Initial patient imaging with an optimised radiotherapy beam for portal imaging

Radiotherapy and Oncology 76 (2005) 63–71 www.thegreenjournal.com

Clinical physics

Initial patient imaging with an optimised radiotherapy beam for portal imaging Stella Flampouria,b, Helen A. McNairc, Ellen M. Donovana, Philip M. Evansa,*, Mike Partridgea, Frank Verhaegend, Christopher M. Nuttinge a

Joint Physics Department, Royal Marsden NHS Trust and Institute of Cancer Research, Sutton, UK, bRadiation Oncology Department, Massachusetts General Hospital and Harvard Medical School, Boston, USA, cRadiotherapy Department, Royal Marsden NHS Trust and Institute of Cancer Research, Sutton, UK, dMedical Physics Department, McGill University, Montreal, Que., Canada, eHead and Neck Unit, Royal Marsden NHS Trust and Institute of Cancer Research, London, UK

Abstract Background and purpose: To investigate the feasibility and the advantages of a portal-imaging mode on a medical accelerator, consisting of a thin low-Z bremsstrahlung target and a thin Gd2O2S/film detector, for patient imaging. Patients and methods: The international code of practice for high-energy photon dosimetry was used to calibrate dosimetry instruments for the imaging beam produced by 4.75 MeV electrons hitting a 6 mm thick aluminium target. Images of the head and neck of a humanoid phantom were taken with a mammography film system and the dose in the phantom was measured with TLDs calibrated for this beam. The first head and neck patient images are compared with conventional images (taken with the treatment beam on a film radiotherapy verification detector). Visibility of structures for six patients was evaluated. Results: Images of the head and neck of a humanoid phantom, taken with both imaging systems showed that the contrast increased dramatically for the new system while the dose required to form an image was less than 10K2 Gy. The patient images taken with the new and the conventional systems showed that air–tissue interfaces were better defined in the new system image. Anatomical structures, visible on both films, are clearer with the new system. Additionally, bony structures, such as vertebrae, were clearly visible only with the new system. The system under evaluation was significantly better for all features in lateral images and most features in anterior images. Conclusions: This pilot study of the new portal imaging system showed the image quality is significantly improved. q 2005 Elsevier Ireland Ltd. All rights reserved. Radiotherapy and Oncology 76 (2005) 63–71. Keywords: Portal imaging; Low-Z target

For a successful radiotherapy treatment, accurate positioning of the patient is required. The patient and target position with respect to the treatment geometry is usually verified with portal images—X-ray images formed with the treatment beam taken prior to or during the treatment. However, portal images obtained routinely in radiotherapy clinics for patient position verification suffer inherently from very low contrast. The photon energy range used in radiotherapy is appropriate for treatment but it is too high to produce high quality images. Solutions proposed to the low contrast problem included detector optimisation methods [2,17,21], radiographs [1] and cone beam computed tomography (CBCT) [8] with diagnostic energy X-ray tubes mounted in the gantry or incorporated in the head of the medical linear accelerator (linac) [3], modifications of the bremsstrahlung target in the linac to produce more low energy photons [4,5,11,14–16,20] and megavoltage

computed tomography (MVCT) [12,18,19]. The approach followed in this study was to modify the linac and the primary detector in order to obtain a megavoltage system that enables high contrast imaging. In previous studies [4,5,20], low Z materials, which reduce photoelectric absorption in the target and consequently achieve the maximum low energy photon yield were used as target materials. Galbraith [5] used thick Be and C targets because the fractional low energy photon yield of thick targets is increased compared to thin targets since energy degraded electrons still contribute significant bremsstrahlung in the forward direction if electron scattering in the target is minimised by using low Z materials. Conversely, Tsechanski et al. [20] used thin Al targets to reduce photoelectric absorption of low energy photons in the target. Flampouri et al. [4] showed that the low-energy photon fluence increases with decreasing target thickness,

0167-8140/$ - see front matter q 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.radonc.2005.04.006

64

Low-energy portal imaging

in favour of Tsechanski. In practice the target thickness (provided that it is thinner than the electron range and therefore it does not stop the low energy photons produced) has very little effect on image quality. For most of the low Z target studies [4,5,11,14,15,20], the primary detector used for imaging with the low Z target beams was a thin, high Z phosphor screen. Film was also used in combination with the detector for all the mentioned previous authors [4,5,11,14, 15,20]. These screens detect selectively the low energy photons and consequently are more suitable for imaging with the modified beams. For this study the primary detector was used in combination with film. Similar improvement in image quality is anticipated when the screen is used with a camera based EPID or an amorphous silicon panel. None of the previous studies was extended to patient imaging. In our previous work [4], a method based on Monte Carlo (MC) simulations and experimental measurements was developed for the optimisation of the accelerator target and the detector for portal imaging. The system was optimised to produce maximum image contrast. A combination of thin aluminium target and thin gadolinium oxysulfide screen/film produced the greatest improvement in image contrast compared to the conventional system (6 MV treatment beam/film sandwiched between two lead plates). The new system increased the contrast produced by 1 cm thick bone equivalent material in 5 and 15 cm water from 5 to 15% and from 3 to 10%, respectively. The contrast improvement decreased as the phantom thickness increased. The new system, which produced the dramatic contrast increase, was used in the current study for patient imaging. Head and neck sites are suitable for imaging with this kind of imaging system for two reasons: (1) thinner sections have been shown to benefit most from this technique and (2) high treatment accuracy is required because of many organs at risk and potential severe side effects. The purpose of this work is to investigate the applicability of the imaging method in the clinic. In the process, the dosimetric properties of the optimum beam for imaging were studied and instruments were calibrated to enable absolute dosimetry. An initial study with a humanoid phantom was carried out. Then the first patient images were taken with the new system and the usefulness of the method was shown. Finally, the visibility of structures in images of a group of six head and neck patients was evaluated.

the scattering foil mounting and was based on the results of the MC optimisation for imaging. It was installed in the machine during annual servicing. The experimental ‘imaging’ beam using this target, named Al6, was configured in the linac. Because the target was placed outside the vacuum, this beam starts as an electron beam, which passes from the Ni vacuum window and hits the external target. The flattening filters and any other structures that harden the photon beams are removed from the beam path. The only structures in the path are the monitor chamber, the mirror and the Mylar exit window. Because the target has an insufficient thickness to stop all the primary electrons, a 1 cm thick PMMA slab was placed in the accessory tray position at the exit of the SL25 head, to absorb the electrons contaminating the beam. The use of such a slab will introduce contamination from scatter within the PMMA. The peak of the energy spectrum from the Al target is at 300 keV. The use of a thicker target would be an alternative method of reducing this primary electron component without introducing a contribution from scatter within the PMMA slab. The primary electron beam is an electron beam with a nominal energy of 4 MeV, which is the lowest energy that this linac can produce. The lowest available electron energy is used because it gives the highest ratio of the low energy photon yield on the whole spectrum [14]. The cross-section of the photoelectric effect, which is necessary for good quality X-ray imaging, has a strong inverse dependence on the energy of the beam. The detector used with the Al6 beam was a commercial mammography system: Min-r 2 screen (Gd2O2S, 34 mg/cm2 back screen) and Min-r 2000 film (Kodak, USA). The imaging system comprised of Al6 and this detector is hereafter referred to as the ‘new system’. For comparison with the new system, images were also taken with the conventional system of the 6 MV beam (same linac) and a radiotherapy verification detector: Fuji EC-A cassette (125 mm front and 250 mm back lead plates) and Fuji new RX-U film (Fujifilm, Japan). The two detectors are more sensitive to different parts of the energy spectrum. MC simulations showed that the radiotherapy verification detector is more sensitive for energies around 300 keV while the mammography system has greater sensitivity for energies below 100 keV. Although the mammography screen is much thinner than the radiotherapy cassette the new system is more sensitive than the conventional because of the high sensitivity of the mammography film. Low dose (!5!10K3 Gy) images with the mammography system and the standard, 6 MV beam have previously been presented [4].

Methods and materials

Beam quality

Equipment

Absolute dosimetry is necessary before patient imaging can commence. The dose distribution in a phantom is typically measured with thermoluminescent dosimeters (TLD). The TLD calibration is performed by comparison with a calibrated ionisation chamber. Hence, the calibration procedure started by calibrating an ionisation chamber system for Al6. The choice of the calibration protocol to be followed was based on the dosimetric characteristics of the beam investigated by experiment and with Monte Carlo simulations [4].

The optimum bremsstrahlung target for imaging [4], 6 mm thick aluminium, was installed in an ELEKTA SL25 medical linear accelerator at the Royal Marsden Hospital. The specific linac is used only for photon therapy, thus all the hardware and software options for electron beams were unused. The target was installed in a vacant position for electron scattering foils, which is 14.4 cm below the vacuum window, placed on the secondary filter carrier carousel. The target was built according to the manufacturer’s designs for

S. Flampouri et al. / Radiotherapy and Oncology 76 (2005) 63–71

The choice of protocol for dosimetry instrument calibration is determined by the quality of the beam. The beam quality specifier for megavoltage beams used by the UK and the international codes of practice [7,13] is the quality index QI. This parameter is a measure of the effective attenuation coefficient of the beam in water beyond the depth of maximum dose. The measurements were performed under recommended conditions from the megavoltage protocol [13]. The QI places the Al6 beam in the megavoltage beam category. However, Al6 has a strong low energy component. Monte Carlo simulations [4] showed that the energy fluence below 200 keV is 6.9% of the total energy fluence for Al6 while it is only 0.2% for 6 MV. This difference, which makes Al6 more suitable for imaging, could be responsible for significant dosimetric differences, particularly at shallow depths. To investigate the extent of the influence of the kilovoltage component to the beam quality, measurements of depth dose data in the build up region (from 0.022 to 1.7 cm) were made with a Markus plane parallel ionisation chamber, for the experimental beam and the normal 6 MV beam. The measurements showed that at 2 mm depth the dose (after Rawlinson correction on the build up area) for the experimental beam was 15% higher than for the treatment beam. This is not important considering that the dose required to form an image with Al6 is 20% less than the dose required to take an image with the conventional system. Half-value layer (HVL) measurement was also performed [9]. The HVL measurements were performed with the same Farmer ionisation chamber placed at the isocentre. The field size was small (2!2 cm2) to minimise scatter. Sheets of pure Al were piled on the plastic electron filter placed at the exit of the linac head (54 cm from vacuum window) while the head of the accelerator was at 1808. The thickness of the Al absorber was varied from 0.05 to 5 cm. It was deduced that the boosted low energy component of Al6 does not change the quality of the beam. Also using the MC model of Al6 (presented in [4]), percentage depth dose curves were calculated in a water phantom when (1) the whole spectrum of Al6 is impinging on the phantom and (2) the low energy photons (below 150 keV) are filtered out before the phantom, showed that the low energy component of the beam affects the depth dose curve less than 1% from the surface to 1 cm depth. These calculations, the HVL, and the QI measurements, showed that for absolute dosimetry Al6 is treated like any other megavoltage radiotherapy beam.

Absolute dosimetry The procedure followed for dosimeter (Farmer ionisation chamber and Farmer electrometer) calibration and determination of absorbed dose was taken from the IAEA 2000 report. According to this protocol, the calibration factor of the dosimeter for a specific quality is calculated from the calibration factor of the same dosimeter for reference beam quality. The reference beam quality used was 60Co. TLDs (LiF:MgTi) were used to measure the dose distribution inside a phantom when the imaging beam was used. To calibrate the TLDs for the Al6 beam, the previously calibrated ionisation chamber system was used. From

65

the TLD reading and the absolute dose measured with the ionisation chamber, the calibration factor was established.

Imaging of humanoid phantom In our previous work [4] images of a tissue equivalent contrast phantom (Atlantis) gave quantitative information about the image quality of the new and the conventional systems. More clinically relevant, lateral images of the head and neck of an anthropomorphic Alderson-Rando phantom were obtained with the new and the conventional systems. The phantom had human skeleton in soft tissue equivalent material and air cavities. The gantry angle was 908, the centre of the beam passed through the centre of the phantom head, the detector was placed on a film holder 40 cm from the isocentre, and the field size used was 25!25 cm2. The dose delivered for the images produced similar film exposures for the two systems. The dose to the phantom to take an image with the new system was measured with the calibrated TLDs. Fifteen pairs of TLD chips were inserted in three slices of the phantom head. Seven pairs were placed in the middle slice and four pairs in each of the off axis slices. The three planes where the chips laid on were 5 cm apart.

Patient imaging Six patients having radical radiotherapy for head and neck cancer were imaged with the new and the conventional imaging systems. A dose check to ensure the beam safety and dose delivered was carried out daily prior to the patient’s treatment using diodes and the DPD510 diode system. The conventional system was the standard departmental system used with linacs without EPI, HR-E 30 Fuji film with lead screens. About 5 MU were delivered for these images. The Royal Marsden NHS Trust ethics committee gave ethical approval. Anterior images and lateral isocentre check films were taken with each imaging system on the same day after treatment delivery. The images were single exposed and were a size of 12!12 cm2. The images were compared and evaluated by a clinician. Two methods of rating the contrast of the films were used. The first used a method analogous to that used by Yin and Kruse [22,10]. Eight radiographers experienced in portal imaging matching evaluated the films. Data sheets were prepared which listed possible structures that may be visible in a head and neck isocentre film. The radiographers were asked to identify if the structure appeared on both films and rate the visibility on a scale of 1–5, with 5 being the clearest. The scores were compared using a non-parametric t-test. The second method required the radiographers to rate on a visual analogue scale (VAS), scale of 0–10, which film they would be more confident in using to evaluate the set-up errors using the same structure. The ratings were evaluated using a one-sample t-test, with a test value of 5, i.e. if there was no difference between the films the score would be equal to 5 and not significant.

Results In this section, phantom and patient images taken with the new system are shown and compared with

66

Low-energy portal imaging

the corresponding images taken with the conventional portal imaging system. Initially, the dosimetric characteristics of Al6 are discussed.

Absolute dosimetry The calibration factor for the ionisation chamber enabled absolute dosimetry and exact measurement of the output of Al6. For a 10!10 cm2 field at 5 cm depth Al6 produced 4!10K5 Gy/MU. The corresponding measurement for the 6 MV beam was 10K2 Gy/MU. This means that the 6 MV beam is 240 times more efficient per MU than Al6. Medical accelerators produce radiation with doserate of few gray per minute at 1 m from the source. When a photon beam is produced (in pulses), the mean dose output is determined by the number and length of the pulses and the electron beam current during a pulse. The electron beam current is of the order of hundreds of milliamps. When the electron beam is used for treatment and not for photon production the electron beam current required to produce a few gray per minute is about 1 mA [6]. The experimental beam is a modification of the nominal 4 MeV electron beam. So the electron beam current is much lower for Al6 than for the 6 MV beam. For Al6 the particle fluence seen by the monitor chamber inside the linac is a combination of photon fluence and contaminating electron fluence (filtered out downstream). This explains why the dose per monitor unit (MU) is much less than 10K2 Gy.

Images of humanoid phantom Images of the humanoid phantom taken with the two systems are shown in Fig. 1. Both images were scanned using a film digitiser (Vidar Systems Corporations, USA) and are of lower quality than the original film images. The horizontal dark lines are air gaps between the slices of the phantom. In the image from the new system (Fig. 1(b)), which obviously has higher contrast than the conventional, the cerebral vertebrae are visible and all air/bone interfaces are well defined. Fig. 2 shows top views of the three slices and the average dose (in units of 10K2 Gy) for each pair of TLDs. The beam entered from the left-hand side of the slices. The maximum dose measured was 6.9!10K3 Gy on the beam axis, 1 cm from the entrance surface. The RMS difference of the readings of the TLD pairs was 2.5%. The dose delivered for the conventional image is 4!10K2 Gy. Since all flattening filters are removed for the path of Al6, the beam and consequently the image are not flat. This problem will be solved when a camera or an amorphous silicon panel replaces the film, and the non-flatness can be automatically corrected.

Patient images The images of the humanoid phantom with the new and the conventional systems showed that the new system improved the image quality whilst less dose was required for image formation. The clinical study of the new low energy portal imaging system was started at the Royal Marsden Hospital (Sutton) in November 2002. The subjects were patients treated for head and neck cancer with radical radiotherapy. The delivery of safe high dose radiation to

Fig. 1. Left lateral head and neck images of the humanoid phantom taken with (a) the conventional and (b) the new systems.

patients with head and neck cancer requires the patient to be reproducibly immobilised each day for a course of treatment lasting for 4–6 weeks. During treatment, portal images are taken to ensure that the radiation beam is correctly delivered. The patients had two additional images taken. This took about 1 min extra time. The additional radiation dose per image was estimated at less than 10K2 Gy (i.e. approximately 0.02% of a total 50 Gy treatment).

S. Flampouri et al. / Radiotherapy and Oncology 76 (2005) 63–71

67

Fig. 2. Absolute dose in units of cGy (10K2 Gy) measured by TLD at the position of the labels on a plane (a) 5 cm above the isocentre, (b) through the isocentre and (c) 5 cm below the isocentre.

Fig. 3 shows the images of the lateral treatment field of the first patient. Fig. 3(a) is the diagnostic X-ray image from the simulator (ELEKTA, UK), (b) is the image taken with the new system (Al6/mammography system) and (c) is the conventional portal image taken with the treatment beam 6 MV and the Fuji radiotherapy verification system. For (b) and (c) the imaging field size was 12!12 cm2 and the detector was placed on a film holder 40 cm from the isocentre. The patient was positioned with an immobilization device (thermoplastic shell). The portal images were taken at the end of the treatment session for retrospective analysis. The dose delivered was 5 MU, which correspond to approximately 5!10K2 Gy, for the conventional image and 150 MU, which correspond to a calculated maximum dose to the patient of w10K2 Gy for the new system. The quality of the scanned images is inferior to the quality of the original films. The new system produced a dramatic contrast improvement compared to the conventional with less dose. The clinician who assessed the two portal images noted the visible anatomical structures and the structures that could confidently be used for patient set up (Table 1). The numbers of Fig. 4 correspond to the structures of Table 1. Structures that are visible on both images are clearer on the new system image. All air/tissue interfaces have higher contrast for the new system compared to the conventional. The sphenoidal sinus is better defined. The posterior choanae that are not distinguishable on the conventional

image could be used for patient set up. Soft palate and epiglottis, even though they cannot be used for set up, are much clearer. Vertebral bodies, disc spaces and spinous processes are observable. The hyoid bone is also discernable. Six patients receiving radiotherapy to the head and neck were consented for this study and had anterior and lateral isocentre check films imaged. The anatomical structures that had more than eight matched observations were compared. Each structure, therefore, had to be seen on at least two sets of films. Tables 2 and 3 show the structures assessed, the number of paired observations and a comparison of score of each film for the lateral and anterior films, respectively. The differences between scores using each imaging system were all significant on the lateral films and all but the orbits significant in the anterior film. However, it must be noted that the orbits had less observations of any of the structures, 13 matched results as compared to 30 for the cervical spine. When the visual analogue scores were evaluated the mean of the structures compared was greater than 5, demonstrating better confidence when using the LE films, but it was not significant for all the structures (Tables 4 and 5). The C-spine, hard palate and hyoid bone significantly more visible in the lateral films but only the maxilla and C spine were significantly different in the anterior films.

Fig. 3. Right lateral head and neck patient images of the isocentre taken with (a) the simulator (diagnostic X-rays) system, (b) the new system and (c) the conventional system.

68

Low-energy portal imaging

Table 1 The visible anatomical structures and the structures that could confidently be used for patient set up from both portal images Structure number

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Structure name

Occipital bone Petrous temporal bone Sphenoidal sinus Posterior choanae Nasal cavity Hard palate Soft palate Teeth Mandible Hyoid bone Epiglottis Nasopharynx Larynx C1 (atlas) C2 (axis) C3 C4 Spinous processes

New system

Conventional

Visible structures

Structures for registration

Visible structures

Structures for registration

# # # # # # # # # # # # # # # # # #

#

# # #

#

# # #

#

# # # # #

#

#

# # # # # # #

Numbers refer to Fig. 4.

Discussion Patient imaging The first patient images showed that the new method produces high quality images with low dose. For the first patient, the two images produced with the new and the conventional systems were compared and scored by a clinician (CN). Air/soft tissue was better distinguished, and bone was more clearly seen. As these structures are the basis of head and neck portal imaging, the new images show potential to improve treatment verification. The improvements in quality of portal images were significant and it was decided to test this imaging method further. A group of six patients were imaged and the image quality scored using a quantitative assessment scale 0–10. The observers were blinded to the method used to obtain the image to avoid bias. The results showed the experimental system to be superior in the majority of cases. The method described, works best for thin and medium radiological thickness treatment sites such as head, neck or chest, which do not completely absorb the low energy photons. However, image improvement is expected for thicker sites since previous experiments [4] showed that image quality improvement compared to conventional imaging systems is successful with water phantoms up to 25 cm thick. The next goal of this work will be to transfer the method to a treatment machine with a flat panel EPID to see if improvements may also be made with an EPID. On-line, interactive portal verification will then be available. Although due to the long exposure time the additive noise from an electronic system may be problematic for a low dose rate and hence, long exposure. Although the dose delivered for patient imaging is minimal, the use of large number of MU’s for imaging

makes the technique slow and unlike conventional verification imaging. This is because a thin target is used that does not stop all electrons that are accelerated and these electrons that pass through the target produce a large signal in the ionisation chamber in the treatment machine head. Possible changes to the linac configuration to resolve this problem include: increasing the target thickness, increasing the electron gun current and recalibration of the MU settings of the ionisation chamber, or positioning of the PMMA electron filter upstream of the monitor chamber. The use of a greater MU efficiency and an electronic detector instead of film is necessary for on-line, interactive verification.

Fig. 4. Simulator X-ray image of the patient with visible anatomical structures numbered.

S. Flampouri et al. / Radiotherapy and Oncology 76 (2005) 63–71

69

Table 2 Lateral film scores. (CS!NS: conventional system worse than new system, CSONS: conventional system better than new system, both system images are of similar quality) Anatomy

No. of paired observations

CS!NS

CSONS

CSZNS

P value

C spine Hard palate Teeth Hyoid bone Maxilla Sphenoid sinus

39 33 31 26 24 19

37 21 29 17 20 10

0 3 0 2 0 1

2 9 2 7 4 8

P!0.001 P!0.001 P!0.001 PZ0.001 P!0.001 P!0.01

Anatomy

No. of paired observations

CS!NS

CSONS

CSZNS

P value

C spine Nasal septum Maxilla Teeth Orbit

30 29 19 14 13

26 13 13 8 7

0 5 0 2 3

4 11 6 4 3

P!0.001 P!0.05 PZ0.01 P!0.05 Not significant

Table 3 Anterior film scores

Geometrical considerations The imaging beam Al6 is produced on a bremsstrahlung target placed 14.4 cm below the normal accelerator target. This results in different beam divergence between the imaging and the treatment beam. As shown in Fig. 5(a), with the same jaw positions the 6 MV beam produces a 10! 10 cm2 field, while Al6 produces a 11.9!12.9 cm2 field at the isocentre plane. At the imager plane (150 cm from the source (original target)) the first field is 15!15 cm2 while the second 18.9!20.48 cm2. This is a significant difference in field size, however the geometrical error of a structure placed at the isocentre plane and imaged at 150 cm is small. As shown in Fig. 5(b), a cubic object (10!10!10 cm3) placed in the middle of the field with its surface at the isocentre produces an image of 15!15 cm2 with the conventional system and 15.8!15.8 cm2 for the new system. For the existing system, the divergence problem could be overcome by comparing the portal images with DRR’s created with the same divergence as the experimental Table 4 Visual analogue scale scores for the lateral films

beam. Another approach is to rescale the field edge position on the film by bringing the leaf edge position towards the central axis. To do this it is necessary to know where the central axis falls on the film however. Elementary geometry shows that if LT is the distance of the new target below the conventional target, LJ is the distance of the collimator jaws below the conventional target and LI is the isocentre distance then the necessary scaling of the field edge position from the centre of the film is: (1KLT/LJ)/(1KLT/LJ). Ultimately, such a system would be used with the low Z target in the position of the standard target and hence the divergence problem would be removed.

Conclusions In this paper the commissioning for patient imaging of a modified megavoltage beam, which has a boosted low energy component, was described. For absolute dosimetry, Table 5 Visual analogue scale scores for the anterior films

Anatomy

No. of paired observations

Mean score

Std deviation

P value

C spine Hard palate Teeth Hyoid bone Maxilla Frontal sinus

36 36

8.1 6.6

2 1.3

P!0.001 P!0.001

4 24

7.8 6.7

0.9 2.1

P!0.01 PZ0.001

23 9

6.7 6.6

2.1 2.4

PZ0.001 Not significant

Anatomy

No. of paired observations

Mean score

Std deviation

P value

C spine Nasal septum Maxilla Teeth Orbit

19 26

7.5 5.3

1.6 2.1

17 8 13

6.7 7.4 5.9

1.7 1.5 2.6

P!0.001 Not significant P!0.001 P!0.01 Not significant

70

Low-energy portal imaging

Fig. 5. (a) Beam divergence for the 6 MV and the Al6 beams. (b) Geometrical error produced by the beam divergence.

the calibration of a field instrument was required. The calibration was based on a megavoltage protocol, as there was no evidence that the low energy component of the beam affect its dosimetric characteristics. As the same instrument was calibrated for 60Co it was not felt necessary to compare its reading with a secondary standard dosimeter. TLDs were also calibrated for the imaging beam. Images of the head and neck of a humanoid phantom were taken with the new imaging system (imaging beam and mammography detector) and the conventional (treatment beam and radiotherapy verification detector). The dose to the phantom (measured with the TLDs) to form an image was less than 10K2 Gy for the new system (4!10K2 Gy for the conventional) and the contrast increase was dramatic. These results led to a clinical trial of the new system. The first patient images taken were compared with the conventional portal images. All anatomical structures that are visible on both films are clearer with the new system. Additionally structures, such as vertebrae are visible only on the mammography film. So we conclude that the contrast on portal images can be dramatically increased by modifying the beam and the detector for low energy imaging and that the imaging method developed is clinically applicable and useful.

Acknowledgements This work is supported by Cancer Research UK (under Grant Reference SP 2312/0201) and the Institute of Cancer Research. We are grateful to Elekta Oncology Systems for useful information and

assistance. We are grateful to the following people for help with various aspects of this work: David Convery, Nick Brigden and Stephanie Reise.

* Corresponding author. Philip Evans, Joint Physics Department, Institute of Cancer Research and Royal Marsden NHS Foundation Trust, Downs Road, Sutton, Surrey SM2 5PT, UK. E-mail address: [email protected] Received 11 November 2004; received in revised form 24 February 2005; accepted 6 April 2005; available online 24 May 2005

References [1] Berbeco RI, Jiang SB, Sharp GC, et al. Integrated radiotherapy imaging system (IRIS): design considerations of tumour tracking with linac gantry-mounted diagnostic x-ray systems with flatpanel detectors. Phys Med Biol 2004;49:243–55. [2] Bissonnette JP, Cunningham IA, Munro P. Optimal phosphor thickness for portal imaging. Med Phys 1997;24:803–14. [3] Cho YB, Kilovision Munro P. Thermal modeling of a kilovoltage x-ray source integrated into a medical linear accelerator. Med Phys 2002;29:2101–8. [4] Flampouri S, Evans PM, Verhaegen F, et al. Optimisation of accelerator target and detector for portal imaging using Monte Carlo simulation and experiment. Phys Med Biol 2002;47: 3331–49. [5] Galbraith DM. Low-energy imaging with high-energy bremsstrahlung beams. Med Phys 1989;16:734–46. [6] Greene D, Williams PC. Linear accelerators for radiation therapy. In: Mould RF, Orton CG, Spaan JAE, Webster JG, editors.. Bristol and Philadelphia: Institute of Physics Publishing; 1997.

S. Flampouri et al. / Radiotherapy and Oncology 76 (2005) 63–71 [7] IAEA. Absorbed dose determination in external beam radiotherapy: an international code of practice for dosimetry based on standards of absorbed dose to water. Technical Report Series 398, IEAE, Vienna; 2000. [8] Jaffray DA, Siewerdsen JH, Wong JW, Martinez AA. Flat-panel cone-beam CT for image guided radiation therapy. Int J Radiat Oncol Biol Phys 2002;53:1337–49. [9] Khan FM. The physics of radiation therapy. In: Passano WM, editor.. Baltimore: Lippincot Williams and Wilkins; 1994. [10] Kruse JJ, Herman MG, Hagness CR, et al. Electronic and film portal images: a comparison of landmark visibility and review accuracy. Int J Radiat Oncol Biol Phys 2002;54:584–91. [11] Krutman Y, Faermann S, Tsechanski A. Use of mammographic film-screen-cassette combination for portal localization imaging with a 4 MV x-ray unflattened beam. Phys Med 2002;XVIII: 55–62. [12] Lewis DG, Swindell W, Morton EJ, Evans PM, Xiao ZR. A megavoltage CT scanner for radiotherapy verification. Med Phys 1992;37:1985–99. [13] Lillicrap SC, Owen B, Williams JR, Williams PC. Code of practice for high-energy therapy dosimetry based on the NPL absorbed dose calibration service. Phys Med Biol 1990;35:1355–60. [14] Mah DW, Galbraith DM, Rawlinson JA. Low-energy imaging with high-energy bremsstrahlung beams: analysis and scatter reduction. Med Phys 1993;20:653–65.

71

[15] Ostapiak OZ, O’Brien PF, Faddegon BA. Megavoltage imaging with low Z targets: implementation and characterization of an investigational system. Med Phys 1998;25:1910–8. [16] Pang G, Beachey DJ, O’Brien PF, Rowlands JA. Imaging of 1 mm diameter radiopaque markers with megavoltage x-rays: an improved online imaging system. Int J Radiat Oncol Biol Phys 2002;52:532–7. [17] Radcliffe T, Barnea G, Wowk B, Rajapakshe R, Shalev S. Monte Carlo optimisation of metal/phosphor screens at megavoltage energies. Med Phys 1993;20:1161–9. [18] Ruchala KJ, Olivera GH, Schloesser EA, et al. Megavoltage CT on a tomotherapy system. Phys Med Biol 1999;44:2597–621. [19] Swindell W, Simpson RG, Oleson JR, Chen CT, Grubbs EA. Computed tomography with a linear accelerator with radiotherapy applications. Phys Med Biol 1983;10:416–20. [20] Tsechanski A, Bielajew AF, Faermann S, Krutman Y. A thin target approach for portal imaging in medical accelerators. Phys Med Biol 1998;43:2221–36. [21] Wowk B, Radcliffe T, Leszczynski KW, Shalev S, Rajapakshe R. Optimisation of metal/phosphor screens for on-line portal imaging. Med Phys 1994;21:227–35. [22] Yin FF, Rubin P, Schell MC, et al. An observer study for direct comparison of clinical efficacy of electronic to film portal images. Int J Radiat Oncol Biol Phys 1996;35:985–91.