optimize the RT treatment parameters. Then the target of a radiation therapy would have to shift from a "physical" dose delivery (Gy) to an actual biological effect delivery. In addition, targeted RT, based on injected medicine, has already been subject to studies.
8. long-term adverse effects of RT: current RT devices aimed at minimising the short- and medium-term adverse side effects of RT treatments. The risk of secondary radiation-induced cancer or long-term non-cancerous diseases (like heart ischaemia for instance) had been already addressed by several research teams, but this had not yet been taken into account in routine to possibly influence treatment choice. A lot of work remained to be performed.
Furthermore the follow-up of all doses received by patient during their life, either by treatment or diagnostic devices, would certainly be of interest.