British Journal of Radiology (2006) 79, S36-S49
© 2006 British Institute of Radiology
doi: 10.1259/bjr/17748030
Localization: conventional and CT simulation
G R Baker, BSc, MSc, MIPEM
Kent Oncology Centre, Maidstone Hospital, Maidstone, Kent ME16 9QQ, UK
 |
Abstract
|
|---|
Recent developments in imaging and computer power have led to the ability to acquire large three dimensional data sets for target localization and complex treatment planning for radiation therapy. Conventional simulation implies the use of a machine capable of the same mechanical movements as treatment units. Images obtained from these machines are essentially two dimensional with the facility to acquire a limited number of axial slices to provide patient contours and tissue density information. The recent implementation of cone beam imaging on simulators has transformed them into three dimensional imaging devices able to produce the data required for complex treatment planning. The introduction of computed axial tomography (CT) in the 1970s was a step-change in imaging and its potential use in radiotherapy was quickly realised. However, it remained a predominantly diagnostic tool until modifications were introduced to meet the needs of radiotherapy and software was developed to perform the simulation function. The comparability of conventional and virtual simulation has been the subject of a number of studies at different disease sites. The development of different cross sectional imaging modalities such as MRI and positron emission tomography has provided additional information that can be incorporated into the simulation software by image fusion and has been shown to aid in the delineation of tumours. Challenges still remain, particularly in localizing moving structures. Fast multislice scanning protocols freeze patient and organ motion in time and space, which may lead to inaccuracy in both target delineation and the choice of margins in three dimensions. Breath holding and gated respiration techniques have been demonstrated to produce four-dimensional data sets that can be used to reduce margins or to minimize dose to normal tissue or organs at risk. Image guided radiotherapy is being developed to address the interfraction movement of both target volumes and critical normal structures. Whichever method of localization and simulation is adopted, the role of quality control is important for the overall accuracy of the patient's treatment and must be adapted to reflect the networked nature of the process.
|
Introduction
|
|---|
The development of the delivery of radiation therapy is closely related to the accuracy with which the target tumour can be located with respect to surrounding anatomical structures. In recent years, the increase in computing power and the development of refined computer graphics have resulted in the ability to perform complex treatment planning in three dimensions and to manipulate images in real time. Early simulators were machines capable of the same mechanical movements as treatment units and were used to confirm treatment set up rather than for localization [1, 2]. Simulators that were developed commercially in the 1960s had the addition of fluoroscopy that was used to set the isocentre with the aid of remotely controlled movements of the couch. Field portals adequate to encompass the target volume to be treated could also be set by remote adjustments to the field defining wires. The introduction of computed axial tomography (CT) scanning in the 1970s was a step change in the ability to define tumours in relation to normal anatomy, and over the ensuing years has been widely adopted in tumour localization. Today it may be used in conjunction with complex graphics software as a virtual simulator. However, the conventional simulator still retains its place in many radiotherapy departments for localization of some tumour sites, either as a result of lack of sufficient access to a CT scanner or for relatively simple techniques not requiring the production of a dose plan. The conventional simulator is also frequently used to verify the more complex treatment plans, producing an image corresponding to a beam's eye view (BEV) from the treatment planning system (TPS) or by verifying the isocentre location from orthogonal films.
|
Brief history
|
|---|
Mould [3] describes the development of simulation, from the use of diagnostic radiographs and skin marks in the 1950s to the introduction of virtual simulation in the 1980s. In 1973, Hounsfield and Ambrose [4, 5] published their work on computerized transverse axial tomography and the potential uses of CT in radiotherapy were quickly recognized [6]. However, access to a CT scanner was often very limited, and in many cases the scanner was not even in the same hospital as the treatment facilities. In addition, a CT scanner was principally a diagnostic tool with limitations for treatment planning imposed by the small aperture and the design of the couch, which frequently prevented the patient from being scanned in the treatment position. Harrison and Farmer [7] recognized the usefulness of being able to acquire a cross-sectional image of the patient in the treatment position using a simulator as a CT scanner and went on to describe the implementation of their idea using a fluorescent screen and an Isocon camera [8]. A number of other adaptations of the simulator to produce cross-sectional images were also proposed at this time [9–12]. This functionality was called Sim-CT and became standard on simulators in the 1990s, but the system had its limitations:
- The heat capacity of the X-ray tube generally meant that only a few slices could be scanned;
- The time taken to scan was limited to approximately one revolution per minute, which introduced motion artefacts resulting in images that were of a poorer quality than those produced on a diagnostic scanner;
- The uncertainty in the Hounsfield units (HU), which depends on the field of view and the phantom/patient size, a result of the beam hardening in the unfiltered X-ray beam from the simulator CT. However, the uncertainty in HU is translated into dose variation not exceeding 3% for photon beams in the range 6–18 MV [13];
- The relatively high dose to the patient which was shown to be approximately 10 times that delivered with a diagnostic scanner under similar conditions [14].
In spite of its limitations, the Sim-CT was a useful tool for planning in a department with limited access to a diagnostic scanner. It was a more accurate way of producing a patient outline than manual methods using callipers and flexicurves and enabled CT numbers to be converted to relative electron densities for tissue inhomogeneity corrections to be applied to a single CT slice in dose calculations. The dose distributions and monitor unit calculations showed good agreement with those obtained with diagnostic scan data [14].
In 1998, Cho et al [15] described the application of digital technology to a radiotherapy simulator in which the imaging system was replaced by a digital spot imager (DSI). The DSI consisted of an image intensifier, digital image processing, display and data transfer facilities. The images were stored during acquisition for later archiving or transfer to workstations. Simulator manufacturers now offer digital capabilities on their machines and conventional image intensifiers have been replaced by flat panel amorphous silicon (aSi) detectors. Their longevity in this application has to be proved and it is possible that the need for regular replacement may have significant revenue consequences. The most recent simulators include anatomical protocol selection, automatic correction for image distortion, last image hold, multileaf collimator (MLC) verification, a variety of image viewing and manipulation tools with annotation, image printing to film or paper, Digital Image Communications in Medicine (DICOM) export to TPS, electronic portal imaging device (EPID), record and verify, and patient management systems. The image manipulation tools enable adjustments to be made to field parameters and image quality on the last-held image, which reduces the screening time and hence patient dose compared with non-digital systems. A wide aperture (typically 90 cm) CT option is available. However, because of the restriction on gantry rotation speed, acquisition times are still slow and reconstruction time does not match that of a diagnostic scanner. In an attempt to overcome this, volume or cone beam CT (CBCT) has been developed. A number of authors describe cone beam reconstructions, based on Feldkamp's original back projection algorithm [16], for the acquisition of volumetric data [17–19].
When first proposed, the size of the detector was a severe limitation on the reconstruction volume and, although promising results were obtained, its use in treatment planning was not realised until aSi flat panel detectors of a reasonable size became available. Commercial systems are now available. For example, the Acuity (Varian, Palo Alto, CA) with cone beam option gives a cone of 17 cm at the isocentre but with added penumbra of 1.9 cm at either end regardless of the scan length. It is therefore not appropriate to acquire a single narrow slice. A single slice takes 45 s and 675 images are acquired per rotation. Early reports (private communications, A Vinall, K Venables, 2005) suggest that the geometric performance and image quality are adequate for radiotherapy planning purposes although the images are not of diagnostic quality. The rotation time of 45 s does, however, result in significant movement artefacts. Figure 1a
shows the streaking that results from the movement of bowel gas during the acquisition of a CBCT scan compared with a CT planning scan. Figure 2 shows similar streaking in the soft tissue around lungs in a CBCT taken during normal breathing. As with the single slice option on the simulator, there seem to be problems with the HU values both in accuracy compared with the calibration and reproducibility on a day-to-day basis. Slice thicknesses of 1–5 mm are available. Reconstruction times vary with the slice thickness and are in the order of 90 s. There is no standard way of quoting doses for these scans. Computed tomography dose index (CTDIw) is a measure of the dose from a CT scan, weighted between the centre and the surface to give an average value across the section. A CTDIw/810 mAs value of 15 mGy has been measured for a 10 cm scan length collimated to 13.8 cm (15 pulse s–1, pulse length 15 ms, 80 mA, 125 kV, 45 s rotation). Setting the scan length to 1 cm in clinical mode gave 54 mGy/810 mAs with the same exposure factors. This compares with the national reference dose of 20 mGy for a multislice scanner [20].
View larger version (105K):
[in this window]
[in a new window]
|
Figure 1. (a) Movement artefacts on an axial slice of a CBCT scan as a result of movement of bowel gas. (b) An axial slice from a planning CT of the pelvis for comparison. (Courtesy of Varian Medical Systems, Palo Alto, CA and Memorial Sloan-Kettering Cancer Centre).
|
|
View larger version (125K):
[in this window]
[in a new window]
|
Figure 2. Movement artefacts on an axial slice from a CBCT acquired during normal breathing.(Courtesy of Varian Medical Systems and Hirslanden Klinik, Aarau).
|
|
|
CT simulation
|
|---|
The alternative to using the simulator and CBCT to acquire a volume data set of the patient in the treatment position was to modify CT scanners to meet the needs of radiotherapy and add software to perform the simulation function.
With the rapid development of computer technology, enabling fast reconstruction of images in three dimensions, the true value of the enormous quantity of data acquired by a CT scanner and its use in radiotherapy planning was recognized.
The development of the concept of the beam's eye view (BEV) into the transmission image from CT scans that would result from any beam orientation paved the way to producing images from CT data that correspond to conventional simulator films [21–23]. These digitally reconstructed radiographs (DRRs) could be overlaid with the outlines of anatomic structures, field shapes and cross wires, and hence could display images similar to simulator radiographs. However, the spatial resolution of DRRs is limited by the voxel size of the CT scans and cannot match that of a simulator radiograph taken with a small focal spot and a short exposure. Even in the early implementation of this process the reconstruction time of the DRRs was reasonable, being in the region of 10 s for a 50 slice study. However, studies were limited by the specification of the CT scanner. The acquisition of a single slice might take 2–3 s with a delay between scans required for repositioning of the scanner and tubes with low heat capacity needed cooling time during the scan [24].
Early critical analysis of the CT simulation process highlighted the areas for improvement [25]. These included the limitations imposed on both treatment technique and the size of the patient by the aperture of the scanner (normally 70 cm), the time required for CT data acquisition and transfer from the scanner to the planning system, time required for outlining and contouring target volume and critical structures and the inconsistent accuracy of portal marking on the patient's skin. Complete field ports were marked on the patient's skin in most cases and novel devices for doing this constituted an important part of the virtual simulation process reported. [26, 27]. These drawbacks have now largely been overcome.
Multislice helical scanning, with high heat capacity CT tubes, has reduced the time required to acquire a CT data set of 100 slices to a matter of seconds. Wide bore scanners have removed most of the constraints of patient size and technique. Increased computing capacity and speed allows for real time reconstruction of the slice images at the scanner and real time manipulation of images in the virtual simulation software. In addition, the DICOM protocol facilitates fast transfer of image data between systems.
|
Current practice
|
|---|
Conformal radiotherapy (CRT) is now accepted best practice for a number of treatment sites, having the advantages of sparing normal tissue and providing the opportunity for dose escalation. Intensity-modulated radiotherapy (IMRT) is the ultimate expression of this, but successful implementation of CRT and IMRT cannot be achieved without three-dimensional information on the location and extent of the target volume and the position of adjacent organs at risk (OAR). The three-dimensionality of virtual simulation is essential to visualize the coverage of the target volume and the avoidance of OARs in the highly complex treatment plans required for CRT and IMRT. For some sites, such as the lung where the relative position of the target and OARs varies with time, this fourth dimension needs to be taken into account.
Sherouse et al [28] introduced the term virtual simulation in 1987 to describe the process of using computer aided design and digitally reconstructed radiographs to replace the process of physical simulation. The process of virtual simulation has been described in detail by Aird and Conway [29] who also gave examples of its application to a number of different sites.
|
The specification of a CT simulator
|
|---|
The fundamental requirements of a CT simulator are a CT scanner with a flat couch, positioning lasers and virtual simulation software.
CT scanner
Advances in the design and capabilities of CT scanners have modified the specifications given by Aird and Conway [29]. Multislice scanners enable very fast scanning times, even for the large studies, with narrow slice thicknesses required for the production of good DRRs. High heat capacity anodes are required for the large datasets that are frequently required for treatment planning applications. One manufacturer (Siemens Medical, Erlangen, Germany) has introduced a new design of directly cooled anode that should eliminate delays due to anode heating and enable fast acquisition of scans with the large number of narrow slices required for good DRRs.
Three manufacturers now produce wide aperture (85 cm) scanners designed for radiotherapy applications. In two, the scanned field of view (SFOV) is 60 cm with an extended reconstructed FOV of 85 cm. It should be noted that in the extended reconstructed FOV the HU numbers may not be consistent with the SFOV. In reality, it is unlikely that the uncertainty in HU translates into a dose discrepancy of more that 1–2% in the target. The third manufacturer claims a true SFOV of 85 cm.
Positioning lasers
A system of three lasers for the accurate positioning and alignment of the patient is required. The lateral lasers may be wall or frame mounted, and may be either fixed or move in a vertical plane. The sagittal laser must be able to move laterally to account for lack of lateral movement on the CT couch. These lasers move under computer control to define the isocentre for the treatment plan in terms of shifts from the reference marks.
Virtual simulation software
The virtual simulation software may either be part of a treatment planning system or may be a stand-alone system. If the latter is chosen, it is essential that connectivity is easily established with the treatment planning system for dose calculation. Since the introduction of DICOM-RT this connectivity is more readily achievable, but the user must be aware that not all manufacturers interpret the standard in the same way and there are frequently hidden licensing issues associated with the connectivity. Essential features of virtual simulation software include automatic contouring of body outlines and semi-automatic contouring of other structures and critical organs such as spinal cord, kidneys and lungs. Particular attention should be paid to treatment of bifurcating structures. Contouring tools should be simple to use and interpolation between non-adjacent slices, with correction as necessary, should be provided to speed the contouring process. The ability to contour in three dimensions, i.e. in sagittal and coronal as well as axial sections, is particularly helpful. Figure 3 shows how three single contours in orthogonal planes produce a three dimensional structure. This functionality can considerably reduce the time taken to outline structures. The shape of the contours can be modified on any slice as necessary. Similar interpolation tools should be available for target volume delineation and true three-dimensional volume margin growth with different margin widths in different directions. Three-dimensional display systems are an essential feature of any virtual simulation software. It should be possible to display axial, sagittal and coronal sections on the same screen and relate each section to the others, and to visualize the DRRs in the same window. An Observer's Eye View, with the patient on the couch and the floor and gantry angles depicted, is an aid to patient setup, as is a light-field displayed on the patient's skin related to skin marks or tattoos. Anti-collision software avoids planning a treatment field which it is physically impossible to reproduce in the treatment room. There are many different ways of rendering the target volume and OARs, but they should be unambiguous and should be rendered in three-dimensions so that coverage can be checked from all aspects. Optimization of MLC leaf positions and collimator angle should be available but adjustable by the planner. For treatment planning where a full dose distribution will not be calculated, a particularly useful feature is the calculation of the equivalent square of an irregular field, the parameter required for simple dose calculations. Increasingly, oncologists are using a number of other imaging modalities such as MRI (see Khoo and Joon in this issue) and positron emission tomography (PET) (see Jarritt et al in this issue) to help in determining target volumes. Most virtual simulation packages include an image fusion function enabling registration of two datasets of the same or different modalities, CT/CT, CT/MRI, CT/PET. Image registration and fusion may be achieved in a number of different ways, both manual and automatic (see Kessler in this issue). Irrespective of the algorithm, there is a variety of display modes to assist in performing and viewing the fusion, some of which are shown in Figure 4. Figure 4a shows the two data sets (MR and CT) fused with information from both sets displayed in the same window. The image can be "faded" between the two showing 100% of the primary data set (CT in this case) through to 100% of the secondary data set (MRI in this example). Figure 4b shows a split screen, with two quadrants displaying the CT data and two showing the MRI data. The point of intersection can be moved around the image to display the intersection at any position on the image. This will assist in delineating the structures using information from both data sets. Figure 4c shows a split screen with the secondary data set fused with the primary in the centre of the image and the primary image on either side. Contours outlining the target or OARs can be drawn on either data set or on the fused images in any of these display modes. These three screens show the fused images in the top three windows and the secondary data set in the lower windows. Figure 4d shows the region of discrepancy between the two fused data sets, in this case two CT studies, as areas of enhancement on the image. Improved localization of a brain tumour when CT and MRI data sets are fused compared with localization on CT alone for treatment planning is demonstrated in Figure 5.
View larger version (84K):
[in this window]
[in a new window]
|
Figure 3. A single contour in axial sagittal and coronal planes defines a three dimensional target in Prosoma.(Courtesy of Oncology Systems Limited, Shrewsbury, UK and Medcom, Darmstadt, Germany).
|
|
View larger version (132K):
[in this window]
[in a new window]
|
Figure 4. (a) Fusion of MRI and CT data sets, fused images in the top windows and MRI images below. (b) A split screen showing fusion between CT and MRI data sets in quadrants. (c) An alternative split screen representation of fusion between CT and MRI data sets. (d) Areas of mismatch between two CT data sets displayed as image enhancement. (Courtesy of OSL and Medcom).
|
|
View larger version (128K):
[in this window]
[in a new window]
|
Figure 5. Improved localization of brain tumour using fused CT and MRI data sets.(Courtesy of OSL and Medcom).
|
|
|
Comparison of conventional and virtual simulation
|
|---|
Conventional and virtual simulation approach the task of localizing the target volume for treatment planning in very different ways, which may result in significantly different treatments. Realisation of the steps performed to provide the data to a treatment planning system is compared for the two modalities in Table 1.
In comparing the two methods of simulation, the first question that arises is whether the two are comparable in terms of accuracy of the treatment set up. There are a number of studies addressing this question for different treatment sites. Bollet et al [30] showed that in a series of 20 patients who were CT scanned and had conventional simulation, the precision of set up evaluations using DRRs was similar to that using simulator films in conformal prostate treatments. They also considered whether errors were introduced at the simulation stage and found a statistically significant systematic error between DRRs and simulator, in both the craniocaudal direction and the anteroposterior direction. In another study of prostate patients Valicenti et al [31] showed that there was no statistically significant reduction in treatment setup error if patients have physical simulation following virtual simulation and concluded that physical simulation may be omitted if virtual simulation is available. In a study of 86 patients undergoing palliative radiotherapy for lung cancer using parallel opposed fields, McJury et al [32] found that setup errors were comparable between the group planned by virtual simulation and that planned using conventional simulation. Similar results are reported at different treatment sites [33–35]. In a detailed study of setup errors in 39 patients undergoing CT planned radiotherapy for lung cancer, de Boer et al [36] concluded that the setup errors introduced at simulation, which become systematic errors if the simulator film is used as the reference image, were comparable with systematic errors at the treatment unit. Hence, omission of the simulation stage would reduce systematic errors on treatment. This conclusion supported a similar result for prostate patients [37].
In comparing the two methods of simulation, studies have shown that the target volumes and field sizes are smaller for virtual than conventional simulation in lung cancer with the associated reduction in irradiation of normal tissue [32, 38]. Smaller field sizes have also been reported for maxillary cancer with a corresponding reduction in long-term side effects [39].
One of the perceived advantages of virtual simulation is the improved coverage of the gross tumour volume (GTV) and the avoidance of OARs as a result of better visualization of soft tissue structures on a CT scan compared with a simulator image, particularly if shielded by bone. This is aided by software functions that remove overlying structures, giving better definition of the region of interest. A study comparing conventional and virtual simulation in the treatment planning of malignant lymphoma showed incomplete coverage of the spleen and spleen hilus in 5 of 15 and 6 of 15 patients, respectively, on conventional simulation and incomplete coverage of the right and left hilus in 4 of 15 and 1 of 15 patients, respectively. In addition, the left kidney was inadequately shielded in 6 of 15 of the conventionally planned patients [40]. Similar improvements in target coverage and OAR avoidance are reported for other anatomical sites [41–44].
Improved visualization of soft tissue structures may bring to light hitherto unsuspected pathology. Mehta and Goffinet [45] reported 17 unsuspected abnormalities in 153 scans (11%) obtained for treatment planning for patients referred for irradiation of the breast or chest wall. Of these, four represented disease that altered the treatment plan.
|
Working practices
|
|---|
The introduction of CT simulation has had a considerable impact on working practices in radiotherapy departments.
Oncologist attendance
The most notable change is that an oncologist is not required to be present during the scanning process. This releases the planning schedule from reliance on the oncologist's timetable, and the oncologists are free to undertake volume definition at a time convenient to them.
Time
A number of centres have reported on the different time allocation between conventional and virtual simulation [25, 28, 35]. Experience at the Kent Oncology Centre has shown that there is little difference in the total time needed for localization between the two modalities for the planning radiographers. With three radiographers in the scanning suite, 20 min appointments are adequate for most patients. Patients undergoing planning for breast radiotherapy are usually allocated 30 min because of the complex immobilization and positioning required with a narrow aperture scanner. These times are shorter than conventional simulation (30 min and 45 min, respectively), but more time is spent in manipulating the acquired data in the virtual simulation software. This includes the registering of reference marks and the production of DRRs for palliative patients, and outlining of target volumes and OARs for radical patients. Reduced simulation time for the patient leads to improved patient compliance, resulting in fewer problems from movement during scanning.
Reference marks
In conventional simulation, using fluoroscopy for localization of the target volume, the isocentre can usually be established and marked at the time of simulation. In CT simulation, a reference point is chosen at the scanning session and the eventual isocentre is defined by movements of the couch from the reference point. If virtual simulation of palliative patients is undertaken with the patient remaining on the couch, the isocentre can be marked immediately from the couch movements indicated.
Verification
It has already been shown that to verify a plan on a conventional simulator after virtual simulation is not only unnecessary, but it could also be a source of systematic errors. However, treatment verification is still required and is of greater importance because of the use of reference marks. Verification takes place on the treatment unit with the electronic portal imaging system. The portal images acquired are then compared with the DRRs produced by the TPS or the virtual simulation software. For complex plans, this may require an extra treatment slot to allow time for the detailed comparison of portal images and DRRs before treatment.
|
Advantages and disadvantages of conventional and CT simulation
|
|---|
The advantages and disadvantages of conventional and CT simulation are summarized in Tables 2 and 3.
The availability of a three-dimensional dataset for all patients has some unexpected benefits. The increased information available may demonstrate previously unsuspected disease that may influence patient management. In palliative patients the extent of bone destruction from osteolytic lesions is easier to visualize on a CT scan than on a simulator film (Figure 6) and the use of software functions to remove overlying structures and display images optimized for different tissue types enables quicker localization of the disease. In breast planning, cardiac and lung volumes are more clearly demonstrated and therefore the fields can be adjusted or shielding employed accordingly.
One disadvantage of CT simulation is the increased patient dose. Doses for CT scanners are quoted as CTDIw with values in the region of 20 mGy. This dose is delivered to regions of normal healthy tissue as well as the tumour volume. Manufacturers of CT scanners provide various methods to reduce the total dose to the patient, taking account of the different dimensions of the patient at different levels and modulating the exposure in response to the detector measurements.
Some challenges still remain. Respiratory motion can affect the position of lung tumours and their relationship to OARs. Fast scanning protocols freeze patient and organ motion giving a snapshot view in time and space which may lead to inaccuracy in target delineation and choice of margins in three dimensions. Imaging techniques to overcome this drawback are an area of active investigation. The conventional method of treatment planning for lung tumours is to use fluoroscopic imaging to determine the maximum migration of the tumour during respiration and adopt large margins around the CTV to ensure that the target remains in the high dose region throughout the breathing cycle. A similar philosophy can be adopted by performing scans at deep inhale and deep exhale [46]. However, a number of other techniques have been suggested involving breath holding and respiratory gating techniques [47]. Deep inspiration breath hold (DIBH) increases the lung volume relative to normal breathing and hence the total volume of lung irradiated will be reduced using this technique [48]. In some patients, DIBH may displace the tumour away from OARs [49], which has the potential for dose escalation to the target for the same level of toxicity to OARs. Gated respiration techniques may either be active or passive. In active breathing control (ABC), the patient is prevented from breathing at a given part of the respiratory cycle during which the scan is performed and subsequent treatment takes place. By acquiring a number of scans at different parts of the breathing cycle, motion of the organ in three-dimensions can be demonstrated. Passive techniques allow the patient to breathe normally and a surrogate for the respiratory induced motion, such as the movement of the anterior chest wall, is monitored. Images obtained from CT scans are sorted according to respiratory phase to produce a 4D CT data set [50–52]. Breath hold and ABC techniques both require the co-operation of the patient and are therefore not appropriate for all patients. Some verbal or visual coaching helps to maintain regular breathing.
An alternative approach to the problem of organ motion is suggested by Murphy [53] who describes the real-time tracking of moving organs. Tracking respiratory motion is a complex procedure as it involves fast movement of organs relative to each other. For real-time tracking to be successful, the system must be able to locate the target, predict the motion to account for any time delays in repositioning the beam and adapt the treatment plan to allow for the change in relative positions of target and OARs. Although respiratory motion appears fairly regular, there are changes in amplitude and period from one cycle to the next which make prediction complicated. Murphy discusses two ways of predicting respiratory movement, by developing a mathematical model and by using an empirical algorithm that is based on measurements of previous breathing cycles. The technical challenges of fast response times to organ motion in continuous real time tracking are presented, but Murphy suggests that in the future it should be possible to treat lung tumours in some patients during free breathing, without needing to include movement margins in the treatment plan.
Respiratory correlation techniques developed to minimize motion artefacts in axial and helical scanning are not applicable to CBCT and different techniques have been developed for the CB application. Sonke et al [54] describe a method for sorting the projections into different phases of the breathing cycle to produce a 4D CBCT scan. Sillanpaa et al describe a method of acquiring megavoltage cone beam CT projection images at the same phase of breathing at all acquisition angles, giving a three-dimensional reconstruction at a single breathing phase [55]. It must be emphasised that gated respiration techniques must be employed at both the localization stage and during treatment.
|
Quality assurance
|
|---|
The accuracy of both conventional and CT simulation has a crucial effect on the overall accuracy of the patient's treatment. Whereas the accuracy of conventional simulation relies mainly on geometric features such as gantry and collimator angles and field defining wire positions, that of CT simulation depends on the image obtained by the scanner and the faithful transfer to the virtual simulation software. This connectivity should be part of any quality assurance (QA) programme.
A detailed description of quality control tests in conventional simulation and their recommended frequency is given by Tuohy [56].
Virtual simulation forms part of the network of the radiotherapy department, the end result of which is the treatment of the patient. The QA of this network should be seen as a process to which the various components of the hardware and software contribute. Guidance for the QA of a networked radiotherapy department is due to be published soon [57]. A QA programme should be established that reflects the importance of the contribution of each component of the system to the accuracy of the patient's treatment. Some components will be checked daily, such as the alignment of the lasers, the accuracy of positioning of any moving lasers and the HU accuracy for water. Others may be checked monthly, annually or after significant upgrades to the system. Special phantoms have been designed to assist with various aspects of QA [58, 59]. The Kent Oncology Centre has produced its own phantom that incorporates checks for a number of parameters in one scan study. These include spatial resolution, HU number, slice thickness, alignment and geometric accuracy.
Mutic et al [60] provide a comprehensive guide to the QA of CT simulators. They stress the need for audit and review of the process and flexibility in the programme as CT simulation evolves.
|
The future
|
|---|
The aim of radiotherapy is to deliver a tumoricidal dose of radiation to the clinical target volume (CTV) whilst sparing normal tissue and critical organs as far as possible. Localization is aimed at answering the question "where is the target?" The gross tumour volume (GTV) is neither a simple line nor an unchanging volume. It is an oncological concept and will vary according to the imaging technique or techniques used, any additional clinical data available and the judgement of the clinician. Each imaging modality displays different information about the GTV. Traditionally, delineation of the GTV has been associated with an anatomical abnormality that is imaged by plane radiography, CT or in some cases MRI. This gives structural, not functional information. However, molecular and physiological imaging techniques are now available which give an indication of the functional state of the tissues. This information can potentially be used in addition to CT and MRI to assist in defining clinically relevant targets more accurately [61]. Ling et al [62] proposed treating a biological target volume defined from anatomical, physiological and/or molecular images. For example, increased glycolysis is a function of a tumour and fluorine-18 fluorodeoxyglucose positron emission tomography (18FDG-PET) studies have been used as an addition to CT for planning patients with poorly defined non-small cell lung cancer (NSCLC) [63, 64], head and neck cancers [65] and malignant gliomas [66] (see Jarritt et al in this issue). Figure 7 shows the fused images from 18FDG-PET and CT acquired in a single session on a PET/CT scanner. The lesion in the left lung is clearly demonstrated in both modalities in this example. Other PET agents may be used to identify areas of hypoxia within a tumour that may benefit from higher doses of radiation such as can be delivered by IMRT. Similar inhomogeneous dose distributions may be applied to regions of the prostate demonstrating a high choline:citrate ratio, indicating a region of active tumour, as demonstrated on MR spectroscopy [67] (see Payne and Leach in this issue). Modalities such as functional MRI (fMRI) and single photon emission computed tomography (SPECT) may also be used to assist in GTV and OAR delineation. SPECT perfusion studies for NSCLC can be used in treatment planning to provide information on normal lung tissue and help to reduce the volume of normal lung irradiated [68].
View larger version (97K):
[in this window]
[in a new window]
|
Figure 7. Fusion of positron emission tomography(PET) and CT images from a CT/PET scanner to localize a left lung tumour.
|
|
Imaging techniques are continually evolving and as they are refined they will reveal more information about the disease to be treated. Collaboration between radiologists and oncologists will be essential if the information contained within these new images is to be maximized for the benefit of the patient.
No consideration of the future of radiation therapy would be complete without mention of image guided radiotherapy (IGRT). IGRT aims to address the interfraction movement of tumours and their relationship to OARs. Of the linear accelerator manufacturers, both Elekta (Crawley, Sussex, UK) and Varian (Palo Alto, CA) provide kilovoltage cone beam CT (CBCT) on the gantry and Siemens (Erlangen, Germany) have installed a CT scanner on rails in the treatment room (see Moore et al and Thieke et al, respectively, in this issue).
These imaging devices provide the ability to localize the tumour immediately prior to treatment and to reposition the patient to correct for interfraction variation in tumour position. Wong et al [69] describe the use of daily scans in the treatment room to reposition prostate patients for the final phase of their treatment. 46% required no isocentre adjustment in the anterior–posterior direction, but 44% required a shift of greater than 5 mm. In the superoinferior direction, 25% required a shift greater than 5 mm and in left-right direction 24% required a shift greater than 5 mm. The shifts were associated with significant changes in the dosimetry. Other authors describe the implementation of CBCT for IGRT [54, 70, 71].
IGRT is a rapidly evolving field and will undoubtedly have implications for treatment planning.
|
Conclusion
|
|---|
Both conventional and virtual simulation have developed in line with the changes in imaging techniques over recent years. The anticipated advantages of virtual simulation have been realised to a great extent and have changed the work flow in treatment planning. The availability of wide bore scanners enables most treatment techniques to be imaged. Fast computer graphics that have reduced image reconstruction times enable the acquisition of large data sets that can be manipulated for respiratory correlated techniques. The rapid development of biological imaging holds the prospect of multimodality localization, which is already being realised for some disease sites such as lung and prostate. The addition of cone beam CT to conventional simulators may add flexibility to departments with both a scanner and a simulator. However localization is achieved, it must be considered as part of the overall process that leads to treatment. The accuracy of the data acquisition and transfer is vital to this process and a comprehensive QA programme is essential.
I would like to thank Dr Ruth Beddows for the design of the Kent Oncology Centre phantom, Ms Alison Vinall and Ms Karen Venables for reports on cone beam CT, David Hill for assistance with image processing and colleagues for discussions during the preparation of this manuscript.
Received for publication June 30, 2005.
Revision received February 28, 2006.
Accepted for publication March 1, 2006.
|
References
|
|---|
- Farmer F, Fowler J, Haggith J. Megavoltage treatment planning and the use of xerography. Br J Radiol 1963;36:426–35.[Abstract/Free Full Text]
- Greene D, Nelson K, Gibb R. The use of a linear accelerator in radiotherapy. Br J Radiol 1964;37:394–7.[Abstract/Free Full Text]
- Mould R. From conventional to virtual simulation: a historical vignette. In: Mould RF, editor. Progress in CT-3D simulation. Bochum, Germany: Medintec, 2003:156–62
- Hounsfield G. Computerised transverse axial scanning (tomography) Part 1. Description of the system. Br J Radiol 1973;46:1016–22.[Abstract/Free Full Text]
- Ambrose J. Computerised transverse axial scanning (tomography) Part 2. Clinical application. Br J Radiol 1973;46:1023–47.[Abstract/Free Full Text]
- Dobbs H, Parker R. The respective roles of the simulator & CT in RT planning: a review. Clin Radiol 1984;35:433–9.[CrossRef][Medline]
- Harrison R, Farmer F. Possible application of a radiotherapy simulator for imaging of body cross sections. Br J Radiol 1976;49:813[Abstract/Free Full Text]
- Harrison R, Farmer F. The determination of anatomical cross sections using a radiotherapy simulator. Br J Radiol 1978;51:448–53.[Abstract/Free Full Text]
- Webb S, Lillicrap S, Steere H, Speller R. Application of a radiotherapy simulator for imaging of body cross sections. Br J Radiol 1977;50:150–3.[Abstract/Free Full Text]
- Kotre C, Harrison R, Ross W. A simulator based CT system for radiotherapy treatment planning. Br J Radiol 1984;57:631–5.[Abstract/Free Full Text]
- Redpath A, Wright D. The use of an image processing system in radiotherapy simulation. Br J Radiol 1985;58:1081–9.[Abstract/Free Full Text]
- Arnot R, Willets R, Batten J, Orr J. Investigations using an x-ray image intensifier and a TV camera for imaging transverse sections in humans. Br J Radiol 1984;57:47–55.[Abstract/Free Full Text]
- Chu J, Ni B, Kriz R, Saxena A. Applications of simulator computed tomography number for photon dose calculations during radiotherapy treatment planning. Radiother Oncol 2000;55:65–73.[CrossRef][Medline]
- Verellen D, Vinh-Hung V, Bijdekerke P, Nijs F, Linthout N, Bel A, et al. Characteristics and clinical application of a treatment simulator with CT-option. Radiother Oncol 1999;50:355–66.[CrossRef][Medline]
- Cho P, Lindsley K, Douglas J, Stelzer K, Griffin T. Digital radiotherapy simulator. Comput Med Imag Graph 1998;22:1–7.[CrossRef][Medline]
- Feldkamp L, Davis L, Kress J. Practical cone beam algorithm. J Opt Soc Am 1984;1:612–9.[CrossRef]
- Cho P, Griffin T. Single scan volume CT with a radiotherapy simulator. Med Phys 1993;20:1292
- Mosleh-Shirazi M, Evans P, Swindell P, Webb S, Partridge M. A cone beam megavoltage CT scanner for treatment verification in conformal radiotherapy. Radiother Oncol 1998;48:319–28.[CrossRef][Medline]
- Midgley S, Millar R, Dudson J. A feasibility study for megavoltage cone beam CT using a commercial EPID. Phys Med Biol 1998;43:155–69.[CrossRef][Medline]
- Shrimpton P, Hillier M, Lewis M, Dunn M. Doses from computed tomography (CT) examinations in the UK – 2003 Review. Didcot, UK: National Radiological Protection Board, 2005
- McShan D, Fraass B, Lichter A. Full integration of the beam's eye view concept into computerised treatment planning. Int J Radiat Oncol Biol Phys 1990;18:1485–94.[Medline]
- Sherouse G, Novins K, Chaney E. Computation of digitally reconstructed radiographs for use in radiotherapy treatment design. Int J Radiat Oncol Biol Phys 1990;18:651–8.[Medline]
- Goitein M, Abrams M, Rowell D, Pollari H, Wiles J. Multi-dimensional treatment planning II. Beam's eye view back projection and projection through CT sections. Int J Radiat Oncol Biol Phys 1983;9:789–97.[Medline]
- Galvin J, Sims C, Dominiak G, Cooper J. The use of DRRs for three dimensional treatment planning and CT simulation. Int J Radiat Oncol Biol Phys 1995;31:935–42.[CrossRef][Medline]
- Perez C, Purdy JA, Harms W, Gerber R, Matthews J, Grigsby P, et al. Design of a fully integrated 3 dimensional computed tomography simulator and prelimary clinical evaluation. Int J Radiat Oncol Biol Phys 1994;30:887–97.[Medline]
- Nishidai T, Nagata Y, Takahashi M, Abe M, Yamaoka N, Ishihara H, et al. CT simulator: a new 3-D planning and simulation system for radiotherapy. 1. Description of system. Int J Radiat Oncol Biol Phys 1990;18:499–504.[Medline]
- Ragan D, Forman J, He T, Mesina C. Clinical results of computerised tomography based simulation with laser patient marking. Int J Radiat Oncol Biol Phys 1996;34:691–5.[CrossRef][Medline]
- Sherouse G, Mosher C, Novins K, Rosenman J, Chaney E. Virtual simulation: concept and implementation. In: Bruinvis IAD, Van der Geissen PH, Van Kleffens HJ, Wittkamper FW, editors. Proceedings of the Ninth International Conference on the use of computers in Radiation Therapy, Amsterdam, The Netherlands. North Holland Publishing Co. 1987:433–6
- Aird E, Conway J. CT Simulation for radiotherapy treatment planning. Br J Radiol 2002;75:937–49.[Abstract/Free Full Text]
- Bollet M, McNair H, Hansen V, Norman A, O'Doherty V, Taylor H, et al. Can digitally reconstructed radiographs replace simulation films in prostate cancer conformal radiotherapy? Int J Radiat Oncol Biol Phys 2003;57:1122–30.[CrossRef][Medline]
- Valicenti R, Waterman F, Corn B, Curran W. A prospective, randomised study addressing the need for physical simulation following virtual simulation. Int J Radiat Oncol Biol Phys 1997;39:1131–5.[CrossRef][Medline]
- McJury M, Fisher P, Pledge S, Brown G, Anthony C, Hatton M, et al. The impact of virtual simulation in palliative radiotherapy for non-small cell lung cancer. Radiother Oncol 2001;59:311–8.[CrossRef][Medline]
- Horst E, Schuck A, Moustakis C, Shaefer V, Micke O, Kronholz H, et al. CT simulation in nodal positive breast cancer. Strahlenth und Onkol 2001;177:511–6.[CrossRef]
- Schiebe M, Hoffman W. CT based virtual simulation using the advantage Sim 4.1 system. Description of reliability and accuracy. Strahlenth und Onkol 2000;176:377–80.[CrossRef]
- Buchali A, Geismar D, Hinkelbein M, Shlenger L, Zinner K, Budach V. Virtual simulation in patients with breast cancer. Radiother Oncol 2001;59:267–72.[CrossRef][Medline]
- de Boer HJ, Van Sorensen de Koste J, Senan S, Vissier AG, Heijmen B. Analysis and reduction of 3D systematic and random setup errors during the simulation and treatment of lung cancer patients with CT based external beam radiotherapy dose planning. Int J Radiat Oncol Biol Phys 2001;49:857–68.[CrossRef][Medline]
- Bel A, Bartelink H, Vijlbrief R, Lebesque J. Transfer errors of planning CT to simulator: a possible source of setup inaccuracies? Radiother Oncol 1994;31:176–80.[CrossRef][Medline]
- Senan S, Van Sorensen de Koste J, de Boer J, et al. The use of CT simulation in digitally reconstructed radiographs (DRRs) in setup verification allows for smaller planning target volumes in lung cancer. Lung Cancer 2000;20 (Suppl. 1):162
- Nagata Y, Okajima K, Murata R, Mitsumori M, Mizowaki T, Tsutsui K, et al. Three-dimensional treatment planning for maxillary cancer using a CT simulator. Int J Radiat Oncol Biol Phys 1994;30:979–83.[Medline]
- Dinges S, Koswig S, Buchali A, Wurm R, Schlenger L, Bohmer D, et al. Comparison of conventional and virtual simulation for radiation treatment planning of malignant lymphoma. Strahlenth und Onkol 1998;174(Suppl. II):28–30.
- Nishioka T, Shirato H, Arimoto T, Kaneko M, Kitahara T, Oomori K, et al. Reduction of radiation induced xerostomia in nasopharyngeal carcinoma using CT simulation with laser patient marking and three field irradiation technique. Int J Radiat Oncol Biol Phys 1997;38:705–12.[CrossRef][Medline]
- Gripp S, Doeker R, Glag M, Vogelsang P, Bannach B, Doll T, et al. The role of CT simulation in whole brain irradiation. Int J Radiat Oncol Biol Phys 1999;45:1081–8.[CrossRef][Medline]
- Driver D, Drzymala M, Dobbs H, Faulkner S, Harris S. Virtual simulation in palliative lung radiotherapy. Clin Oncol 2004;16:461–6.[CrossRef]
- Chao M, Gibbs P, Tjandra J, Darben P, Lim-Joon D, Jones IT. Evaluation of the use of computed tomography versus conventional orthogonal x-ray simulation in the treatment of rectal cancer. Australas Radiol 2005;49:122[CrossRef][Medline]
- Mehta V, Goffinet D. Unsuspected abnormalities noted on CT treatment planning scans obtained for breast and chest wall irradiation. Int J Radiat Oncol Biol Phys 2001;49:723–5.[CrossRef][Medline]
- Shih H, Jiang S, Aljarrah K, Doppke K, Choi N. Internal target volume determined with expansion margins beyond composite gross tumor volume in three dimensional conformal radiotherapy for lung cancer. Int J Radiat Oncol Biol Phys 2004;60:613–22.[CrossRef][Medline]
- Mageras G, Yorke E Deep inspiration breath hold and respiratory gating strategies for reducing organ motion in radiation treatment. Semin Radiat Oncol 2004;14:65–75.[CrossRef][Medline]
- Rosenzweig K, Hanley J, Mah D. The deep inspiration breath hold technique in the treatment of inoperable non small cell lung cancer. Int J Radiat Oncol Biol Phys 2000;48:81–7.[CrossRef][Medline]
- Stromberg J, Sharpe M, Leonard H. Active breathing control (ABC) for Hodgkin's disease: reduction in normal tissue irradiation with deep inspiration and implications for treatment. Int J Radiat Oncol Biol Phys 2000;48:797–806.[CrossRef][Medline]
- Ford E, Mageras G, Yorke E, Ling C. Respiration correlated spiral CT: a method of measuring respiratory induced anatomic motion for radiation treatment planning. Med Phys 2003;30:88–97.[CrossRef][Medline]
- Pan T, Lee T, Rietzel E, Chen G. 4D-CT imaging of a volume influenced by respiratory motion on multi-slice CT. Med Phys 2004;31:333–40.[CrossRef][Medline]
- Vedam S, Keall P, Kini V, Mostafavi H, Shukla H, Mohan R. Acquiring a four dimensional computed tomography dataset using an external respiratory signal. Phys Med Biol 2003;48:45–62.[CrossRef][Medline]
- Murphy M. Tracking moving organs in real time. Semin Radiat Oncol 2004;14:91–100.[CrossRef][Medline]
- Sonke J, Zijp L, Remeijer P, van Herk M. Respiratory correlated cone beam CT. Med Phys 2005;32:1176–86.[CrossRef][Medline]
- Sillanpaa J, Chang J, Mageras G, Riem H, Ford E, Todor D. Developments in megavoltage cone beam CT with an amorphous silicon EPID: reduction of exposure and synchronisation with respiratory gating. Med Phys 2005;32:819–29.[CrossRef][Medline]
- Tuohy JB. Radiotherapy imaging devices: treatment simulators. In: Mayles WPM, Lake R, McKenzie A, Macaulay EM, Morgan HM, Jordan TJ, et al. Physics aspects of quality control in radiotherapy. York, UK: Institute of Physics and Engineering in Medicine, 1999:34–43
- Kirby M, Carpenter D, Lawrence G, Poynter A, Studdart P. Guidance for the commissioning and quality assurance of a networked radiotherapy department. A report produced by the Institute of Physics and Engineering in Medicine working party. In Press
- McGee K, Das I, Sims C. Evaluation of digitally reconstructed radiographs (DRRs) used for clinical radiotherapy: a phantom study. Med Phys 1995;22:1815–27.[CrossRef][Medline]
- Craig T, Brochu D, Van Dyke J. A quality assurance phantom for three dimensional radiation treatment planning. Int J Radiol Oncol Biol Phys 1999;44:955–66.[CrossRef][Medline]
- Mutic S, Palta JR, Butker E, Das I, Huq M, Loo LD, et al. Quality assurance for computed-tomography simulators and the computed-tomography-simulation process: Report of the AAPM Radiation Therapy Committee Task Group No.66. Med Phys 2003;30:2762–92.[CrossRef][Medline]
- Apisarnthanarax S, Chao K. Current imaging paradigms in radiation oncology. Radiat Res 2005;163:1–25.[CrossRef][Medline]
- Ling C, Humm J, Larsan S, Amols H, Fuks Z, Liebel S, et al. Towards multidimensional radiotherapy (MD-CRT) biological imaging and biological conformality. Int J Radiat Oncol Biol Phys 2000;47:551–60.[CrossRef][Medline]
- Mah K, Caldwell C, Ung Y, Danjoux C, Balogh J, Ganguli N, et al. The impact of 18FDG-PET on target and critical organs in CT based treatment planning of patients with poorly defined non small cell lung cancer: a prospective study. Int J Radiat Oncol Biol Phys 2002;52:339–50.[CrossRef][Medline]
- Bradley J, Thorstad W, Mutic S, Miller T, Dehdashti F, Siegel B, et al. Impact of 18FDG-PET on radiation therapy volume delineation in non-small-cell lung cancer. Int J Radiat Oncol Biol Phys 2004;59:78–86.[CrossRef][Medline]
- Heron D, Andrade R, Flickinger J, Johnson J, Agarwala S, Wu A, et al. Hybrid PET-CT simulation for radiation treatment planning in head and neck cancers. A brief technical report. Int J Radiat Oncol Biol Phys 2004;60:1419–24.[CrossRef][Medline]
- Gross M, Weber W, Feldman H, Bartenstein P, Schwaiger M, Molls M. The value of 18FDG-PET for the 3-D radiation treatment planning of malignant gliomas. Int J Radiat Oncol Biol Phys 1998;41:989–95.[CrossRef][Medline]
- Beasley M, Driver D, Dobbs H. Complications of radiotherapy: improving the therapeutic index. Cancer Imaging 2005;5:78–84.[Medline]
- Munley M, Marks L, Scarfone C, Sibley G, Patz E, Turkington T, et al. Multimodality nuclear medicine imaging in three dimensional radiation treatment planning for lung cancer: challenges and prospects. Lung Cancer 1999;23:105–14.[CrossRef][Medline]
- Wong J, Grimm L, Uematsu M, Oren R, Cheng C, Merrick S, et al. Image guided radiotherapy for prostate cancer by CT linear accelerator combination: prostate movements and dosimetric considerations. Int J Radiat Oncol Biol Phys 2005;61:561–9.[CrossRef][Medline]
- Sykes J, Amer A, Czajka J, Moore C. A feasibility study for image guided radiotherapy using low dose, high speed cone beam x-ray volumetric imaging. Radiother Oncol 2005;77:45–52.[CrossRef][Medline]
- Jaffray D, Siewerdsen J, Wong J, Martinez A. Flat panel cone beam computed tomography for image guided radiation therapy. Int J Radiat Oncol Biol Phys 2002;53:1337–49.[CrossRef][Medline]
Top
Advantages and disadvantages of...
Quality assurance
