PARTNER. Grant Agreement Number WP13 - D.1 Report on basic principles for treatment planning software for light ions.

Transcription

1 PARTNER Grant Agreement Number WP13 - D.1 Report on basic principles for treatment planning software for light ions. Joanna Gora Host Organisation EBG MedAustron Supervisors: Ao. Univ.-Prof. Dr. Ramona Mayer Univ. Doz. DI Dr. Dietmar Georg Date: April 2010

2 1. Introduction Treatment planning system (TPS) is a vital part of radiotherapy process, modeling the interaction of beams with the virtual patient created with the diagnostic image data sets. Computerized optimization of a treatment plan has been introduced in radiation therapy since computers become available to hospitals in the 1960s. However at that time, those early approaches had many limitations (like low computing power) and did not find their way into widespread clinical practice [1]. The common way of treatment planning was the manual manipulation of standard isodose charts on the patient body contours that were generated by direct tracing or lead wire representation, and relied on judicious choice of beam weight and wedging by an experienced dosimetrist and physicist [2]. Evolution in computer technology led not only to development of treatment planning system but also diagnostic imaging. The simultaneous progress in computed tomography (CT) and increasing computing power in 1970s was the basis for computerized treatment planning as it is understood today. CT scanning was ideally suited for radiation therapy planning, since it provides for the first time, an easy ability to localize the tumor and surrounding normal tissues on the patient s axial anatomy. Moreover, it provided an ability for calculating dose distributions, which account for the real tissue densities within the patient [3,4]. Successive improvements in treatment planning hardware and software have been most notable in the graphics, calculation and optimization aspects of current systems. Today, treatment planning has evolved such that full 3-D planning capabilities are possible including patient data from a variety of different sources. Image data from CT and MRI, PET or in some cases ultrasound can be used to define planning target volumes in 3-D with high precision, which in particle therapy is even more important than in photon therapy. Modern 3-D treatment planning systems allow for virtual simulation of the patient by superposition of radiation beam geometries at any orientation on any image, i.e. combination of beam s eye view (BEVs) with the anatomy. Furthermore, plan analysis tools such as dose-volume histograms (DVH) and biological objectives are assisting the choice of techniques and provide a tool for plan optimization [3]. Dose calculation has evolved from simple 2-D models through 3-D to Monte Carlo techniques. Treatment planning is a long and complex process and treatment planning systems (TPS) are a huge help for related professionals in order to design and deliver a proper treatment. 2. Treatment planning system hardware The main hardware components of each TPS are: Central processing unit (CPU), A graphics display, Memory, Digitizing devices, Archiving and network communication devices, Output devices. CPU should have at least the memory and speed required by the operating system and treatment planning software. Graphics display is normally sufficient for accommodating the patient transverse anatomy on a1:1 scale or larger with the highest possibly resolution. Graphics speed may be enhanced with video cards or hardware devices. Memory and archiving functions are carried out through either removable media or network on the remote computer or server. Archiving operation may include beam data and parameters, patient related data such as CT scans and dose distributions, and data used for setting up the patient for the treatment on a linac with record and verify systems. Digitizing devices (today less popular) are used to acquire manually entered patient data such as transverse contours and BEVs of irregular shapes. And finally for hard copies, output devices like printers and plotters are needed. As hardware capabilities tend to change quickly, the general approach is to purchase equipment having the highest current specifications with ability for future upgrade [3]. 3. Treatment panning system software 3.1 Data acquisition and entry PARTNER GA number Prof. Manjit Dosanjh 2

3 3.1.1 Patient data The first step of the TPS concerns is the acquisition and manipulation of computed tomography (CT) data and contours, i.e. target volume (GTV, CTV, PTV) and the organs at risk (OARs) in order to generate a three-dimensional digital model of the irradiation region. The common way of transferring patient image data to the TPS is via the Digital Imaging and Communications in Medicine (DICOM) data format. To ensure accurate dose calculation, the CT numbers must be converted to electron densities and powers. The conversion of CT numbers to electron density and power is usually performed with a user defined look-up table, which in turn is generated using a water equivalent circular phantom containing various inserts of known densities simulating normal body tissues such as bone and lung [2,5]. The definition and delineation of target and OARs can be achieved either by manual contouring on TPS (using software tools provided by vendor) or by using auto-segmentation s. Those automatic contouring routines can help in outlining organs or regions of bulk density. According to the ICRU report 78 target volumes in particle therapy are defined as follow: GTV- Gross tumour volume (generally the volume discernable from imaging), CTV- Clinical target volume (expansion of the GTV for uncertainty about tumour extent), PTV- Planning target volume -expansion of the CTV for geometric uncertainty (setup variation, organ motion). Not only target volumes have to be delineated but also normal tissues and organs (OAR) whose radiation sensitivity can significantly influence treatment planning and the prescribed dose [6]. Patient anatomy may be displayed using the BEV capability of the TPS. The rendering of patient anatomy from the point of view of the radiation source is useful in viewing the path of the beam [2] Machine and beam data In order to simulate the possible mechanical motions and limits of the treatment machine, the various mechanical specifications of the machine must be specified in the TPS. The gantry, table and collimator rotation conventions used in a particular institution must be described accurately and the angle convention must be fully understood by the user. The TPS must also be able to distinguish between jaw pairs and to accurately model the limits of the jaw over-travel. Physical and virtual wedges use by the TPS will be limited by field sizes that are smaller than the maximum field setting in both the transverse and longitudinal directions. Linacs capable of producing IMRT fields may do so via step and shoot or fully dynamic techniques. For these types of treatment the TPS requires data regarding the maximum speed and the characteristics of the maximum rise in the beam-on time and information on maximum dose rates. The TPS may also require information regarding isocenter, source to cone distance (SCD), source to skin distance (SSD), nominal reference distance or percent depth dose data (PDD). Finally specific information related to the beam characteristics like data concerning the direction of measurement, energy, field size, wedge type and orientation and other relevant parameters need to be implemented [2,7]. 3.2 Beam design and dose calculation s Beam delivery techniques In particle therapy, a number of different beam shaping and delivery techniques can be used, and these techniques strongly affect the selection of beams and their resulting dose distributions. The planning software must therefore be able to simulate all techniques of particle delivery available to the user: Scattered beams: broad beam, generally to produce a near-uniform dose distribution within the target volume, Wobbled beam: where a spot beam extracted from an accelerator is laterally spread out by using a pair of wobbler magnets and a scatterer. This technique produces a near-uniform dose within the target volume for each beam, Scanned beam (continues or discrete): can either produce a near-uniform dose distribution or, more usually, a highly non-uniform dose distribution within the target volume for each beam, it is thus suitable for intensity modulated particle therapy [6]. PARTNER GA number Prof. Manjit Dosanjh 3

4 It is common to select the direction of a beam so as to entirely or partially avoid OARs in all radiation modalities. In charged particles therapy, it is also desirable to avoid beam directions that pass through the complex or high-z heterogeneities or that lie parallel to a tissue-air interface. For particle therapy, beam modifiers devices around the target must be created, especially when broad beams are used. Every clinically used TPS should be able to design those and take them into consideration during dose calculation. Blocks: Field shielding is accounted for in the TPS by considering the effective attenuation of the block to calculate the total dose. TPSs are able to generate files for blocked fields that can be exported to commercial block cutting machines. MLC (multileaf collimator): An MLC is a beam shaping device that can replace almost all conventional blocks, with the exception of island blocking and excessively curved field shapes. Most modern linacs are now equipped with MLCs [2]. Range Compensator: it is a 3D device made of low-z material to attenuate radiation, such that the prescribed dose conforms to the distal surface of the target. TPSs are able to generate files for compensators that can be read by commercial compensator cutting machines. This is for scattered and wobbled beams. If a scanned beam is applied, a physical compensator could still be used. However, in practice, the variation in particle beam penetration is achieved by upstream changes in pencil beam energies [6,10]. In the particle therapy a TPS must take into the account the gap between the physical compensator and patient. In practice, an air gap is left between compensator and patient surface, the larger air gap is the greater effects of increasing penumbra and blurring can be observed. For scanning the following additional information has to be taken into the account by the TPS: Grid Spacing: The grid spacing is determined by the requirement that a sufficient overlap exists between the dose distributions delivered by adjacent pencil beams, to allow for construction of smooth dose distributions [11]. A good optimization strategy might be to start with low-resolution grid and re-optimize it with a high-resolution grid (2-3mm). However, it is important to remember that for the lower resolution less spots might be identified to be in the target volume then for high resolution [10]. When designing a spot scanning beam scan volume parameters must be entered in the TPS such as shape, size, depth of the scanning pencil beam and number of layers the target will be treated with Dose calculation s Dose calculation s are the most critical software component in a computerized TPS. The intent of a dose calculation is to predict, with as much accuracy as possible, the dose delivered to any point within the patient. Due to the complexity of radiation interaction with the human tissues and due to the practical need for rapid calculation times, such dose calculation s have limitations due to the approximations used in the physical models. Usually the most complex s have fewer uncertainties compared to the simpler s although this is usually at the expense of longer calculation times [3]. There are three primary models for the calculation of dose in particle therapy, namely: Uniform-intensity beam s Uniform-intensity beam dose s provide the simplest, fastest, and least accurate approach for estimating dose. The input data to the dose calculation are either experimentally measured data, or numerical fits to experimentally measured data, which describe dose distribution of various uniform-intensity beams irradiating a flat-surfaced water-equivalent phantom. Typically for particle therapy, these may consist of depth-dose curves for spread-out Bragg peak (SOBP) together with lateral dose profiles at representative depths. From these data, the dose at any point in the distribution can be derived by calculating its water equivalent depth (the integral in depth of the water equivalent densities, from the patient surface down to the point of interest), usually performed using a ray tracing, and interpolating the resulting dose from the measured depth-dose data. Uniform-intensity beam s can be used for both scattered and wobbled beams [6]. Pencil beam s In order to use the advantages of particle therapy, active scanning treatment methods have been developed and pencil beam s are used here to compute the dose. Compared to passive techniques, where a small number of broad fields are used, many (up to tens of thousands) narrow particle beams (spots) are used for irradiation in active scanning. With pencil beam s more accurate modelling of dose can be achieved. Typically, the incident beam is modelled using a number of closely spaced finite pencil beams, with each pencil beam PARTNER GA number Prof. Manjit Dosanjh 4

5 being assigned a weight that is directly proportional to the particle fluence of the beam for the pencil s position. Each pencil beam broadens because of multiple Coulomb within the patient, and its lateral shape can be modelled using measured or calculated data. The resultant dose at any point is then computed by summing the contributions from each of the pencil beams, with each calculated point taken to be its actual water-equivalent depth [6,7]. These approaches have a number of advantages. Because pencil beam passing outside of the collimator aperture are rejected, the penumbral effect of the collimator is automatically calculated. In addition, inhomogeneous beam intensities can be easily modelled, which becomes important when intensity-modulated beams are to be design or their dose distributions to be calculated [6]. Monte Carlo s (MC s) The most reliable method for calculating the detailed transport of particle beams in tissues is based on MC simulations. In this approach, individual particles are tracked as they penetrate trough the patient and interact with the material through which they pass. The likelihood of an interaction, and its consequences, is sampled using random numbers, from the best available probability distributions. For MC s it is easy to model Coulomb interactions that lead to energy loss and of particles. However, to obtain sufficient statistical accuracy for useful dose distributions in practical situations, tens of millions of histories usually must be traced. Such calculations can take hours or even days to process [5,6]. In near future MC s will be preferred for treatment planning in particle therapy. Recently, a simplified Monte Carlo (SMC) planning for proton beams based on measured depth dose distributions in water has been developed. Relatively short calculation time SMC agrees well with experimental results in a heterogeneous phantom. However, further investigations will be necessary to quantify the calculation time as well as the influence of these simplifications on the results. Another option is the implementation of established general purpose MC codes like GEANT4 or FLUKA for treatment planning. However, these codes are still too slow for clinical dose calculations. For particle therapy treatment planning, fast and accurate MC dose calculation s are currently being developed [6,8,9] Compensation for heterogeneities Particle ranges are very sensitive to the material crossed. The complex density heterogeneities encountered by particles in patient bring special problems. As the range is depend on the heterogeneities, than the actual delivered dose will be sensitive to the exact positioning of those heterogeneities in relation to the particle beam [13]. Because of the influence of heterogeneities, a map of heterogeneities along the beam path must be made and compensated for. Finally the dose distribution must reflect the remaining effects of the heterogeneities. The map of heterogeneities is built up from fine resolution CT images converted to water equivalent densities in order to compute 3D dos distribution. The resulting requirements for planning particle therapy imply the following: To ascertain the CT Hounsfield number to water equivalent density conversion table, To compensate, either physically or virtually, for heterogeneities, including metallic implants when present, To take into account uncertainties associated with the possible misalignment of the compensator with the patient, Be aware of the possible hot-and-cold spots due to lateral effects, Take into account uncertainties in particle beam penetration. For example, choose the right beam direction (i.e. avoid beam direction that pass through complex of high-z heterogeneities) or designing suitable margins. 3.3 Image display and DVHs BEVs and room eye views (REVs) are used by modern TPSs. BEVs are projections of treatment beam, field limits and outlined structures through the patient. The REV gives the user a perception of the position of the gantry and table with respect to each other and may help in avoiding potential collisions when moving from the virtual plan to the actual patient set-up. DRRs are projection images generated by mathematically passing ray lines through the patient CT data. Electronic Portal Imaging & Dosimetry (EPID) generation can be accomplished by TPSs by substituting energy shifted attenuation coefficients for CT data sets. These virtual portal images with the treatment field superimposed can be used for comparison with the portal images obtained with the patient in the treatment position on the treatment machine [2]. PARTNER GA number Prof. Manjit Dosanjh 5

6 There are many ways of describing dose-summarizing parameters. The most popular one is displaying dose-volume histograms (DVH). A cumulative DVH is a graph of the volume of specified VOI (volume of interest) that receives at least a given dose as a function of that dose. DVHs can employ both relative or absolute volumes, and relative of absolute doses. The DVH is a helpful quantitative tool to summarize the dose distribution to a given VOI. However, the disadvantage of it is that all spatial information is lost [6]. 3.4 Optimization Optimization routines are provided by TPSs with varying degrees of complexity. Algorithms can modify beam weights and geometry or calculate beams with a modulated beam intensity to satisfy the treatment objectives. Plan optimization is a process of iteratively generating and then automatically assessing a large number of plans and choosing the best among them. The mathematical meaning is to choose those values for treatment-delivery variables that would result in an extremum of the score function. Software may design a treatment plan automatically, given the following: treatment objectives, a method of giving a score to a plan, definition of what variables to adjust, an initial guess at a plan and a method of searching for that set of plan variables. Although the process of having a computer to generate best plan is automatic, modifying constrains, goal and trying is commonly perform by the user. A good optimization should be insensitive to the starting values used for the variables being adjusted. Variables that are not adjusted in the iterative process do need to be specified with care as they will remain in the final plan [6]. Intensity modulated proton therapy (IMPT) is an active scanning technique using the approach of inverse treatment planning. Each spot is weighted in a computer-based optimization procedure to find the optimal solution for the prescribed dose in the target volume and for sparing critical organs as much as possible [8]. Several optimization s have been proposed for radiation therapy. In my project I am using software provided by Elekta CMS called XiO. Below, as an example, are explained IMPT prescription parameters for this system, which must be defined by the user: The XiO system lists all structures that have been contoured for particular patient. Each structure to be optimized must be assigned a type. It can either be a Target or an OAR. The next parameter is Rank: each of structures should have a rank, which determines how the IMRT optimizer will treat the voxels in the volume where structures overlap. The structure that is assigned with the higher priority will own the voxels in the overlap region. Therefore, the objective function for the higher priority structures will include the contribution of these voxels but the objective function for the lower priority structure will not include these voxels. One (1) is considered as the highest rank, higher numbers (2,3,4 ) have lower priority. For example, the patient structure would always be given the lowest rank priority. The next step is to assign the Objective, Dose (cgy/gy) and Volume (%). For a target minimum and maximum dose objectives can be set, these are min and max doses we want to achieve within that structure. The optional parameter here is a Goal, by default; the goal is set half way between the max and min. The goal is the point between the minimum and maximum where there is no penalty applied. For OAR only maximum dose can be chosen. This objective sets the maximum dose we want to achieve within that structure; the min dose will be zero. Placing dose volumes objective (volume % value) can help shape the DVH curve. An optional parameter is the threshold dose. The default threshold is zero, since ideally we wish all doses to OAT to be zero. Weight is an option to increase the relative importance of an objective. It might be use when decrease of the dose over the entire volume of structure is needed. Power is an option to increase the penalty to those voxels with doses in violation of the structure s objective. A small increase in power can make a large change in the objective function. The optimizer, in seeking to minimize the overall numerical penalty, will try harder to achieve the cord dose objective [10]. 3.5 Relative Biological Effectiveness (RBE) Different radiation qualities have different degrees of effectiveness in producing effects in biological systems. When radiation is absorbed in biological materials, the energy is deposited along the tracks of charged particles in a pattern that is characteristic of the type of radiation involved. Figure 1 shows various RBE values for different types of radiation [17]. Clinical proton beam therapy has been based on the use of a generic relative biological effectiveness (RBE) of 1.0 or 1.1. Since protons are low-let particles, they behave radiobiologically like photons; there is no clear biological advantage of using protons over photons. However, it is well known that the RBE increases with depth in the spread-out Bragg peak (SOBP) (Paganetti 2003 Technol. Cancer Res. Treat ) and can reach values of in vitro at the distal edge. This needs to be considered in treatment planning, particularly for single field plans or in plans where the particle end of range is close to a critical structure. PARTNER GA number Prof. Manjit Dosanjh 6

7 Figure 1. RBE of various radiation types [17] Even greater need for considering RBE in the treatment planning process is for carbon ions. Those high-let particles have higher biological effectiveness compared to photons. This is the main rationale for using carbon in radiation therapy. Treatment planning for ion beams has to be based on the RBE corrected dose instead of the physical dose. There are two main approaches of including radiobiological effects into treatment planning for ion beam therapy. In NIRS (National Institute of Radiological Science) in Chiba, Japan, the passive technique is used. A fixed RBE value is there applied, which is obtained from clinical neutron experience and experimental data (LET-RBE spectra from killed cells exposed to different ion beams) [14]. A different approach is used at GSI (Gesellschaft für Schwerionenforschung mbh) and HIT (Heidelberger Ionenstrahl-Therapiezentrum), Germany. RBE values are estimated by using the Local Effect Model (LEM). The basic principle of the local effect model LEM is to convolute the nonhomogeneous dose distribution in the particle track with the non-linear photon dose effect curve. With this procedure the effects of the particle can be calculated on the basis of the photon dose effect curve [15]. The LEM is ideally suited for use as an integral part of treatment planning code systems for active dose shaping devices like the GSI raster scanning system. Thus it has been incorporated into GSI homemade treatment planning system for ion therapy (TRiP) [16]. 3.6 Normalization and Monitor Units (MU) calculation Beam time and MU calculations by TPSs are optional. The calculation process is directly related to the normalization method. The output dose per MU (d/mu) of a therapeutic radiation beam is traditionally calibrated under specific reference conditions. These conditions include beam energy, field size, suitable depth in water or water equivalent phantom in a low dose gradient region with known relative depth dose, and source to point of calibration distance [12]. Ideally, the dose distribution produced by a planning calculation should provide a 3D map of the absolute dose throughout the patient volume that would result from specified beam input intensity without further normalization. This is not always the case, however. Some s provide only the relative dose distribution. This is often normalized to some well defined point (ICRU reference point), which is said to receive 100% dose, but it can be normalized in many other ways [6]. 4. Commissioning and Quality Assurance 4.1 Sources of uncertainties Uncertainties and errors in TPSs may arise from any of the many steps involved in the treatment planning process. Sources of uncertainties in treatment planning process include: Mechanical treatment machine related uncertainties: these include gantry rotation, collimator rotation, shielding block repositioning, and field size settings. PARTNER GA number Prof. Manjit Dosanjh 7

8 Dosimetric treatment machine related uncertainties: these include accuracy and reproducibility of the monitor ionization chamber, reproducibility of field flatness (for scattered beams only), collimator opening. Imaging related uncertainties: issues related to imaging accuracy such as geometrical distortion in MR or beam hardening in CT, issues related to data transfer and conversion from CT numbers to electron densities, inaccuracies in image registration and resolution limitations. Patient related uncertainties: include patient repositioning and organ motion during one of the steps of the planning and treatment process such as CT scanning, simulation and treatment. Patient changes in weight and tumour shrinkage. Uncertainties in the definition of target volume and normal tissue localization: it has been shown that target volume determination is one of the larger uncertainties in the entire treatment process. Beam measurement uncertainties: include uncertainties related to detectors, the size of the detectors, the precision and accuracy of the measurements. Dose related uncertainties: these are related to beam measurement uncertainties as well as limitation in the calculation s. Dose evaluation uncertainties: DVHs provide a means of optimizing treatment plans. Inaccuracies in volume determination and dose binning procedures could impact the accuracy of DVHs. Biological modelling: because of complexity of biological response and the simplicity of the most existing models, these models have major limitations in their capability of predicting tumour control probabilities and normal tissue control probabilities [3]. 4.2 Quality Assurance (QA) In general, quality assurance involves tree steps: - The measurement of the performance, - The comparison of the performance with a given standard, - The action necessary to maintain or regain the standard. For TPS, the commissioning provides the standard by which the system must be maintained. Thus, once the TPS is commissioned for clinical use, on-going QA must be performed to ensure the integrity of the data files and the reproducibility of the calculations. The frequency of these tests and the acceptance criteria should be established based on the user s specific needs or on national or international norms [2,3]. Below is a list of publications, where guidelines and recommendations can be found regarding QA for treatment planning systems: IAEA TRS No.430, 2004: Commissioning and Quality Assurance of Computerized Planning Systems for Radiation Treatment of Cancer, IAEA-TECDOC-1540, 2007: Specification and Acceptance Testing of Radiotherapy Treatment Planning Systems, IAEA-TECDOC-1583, 2008: Commissioning of Radiotherapy Treatment Planning Systems, AAPM Report 55, TG 23, 1995: Radiotherapy Treatment Planning Dosimetry verification, Van Dyk J, Commissioning and QA of treatment planning computers, IJROBP 26, p26, 1993, AAPM Report, TG 53, 1998: QA for Clinical Radiotherapy Treatment Planning. 5. Overview on commercially available TPS With the growing popularity in using particle therapy, continuing development and improvement in treatment planning software (TPS) has been observed. TPSs are advancing mainly with improved dose calculation and optimization s, increased calculation speed and higher functionality. Currently, there are a few leading companies on the market, which provide software for particle therapy treatment planning (see Table 1). All of them provide also solution for advanced photon and electron beam therapy, however those are not considered in the following. PARTNER GA number Prof. Manjit Dosanjh 8

9 Table 1 Commercial TPSs Company TPS name For protons For carbon ions Dose Supported treatment modalities Other information Webpage ELEKTA XiO Pencil beam Varian medical systems Eclipse Pencil beam, broad beam Phillips - Pencil beam Siemens Nucletron RaySearch Labolatories Clatterbridge Centre for Oncology Syngo PT Planning Helax- TMS Ocentra MasterPl an Proton RayStatio n Pencil beam Pencil beam Pencil beam Pencil beam, Monte Carlo EyePlan Simplified broad beam PerMedics Odyssey Convolution dose developed at Loma Linda University scanning and passive for protons, for carbon ions only method scanning and passive modalities scanning scanning Passive scanning and passive scanning and passive Passive Passive, active scanning in progress Will be available soon One of the first TPS for proton on the market Cooperation with RaySearch labolatories, not available yet IMPT optimization accounting for range and position uncertainties Treatment of ocular lesions elekta.com/ varian.com/ healthcare.p hilips.com/in /products/ro s/ medical.siem ens.com/ nucletron.com r aysearchlabs.com/ - permedics.co m/ At present, there are more proton facilities operating all over the world than carbon ones. This is reflected in the number of available treatment planning systems for heavy ions. Only two companies ELEKTA and Siemens provide suitable software solutions. XiO (from ELEKTA) is suitable for both, proton and carbon therapy. It supports active scanning and passive modalities for protons, using pencil beam for a dose calculation, while for carbons only method is available. This type of TPS is for example in NIRS, Japan but they use their in house developed dose calculation and optimisation. On the other hand, the Syngo PT Planning (Siemens) was design exactly for active scanning purposes. The dose is calculated for both particles by pencil beam s. Moreover, for carbon ions the biological effects are taken into the account by applying the LEM model, while for protons a fixed value is used 1.1. This type of TPS can be found, for example in CNAO (Centro Nazionale di Adroterapia Oncologico), Italy. Among proton TPSs there is a higher variety. Helax-TMS (Nucletron) was one of the first TPSs for protons available on the market. Currently Nucletron and RaySearch Laboratories have decided to join their forces in order to develop leading-edge software for proton planning. Their agreement will result in a new treatment planning and optimization module with protons, which will be fully integrated in Oncentra MasterPlan. The MasterPlan proton planning will support both passively scattered beams and several modes of beam scanning, including Intensity Modulated Proton Therapy (IMPT). The next company which is planning to release a new product, dedicated to proton planning, is Phillips. Similarly, their TPS will have a pencil beam and support active scanning. RaySearch develops software for radiation therapy of cancer, their products are PARTNER GA number Prof. Manjit Dosanjh 9

10 specially designed to optimize radiation therapy. For proton treatment planning they propose RayStation, an advanced treatment planning tool. Their proton therapy optimization supports a variety of proton treatment modalities and beam lines, such as pencil beam scanning, uniform scanning. The dose calculation can be performed with either their own Monte Carlo or a Fermi-Eyges based pencil beam. For pencil beam scanning, the optimization tools include IMPT, SFUD (Single Field Uniform Dose) and Distal Edge Tracking. Quite unique in RayStation is the robust optimization, which makes the plans robust to geometrical and patient density uncertainties [18]. Odyssey (formerly OptiRad) was developed at Loma Linda University Medical Center and it supports combined proton, photon and electron planning. Odyssey has precise, yet fast dose s. It uses a convolution dose developed at Loma Linda University, which handles scatter variables in volumetric manner. Works in progress include the improvement of IMPT method and 4D Planning. EyePlan developed at Clatterbridge (Goitein et al 1983) is today the most widely used. It is the standard tool for treatment planning of ocular tumours and the treatment itself is well established and very successful; tumour control rates over 95% have been reported. The dose is calculated on a spherical model of the patient s eye and adopts several approximations for proton dose planning. Its is based on a very simplified broad beam method [19]. As discussed above there are several commercially available treatment planning systems for ion beam therapy. However some facilities, especially research institutes prefer to use their own, in house developed software. One of the examples is PSI (Paul Scherrer Institute), Switzerland. PSI developed the spot scanning technique together with the corresponding software for proton treatment planning, where pencil beam for dose calculation is used. The other example is the program TRiP (TReatment planning for Particles), which handles all ion specific tasks. TRiP was developed for an active beam shaping system for carbon ions at GSI facility (Gesellschaft für Schwerionenforschung), Germany. Dose calculation is performed with a pencil beam and the biological effects are taken into the account with use of LEM model (developed also in GSI). Moreover, the standard version of TRiP has been extended to full 4D functionality. The 4D version of TRiP allows to calculate dose distributions in the presence of motion. Furthermore, for motion mitigation techniques tracking, gating, rescanning, and internal margins optimization of treatment parameters has been implemented. 4D calculations are based on 4D computed tomography data, deformable registration maps, organ motion traces, and beam scanning parameters [20]. The treatment planning system is an extremely dynamic field of radiotherapy. The information presented above is only a snapshot of the current state of the TPSs and in the next few years further development can be expected. Although for proton and carbon ions TPSs evolves quickly, very little work has been done for other particle (e.g. helium, neon, oxygen). The reason being is that so far only proton and carbon ions found application in clinical practice but advantages of light ion therapy are still investigated. 6. Summary Advanced treatment planning tools play a key role in today s particle therapy.treatment planning is the heart of the radiation therapy process. Since TPSs have been introduced in clinical use, it has been further improved. TPS is essential and powerful tools and today radiotherapy could not be performed without them. It is a combination of hardware and software components that allow user to produce and display calculated dose distributions of prescribed treatment. A TPS is a rapidly evolving modality and in the future the significant development can be expected. The modern treatment planning solutions will have a full support for adaptive therapies taking the full advantage of the 4D imaging. Improvement will be also seen in dose calculation s making them more precise. As the computer speeds and memories increase also the ability to use MC dose calculation for routine treatment planning is fast approaching. TPSs seek further improvement also in radiobiological optimization because taking biological effects into the account is especially important with heavy ion therapy. Father development of treatment planning systems is essential for using the particle therapy in its whole potential. PARTNER GA number Prof. Manjit Dosanjh 10

11 References: [1] Schlegel W, Bortfeld T, New Technologies in Radiation Oncology, Springer Berlin Heidelberg, 2006 [2] Podgorsak E. B, Radiation oncology physics, A handbook for teachers and students, IAEA, Vienna, 2005 [3] Van Dyk J, The Modern Technology of Radiation Oncology: A Compendium for Medical Physicists and Radiation Oncologists, Medical Physics Publishing Corporation (October 1999) [4] Battista J, Rider W, Van Dyk J, Practical aspects of radiotherapy planning, Int. J. Radiat. Oncol. Biol. Phys. 6: (1981), [5] Bourhaleb F, Marchetto F, A treatment planning code for inverse planning and 3D optimization in hadrontherapy, Computers in Biology and Medicine 38 (2008) [6] ICRU Report 78, Prescribing, Recording, and Reporting Proton-Beam Therapy, Journal of the ICRU, Oxford University Press, 2007, Vol. 7 No 2. [7] XiO Manual, Computerized Medical Systems, Inc, 2002, [8] Soukup M, Fippel M, A pencil beam for intensity modulated proton therapy derived from Monte Carlo simulations, Phys. Med. Biol. 50 (2005) [9] Kohno R, Takada Y, Experimental evaluation of validity of simplified Monte Carlo method in proton dose calculations, Phys. Med. Biol. 48, ~2003!. [10] XiO Proton Training Guide, Computerized Medical Systems, Inc, 2002, [11] Weber D, Rutz H, Results of spot-scanning proton radiation therapy for chordoma and chondrosarcoma of the skull base: The Paul Scherrer Institut experience, International Journal of Radiation Oncology Biology Physics Volume 63, Issue 2, 1 October 2005, Pages [12] Sahoo N, Zhu XR, A procedure for calculation of monitor units for passively scattered proton radiotherapy beams, Med Phys Nov;35(11): [13] Lomax A J, Intensity modulated proton therapy and its sensitivity to treatment uncertainties 1: the potential effects of calculational uncertainties, Phys. Med. Biol. 53 (2008) [14] Furusawa Y, A method to estimate LET-RBE on cell killing for unknown heavy ion particles, II NIRS-CNAO Joint Symposium on Hadrontherapy, 2010 [15] gsi.de/documents/doc-2007-jul pdf [16] Krämer M, Treatment planning for heavy-ion radiotherapy: calculation and optimization of biologically effective dose, Phys. Med. Biol. 45 (2000) [17] nirs.go.jp/eng/research/charged_particle/index.shtml [18] Fredrikkson A, Stochastic and robust methods for handling of range uncertainties in proton therapy optimization, Poster session P6-10, PTCOG 49, Japan [19] Cirrone G, Monte Carlo validation of EYEPLAN proton therapy treatment planning, Nuclear Physics B (Proc. Suppl.) 172 (2007) [20] Bert C, 4D treatment planning for scanned ion beams, Radiation Oncology 2007, 2:24 PARTNER GA number Prof. Manjit Dosanjh 11