Principles and practice of radiation treatment planning

Chapter 10 Principles and practice of radiation treatment planning




Chapter contents



Introduction


Patient data


Megavoltage photon therapy
















Planning tools, calculation and display


















Beam arrangements
















Kilovoltage photon therapy


Electron therapy











Differences between kilovoltage and electron therapy


Special techniques







Treatment plan checking




References


Further reading



Introduction


The goal of treatment planning of external beam radiotherapy is to produce a dose distribution within the patient that will destroy the tumour while sparing as much healthy tissue as possible. The planning role is usually carried out by radiographers, physicists, or technologists with expertise in producing either an optimum plan, or a selection of compromise plans for the clinician to approve. The planner must have sufficient clinical knowledge to understand where compromises may have to be made, either in sparing some of the target volume dose or in accepting higher than desirable doses to radiosensitive healthy tissue.


Most treatment sites will have locally agreed standard treatment protocols and the planner will normally have a limited range of variables with which to optimize the plan.


The planner must take the clinical requirements stated in a prescription and planning request and translate them into the best plan for the individual patient. This must then be presented to the clinician in a clear format so that what in many cases is a complex three-dimensional plan can be readily appraised. Over the years, tools and techniques have evolved to enhance this critical communication between staff groups. The ICRU reports 50 and 62 [1, 2] referred to in Chapter 9 provide an effective methodology of prescribing the volume to be treated and the means of displaying the dose distribution produced by the treatment plan.


The intention of the treatment can drive the complexity of the plan, given that most radiotherapy centres now possess therapy machines capable of delivering highly sophisticated treatments. The cost (i.e. complexity) of the treatment must be balanced against the benefit and, in many instances, the simplest plan is the best plan. The cost may be measured in increased treatment time, which could lead to increased waiting times for other patients. Complex and non-standard plans can also increase the risk of errors.


Where the treatment intent is palliation, for example relief from pain arising from metastatic disease, then a single treatment beam will usually fulfill the purpose. This can be justified as the most important consideration is to give the treatment as soon as possible, with minimal distress to the patient, and the consequences arising from irradiating excess normal tissue do not apply to a patient with a short life expectancy.


Where the treatment intent is curative ‘radical’ treatment, consideration must be given to short- and long-term side effects arising from the irradiation of healthy tissue. The complexity of a treatment plan is likely to increase, especially when an organ at risk (OAR), also referred to as a critical structure, lies adjacent to the planning target volume (PTV), and the prescribed dose greatly exceeds the tolerance dose of the OARs.


It is important that the planner appreciates the radiosensitivity of normal tissue. OARs can be considered as serial (e.g. spinal cord), parallel (e.g. lungs), or a combination of serial and parallel (e.g. the heart). In a serial organ, if any part of the organ receives a dose above its threshold then there will be total loss of function, e.g. paralysis in the case of the spinal cord. In a parallel organ, part of the organ may be severely damaged but the rest continues to function. Therefore, it is important to consider the volume of the parallel organ that is receiving the damaging dose. For example, the severity of lung pneumonitis significantly increases if the dose exceeds 20   Gy so the plan would be evaluated by considering the volume of lung exceeding this dose – ‘V20’. For a patient with two healthy lungs, it would be normal practice to consider both lungs as a single organ.


Table 10.1 gives typical tolerance doses for organs. However, these values should not be taken as absolute, as fractionation affects the ability of normal tissue to repair damage. The quoted dose may represent an increased risk rather than a stochastic threshold above which all patients will be affected. The patient’s clinical condition and drugs may affect radiosensitivity of some organs. The complete clinical picture is also important, for example in the abdomen region, the planner must know whether one of the kidneys is already non-functioning as this will enable them to design a plan that may reduce the dose to the healthy kidney at the expense of dose to the already damaged one.



Treatment planning computers offer a variety of tools used to localize treatment volumes, ‘virtually simulate’ the field arrangement, and to calculate the dose distribution. This chapter covers the production of desired dose distributions to treat a specified PTV and the tools relevant to this process. The majority of radiotherapy treatments are delivered using megavoltage x-rays or electrons from linear accelerators. However, there are some specialized treatment techniques appropriate to a limited range of diseases. These techniques may be delivered with a conventional linear accelerator, or may require specialized equipment such as the Tomotherapy® device or Gamma knife®. The use of modern treatment delivery equipment with dynamic control of multiple simultaneous movements, such as couch, gantry, and multileaf collimators, may be restricted by the treatment planning process as complex software is required to simulate and calculate dose distributions. Specialist software is required to plan treatments for devices such as Tomotherapy and Gamma knife.


The reasoning behind the prescribed dose required to kill a specific tumour type is discussed in Chapter 17. The treatment planner is given this in the plan request and is solely concerned with delivering this dose at the specified fractionation. However, it is important to have the dose and dose per fraction available at the time of planning as the dose limits to critical structures will depend on the dose rate.


Much of radiotherapy dose and fractionation is based largely on clinical results evaluated over decades. However, better knowledge of how specific tumours and normal tissues respond to radiation, in individuals, or groups of patients, could lead to more effective treatments. If reliable radiobiological data relating to both the tumour and normal tissues are available then it may be possible to prepare a plan based on the effectiveness of the dose delivered rather than the physical dose alone. Inverse planning may be performed by optimizing the dose distribution through the use of normal tissue complication probability (NTCP) and tumour control probability (TCP).


Modern treatment planning is a complex process. As well as striving to meet the optimum clinical objectives, it is also important to comply with local practice for the patient set-up instructions, and local protocols established for reliable and safe treatment delivery. For the beginner, this can be frustrating as, even for basic plans, to some extent the process can only be learnt from one’s own mistakes.



Patient data


The patient contours may be acquired in the form of multiple computed tomography (CT) slices or, in some cases, derived from external contours alone, possibly with some estimation of the dimensions and density of significant internal heterogeneities, such as the lung. The detailed anatomy derived from CT images is the optimum means of obtaining accurate dose distributions as the CT numbers can be converted to electron density maps, which the treatment planning computer (TPS) will use to calculate how the beams are attenuated and scattered. However, CT- density tables can also be manipulated to obtain enhanced digitally reconstructed radiographs (DRRs), so it is important to use a table that is specifically intended for dosimetry rather than imaging (see Chapter 9).


CT is the best imaging modality for radiotherapy in terms of dosimetry and spatial accuracy, but does not always provide the best images for tumour or critical structure localization. Normal practice is to register the diagnostically superior magnetic resonance (MR) or positron emission tomography (PET) images onto the CT images (see Chapter 9). However, it is possible to use the MR images directly for some sites, by applying distortion correction to the MR images and assigning ‘bulk heterogeneity’ corrections to visible structures such as bone.


When using a TPS with 3D scatter correction, it is important that the CT scan extends beyond the PTV, ideally by at least 2   cm, so that the adjacent anatomy is accounted for in the calculation and not assumed to be air by the TPS.


The contours used for planning must represent the patient contour during treatment delivery – bearing in mind that treatment will take several minutes, whereas CT scans are acquired in seconds. Most contouring techniques, whether CT or external outlines only, are very rapid ‘snapshots’ compared to treatment delivery times. Therefore, either the planning procedure, or the treatment delivery, must account for patient movement. For the vast majority of treatments, the treatment plan must be designed to account for this patient movement by the addition of margins around the clinical target volume. However, modern technology now offers practical methods of ‘gated’ treatments that only deliver the dose to the patient while they are in the same position as during the therapy CT scan, or ‘tracking’ treatments, that can follow the PTV movement. These are discussed in the final section of this chapter.


Figure evolve 10.1image shows the effect of patient breathing on a thoracic CT image acquired in spiral mode. The appearance of the liver clearly indicates the breathing motion that otherwise might not be apparent.


Radiotherapy patients may have prostheses, dental fillings, breast implants, and other artifacts present externally or internally. Sometimes these will be present during imaging but may be removed by the time treatment commences (for example, drainage tubes). Where the artifact is made of material similar to tissue density such as breast implants (although some breast implants contain high-density magnets), the CT-density table will correctly allocate the density required. However, for metal implants, the CT scan must be acquired with an extended CT scale and the TPS CT-density table must be extended typically to give Hounsfield numbers of up to 32   000 if metal implants are to be identified and the correct density assigned. Even with artifact correction tools on CT scanners, there are usually significant artifacts present surrounding metal objects – especially for bilateral hip implants, and a manual overlay of the density may be required (Figure evolve 10.2image). The use of MR imaging in the presence of non-ferrous implants may provide useful information that may be transferred to the distorted CT dataset.


Cardiac pacemakers present a problem, not only in perturbing the treatment beam, but also in their sensitivity to radiation and electromagnetic interference from linear accelerators. Ideally, pacemakers should not lie in the path of a treatment field and should only receive scattered radiation dose of less than 2   Gy. The cardiologist should be consulted to ascertain the required level of monitoring during and after the treatment.


Contrast may be present in a CT scan, for example, to visualize nodes in head and neck images (Figure evolve 10.3image), or may be present in the bladder as a result of contrast used elsewhere.


This contrast will not usually be present during the actual treatment, but its presence during planning does not usually have a significant impact on the dosimetric accuracy. However, this should be verified whenever a new contrast-based localization technique is introduced.


Care should be taken to ensure that the CT image represents the treatment set-up. Sometimes, the patient cannot be scanned in the same position as for treatment due to restrictions of the CT scanner aperture, particularly when immobilization devices are employed. For some sites, the scan may be performed feet first into the scanner while the patient may be treated in the head first orientation. In this case, the TPS or scanner software may be utilized to mirror and restack the scans into the treatment position. When such techniques are employed, there must be clearly visible quality assurance (QA) tools to verify that the software has processed the images correctly, such as couch markers identifying the couch superior-inferior and lateral orientations.


The CT couch, immobilization supports, bolus, or non-treatment-related devices may be present in the CT scan. The user should be aware of how the TPS will handle, or ignore, structures that lie outside the patient’s external contour and how to compensate for TPS calculations that do not account for beams passing through these structures. High density couch bars on the treatment unit couch may need to be accounted for in plans with beams passing through the couch and, ideally, the positions of these bars should be overlaid on the CT image so that the plan is designed to avoid them.


The planner should also be aware of any anatomical abnormalities in the CT scan that may affect the treatment plan. It may be preferable to avoid regions where surgery has taken place, or to avoid abnormalities unrelated to the disease being treated, such as a hernia.


It is important to appreciate the shortcomings of the TPS algorithms, especially in the presence of heterogeneities. The build-up effect at internal density boundaries may not be accurate and the 3D scatter distribution may not be accurately modelled. These effects typically lead to the TPS overestimating doses in soft tissue regions (e.g. tumours) surrounded by low density tissue (e.g. lung).


For simple point dose calculations, a water phantom can be used to approximate the patient anatomy. This may be performed manually using tabulated data direct from plotting tank measurements, or by a TPS assuming the patient to be an infinite water phantom. In either case, the actual patient contour may differ significantly from the phantom and where the point dose lies near the patient surface or internal heterogeneities, discrepancies of several percent may exist if comparing against calculations performed with full CT data.



Megavoltage photon therapy



Treatment beams


The characteristics of a megavoltage beam must be employed to produce an optimum plan for the variety of clinical requirements.







Wedges


Megavoltage treatment machines have integral beam modifying techniques which are referred to as wedges (see Chapter 8). They alter the uniform dose distribution of the radiation beam in a simple and controlled manner to produce a dose gradient across the field.


Wedges are essential for most multifield plans to produce a uniform PTV dose and serve two purposes.


First, they allow one field to compensate for the depth dose fall off of another field in the plan (Figure evolve 10.4Aimage). Secondly, they compensate for a ‘density gradient’ in the path of the beam (Figure evolve 10.4B, Cimage). The simplest application of this is ‘missing tissue’ when a beam enters an oblique patient contour, but heterogeneities within the patient may also require correction especially the lungs and large pelvic bones.


There are several techniques used in treatment machines for generating wedge-shaped isodoses (see Chapter 8) and the technology used can influence the plan. Fixed wedges may be positioned in the beam for the entire delivery of the beam. This is less commonly used as it requires each field to be manually set up inside the treatment room but, in some machines, may be the only method of permitting a wedge to be orientated in all four directions in the field. These are described in terms of wedge angle of the isodose distribution that they produce (not the physical shape of the wedge itself).


A ‘motorized wedge’ is a physical wedge-shaped filter positioned inside the treatment head and is designed to be in the beam for part of the treatment time. This gives the planner a continuous range of wedge angles from zero (wedge out of the field) to typically 55° (when the wedge is in the beam for the entire delivery of that field). These will typically be described by the TPS as the open/wedge segments of the beam.


A ‘dynamic wedge’ uses no physical wedge in the path of the beam, but generates the wedged isodose effect by moving in one of the collimators across the field during the beam on time. The machine is programmed to deliver either a continuously variable wedge angle or a set of discrete wedge angle values typically 15°, 30°, 45°, 60° and described by the isodose wedge angle that is generated.


For the planner, there are a number of consequences arising from these different technologies.


‘Dynamic wedges’ tend to produce more penetrating isodoses at the ‘thin end’ of the wedge with higher doses in the superficial region of the thin end than physical wedges. Physical wedges (fixed or motorized) tend to have more rounded, blunt, profiles.


The wedge direction may be limited, i.e. not both sets of jaws can generate dynamic wedges, or physical wedges cannot be rotated into both collimator planes. When planning with multileaf collimators, this may limit the effectiveness of the multileaf collimator (MLC) shielding if the wedge orientation restricts the planner to a suboptimal collimator rotation (Figure evolve 10.5image).


Physical wedges (fixed or motorized) significantly attenuate the beam at the central axis and a 60° wedge may typically transmit 25% of the open field dose. Therefore, the treatment time is increased 3–4 fold and this factor can often justify the use of segmented treatments which, although more time consuming to plan, do offer improved PTV dose homogeneity at no additional cost during the treatment delivery.








Intensity modulated radiotherapy


Intensity modulated radiotherapy (IMRT) is a further refinement of this technique in which each of the treatment fields is made up of many segments. The objective is to produce a higher conformity than is achievable with conformal blocked fields. By delivering typically five to seven beams with an individualized intensity profile, the dose is conformed to the volume and normal tissue and critical organs spared. The individual treatment beams for IMRT can be delivered using MLCs in either dynamic or multiple-segment (‘step-and-shoot’) mode. There are advantages and disadvantages of each method. The design of IMRT plans is impractical by conventional manual techniques and is normally done by ‘inverse planning’. The planner specifies the requirements of the plan and the intensity profile is determined by computation to achieve the required dose distribution. The specification must include the relative importance of each requirement as many parameters will be conflicting. For example, if a critical structure lies very close to the PTV and the request is for 10% of the prescription dose to the critical structure but a uniform full dose to the PTV, then this will be impossible to achieve.


Multisegment IMRT is clearly the simpler form of MLC-based IMRT, particularly when a relatively small number of segments are used. As such, the dosimetry is more easily handled, techniques can be implemented using standard 3D planning systems, no complex dynamic collimation control is required so that current MLCs can readily be used. Also, conventional field verification using portal imaging is standard. For complex field modulation, however, many static field segments may be required. There may then be a significant time penalty arising from leaf motion between segments and beam start-up time at the beginning of segments. In particular, the use of small monitor unit increments could result in significant deviations in total dose due to inaccuracies in the dose per monitor unit at small MU settings. Additionally, beam flatness and symmetry should be assessed for small MU increments before such fields are used.


In contrast, while dynamic collimation is potentially more efficient, this advantage may be outweighed by the complexity of the delivery method and beam dosimetry, reproducibility of delivery, maximum leaf speed, the lack of conventional image verification and the need for additional QA dose and fluence measurements.





Planning tools, calculation and display



Algorithms for dose calculation in treatment planning systems


There are several algorithms in use in different treatment planning systems in order to calculate dose to a patient. They have varying degrees of accuracy and, in some treatment planning systems, present a choice between a fast calculation and a more accurate dose calculation. For example, a simple fan-line model may be selected to determine field size, gantry angle, wedge etc. and then a more accurate calculation can be used for the final patient treatment plan.


Algorithms can be represented as follows.



Stored beam data models


Stored beam data models are based on the use of measured data which, typically, are stored as fan-line matrices divergent from the source. Generally, this requires a large number of beam measurements to be taken in order to build up a library of radiation beams of different open and wedged field sizes. The Milan-Bentley algorithm was one of the first models of this type and is still in use today in some planning systems.


This algorithm is also frequently used as the method of point dose checking on an independent system. Measurements required are percentage depth doses (PDD’s) along the Central Axis (CAX) of the beam and off-axis ratios (OAR’s) in the form profiles across the beam (as shown in Figure 10.7A and B), output factors as a function of field size, wedge factors for each wedge, cGy/MU for reference situation with a 10 × 10   cm field and a knowledge of dmax depth. The planning computer can then interpolate these basic data and apply corrections to produce isodose curves in a tissue equivalent phantom. In order to calculate the dose to a patient, a heterogeneity correction is often applied. This can be done by various different methods, such as bulk density equivalent path length, pixel-by-pixel equivalent path length, power law method (Batho) or the ETAR method. Irregular fields are calculated using scatter integration techniques, such as the Clarkson method.






Treatment plan evaluation tools


The ideal radiotherapy treatment plan produces a uniform coverage of the target volume without giving significant dose to surrounding normal tissue. 3D planning is a realistic option for many cases, due to the increased availability of radiotherapy CT scans, virtual simulation, MR, fast computer hardware and improved algorithms. Treatment plans are therefore more difficult to compare and evaluate, hence the need for plan evaluation tools.


Several such tools are in use as discussed.



Isodose distributions


Isodose information is presented as lines or surfaces of equal dose. They show either relative or normalized dose, if the dose is expressed as a percentage of a reference dose or they give absolute dose, if the dose prescription has been included in the treatment plan. It is a matter of local protocol as to which method is used, although prescriptions specifying different doses to different volumes within the same plan are best displayed in absolute dose.


In order to calculate dose to a surface or volume, a matrix of reference points spread over the volume of interest is required. The number of points and their spatial resolution is always a trade-off between speed and accuracy.


The x and y coordinates of these points are in the transverse plane and the z coordinate is in the superior-inferior plane. The z coordinate is frequently interpolated as the slice thickness of the planning CT scan can vary. The isodose contours are produced by linear interpolation between points, therefore, the separation of these points needs to be small enough to allow sufficient dose detail at organ and PTV boundaries.


Generally, isodose curves are normalized to give 100% to a reference point (such as the ICRU 62 reference point). This makes both the assessment of PTV coverage and the dose uniformity across the volume easier. The aim is usually to cover the PTV with the 95% isodose curve while ensuring the maximum dose within the volume does not exceed 107% [1, 2].


The dose distribution can be displayed on multiple views of reconstructed sections (e.g. transverse, sagittal and coronal sections on a single screen), and also on a 3D reconstruction of anatomy with isodose surfaces displayed.


Useful features are:



Mar 7, 2016 | Posted by in GENERAL RADIOLOGY | Comments Off on Principles and practice of radiation treatment planning

Full access? Get Clinical Tree

Get Clinical Tree app for offline access