In vivo dosimetry for lung radiotherapy including SBRT

Physica Medica xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Physica Medica journal homepage: http://www.physicamedica.com

Review paper

In vivo dosimetry for lung radiotherapy including SBRT Boyd M.C. McCurdy a,b,c,⇑, Peter M. McCowan a a

Division of Medical Physics, CancerCare Manitoba, Winnipeg R3E 0V9, Canada Department of Physics and Astronomy, University of Manitoba, Winnipeg R3M 2N2, Canada c Department of Radiology, University of Manitoba, Winnipeg R3M 2N2, Canada b

a r t i c l e

i n f o

Article history: Received 2 February 2017 Received in Revised form 20 May 2017 Accepted 22 May 2017 Available online xxxx Keywords: In vivo dosimetry Dose verification Treatment verification Lung SBRT

a b s t r a c t SBRT for lung cancer is being rapidly adopted as a treatment option in modern radiotherapy centres. This treatment is one of the most complex in common clinical use, requiring significant expertise and resources. It delivers a high dose per fraction (typically 6–30 Gy/fraction) over few fractions. The complexity and high dose delivered in only a few fractions make powerful arguments for the application of in vivo dosimetry methods for these treatments to enhance patient safety. In vivo dosimetry is a group of techniques with a common objective – to estimate the dose delivered to the patient through a direct measurement of the treatment beam(s). In particular, methods employing an electronic portal imaging device have been intensely investigated over the past two decades. Treatment verification using in vivo dosimetry approaches has been shown to identify errors that would have been missed with other common quality assurance methods. With the addition of in vivo dosimetry to verify treatments, medical physicists and clinicians have a higher degree of confidence that the dose has been delivered to the patient as intended. In this review, the technical aspects and challenges of in vivo dosimetry for lung SBRT will be presented, focusing on transit dosimetry applications using electronic portal imaging devices (EPIDs). Currently available solutions will be discussed and published clinical experiences, which are very limited to date, will be highlighted. Ó 2017 Published by Elsevier Ltd on behalf of Associazione Italiana di Fisica Medica.

Contents 1. 2. 3.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rationale for in vivo dosimetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Currently available solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Point detectors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. EPID detectors (transit dosimetry) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1. Technical challenges. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.2. 2D transmission image comparisons (measurement-to-predicted). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.3. Dose reconstruction from transmission images (0D, 2D, 3D) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Clinical experience to date . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.1. Non-SBRT lung . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.2. SBRT lung . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Outlook and future directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Funding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00

⇑ Corresponding author at: Division of Medical Physics, CancerCare Manitoba, Winnipeg R3E 0V9, Canada. E-mail address: [email protected] (B.M.C. McCurdy). http://dx.doi.org/10.1016/j.ejmp.2017.05.065 1120-1797/Ó 2017 Published by Elsevier Ltd on behalf of Associazione Italiana di Fisica Medica.

Please cite this article in press as: McCurdy , McCowan . In vivo dosimetry for lung radiotherapy including SBRT. Phys. Med. (2017), http://dx.doi.org/ 10.1016/j.ejmp.2017.05.065

2

B.M.C. McCurdy, P.M. McCowan / Physica Medica xxx (2017) xxx–xxx

1. Introduction Stereotactic body radiation therapy (SBRT) for lung cancer is being rapidly adopted as a standard treatment option in modern radiotherapy centres. This treatment is one of the most complex in common clinical use, requiring significant expertise in respiratory motion management, 3D and/or 4D imaging in both pre-treatment and on-treatment settings, dose calculation algorithms, inverse treatment planning, IMRT and/or VMAT delivery techniques, and potentially deformable volumetric image registration. This complexity, combined with a high prescription dose per fraction (typically 6–30 Gy/fraction), make powerful arguments for the application of in vivo dosimetry methods for these treatments. By estimating the dose delivered to the patient through a direct measurement of the treatment beam(s), ie. in vivo dosimetry, medical physicists and clinicians may have a much higher degree of confidence that the dose has been delivered to the patient as intended. In vivo dosimetry is recognized and recommended by several international organizations (e.g. IAEA and WHO) as an important quality assurance tool [1–3]. And in some countries, for example Sweden, France, and the United Kingdom, in vivo dosimetry is included within national radiotherapy guidelines [4–6]. In vivo dosimetry is very powerful in that it can catch errors that many existing quality assurance (QA) techniques, including pre-treatment QA, will miss [7]. Analysis of incident reporting system data has quantitatively demonstrated that in vivo dosimetry is a highly effective addition to the common array of quality assurance techniques, providing a significant increase in error sensitivity, as well as being rated one of the most effective checks [8]. Mijnheer et al., Kron et al., and references therein provide recent and detailed overviews of in vivo dosimetry [7,9]. Much research effort has gone into developing and exploring in vivo dosimetry methods over the past two decades, mainly focusing on transit imaging approaches. In general, the measurement methods can be categorized as point-based measurements (ie. a single point dose measured with a diode, thermoluminescent dosimeter, MOSFET, or other point detector) or image-based measurements (ie. a megavoltage planar imaging system). Image-based measurements have the potential to provide much more information compared to point-based measurements. Image-based methods can be further classified into i) 2D image comparisons or ii) dose reconstruction methods. The 2D image comparison methods directly compare 2D measured transmission images to 2D predicted transmission images. In contrast, dose reconstruction methods make estimates of the actual delivered dose within a patient model, and these estimates can be 0D (a point), 2D (a plane), or 3D (a volumetric dose distribution). Most of the image-based methods can potentially be employed as a function of irradiation time to obtain time-resolved patient dosimetry information. Note that several techniques presented in the literature utilize some real-time data acquired during treatment, but assume the treatment plan or some portion of the treatment plan is delivered exactly as intended and therefore for the purpose of this review are not considered fully in vivo dosimetry approaches. For example the MAASTRO group (The Netherlands) developed a method to estimate the 3D patient dose from non-transmission EPID images [10,11], which assumes the treatment plan is delivered faithfully. In their approach, EPID images of the delivered treatment fields are acquired without the patient present, and are converted to a water-equivalent portal dose image. This water-equivalent dose image is deconvolved with a dose deposition kernel, yielding an estimate of energy fluence, which is then back projected to a plane within the linear accelerator head. The patient dose is then calculated on a computed tomography (CT) or cone-beam CT (CBCT)

model of the patient using XVMC [12] assuming that the incident photons are generated from a point-source at the linac focal spot location, with energy sampled from the energy fluence distribution derived from the non-transmission EPID field image(s). This method was demonstrated on four lung SBRT patients using megavoltage CBCT image data sets [13], while Persoon et al. presented 3D dose estimates for five example VMAT, non-SBRT patient cases using kilovoltage CBCT data sets [14]. In the latter study, treatment plans were modified in four out of five example cases, including one patient where significant changes in atelectasis were identified. Other groups are interested in dose estimates for lung tumours which account for the tumour motion. For example, lung tumour trajectories may be tracked with real-time planar imaging combined with automatic tumour segmentation techniques, and then the patient dose calculated for the tumour at its estimated position using the original treatment plan [15–17]. The estimated real-time lung tumour position can be obtained in other ways, for example via an external surrogate [18]. Lin et al. [19] used implanted fiducials to estimate lung tumour position in real-time, and also utilized real-time megavoltage transmission images to estimate MLC positions and extracted delivered monitor units as a function of time from linac log files, post-irradiation. That work also demonstrated that by considering the timing of the tumour position and the MLC positions, the interplay effect could be accounted for. While these foregoing approaches can be very useful applications for estimating dose within the tumour or patient for lung SBRT, for the purposes of this review, they are not considered fully in vivo dosimetry since they make assumptions regarding the reproducibility and accuracy of linear accelerator performance. There are also some commercial devices available that monitor entrance fluence to the patient (i.e. arrays mounted on the head of the linac, positioned below final collimation), but measurements with these devices do not include effects of real-time patient anatomy or position changes during the treatment, or immobilization devices, and thus are also not considered fully in vivo dosimetry for this review. Currently only a handful of academic centres have substantial clinical experience implementing in vivo dosimetry programs, including for lung SBRT treatments. In-house developed software and (often) customized hardware is utilized. There are some commercial packages available including EPIgray (Dosisoft, Paris, France) and DISO (Università Cattolica S. Cuore, Rome, Italy) which perform single or few point in vivo dosimetry, as well as Dosimetry Check (Math Resolutions, Columbia, MD, USA) and recently iViewDose (Elekta AB, Stockholm, Sweden) both of which handle 3D in vivo dosimetry, although some limitations of the former have been reported [20]. As more commercial solutions become available, clinical experience with in vivo dosimetry for all disease sites will grow rapidly. In this review, currently available solutions proposed in the literature for in vivo dosimetry, specifically those that have provided examples of lung radiation treatment or lung SBRT applications, will be presented. The few publications providing clinical in vivo dosimetry results for lung SBRT will also be summarized. 2. Rationale for in vivo dosimetry The rationale for in vivo dosimetry of any disease site in radiotherapy is to catch errors that would otherwise be undetected. For lung SBRT, plan and anatomy complexity as well as large doses in fewer fractions, further increase the need for in vivo dosimetry for this patient population. By estimating the dose delivered to the patient through a direct measurement of the treatment beam(s), medical physicists and clinicians may have a much higher degree of confidence that the dose has been delivered as intended.

Please cite this article in press as: McCurdy , McCowan . In vivo dosimetry for lung radiotherapy including SBRT. Phys. Med. (2017), http://dx.doi.org/ 10.1016/j.ejmp.2017.05.065

B.M.C. McCurdy, P.M. McCowan / Physica Medica xxx (2017) xxx–xxx

In addition to dose verification, there may be useful roles for in vivo dosimetry to play in quality assurance for clinical trials [9] as well as in adaptive radiotherapy applications, where estimates of the actual delivered dose can be utilized to better refine future treatment fractions.

3. Currently available solutions 3.1. Point detectors There are a wide variety of point dosimeters available for use as in vivo dosimeters. Thermoluminescent dosimeters (TLDs), optically stimulated luminescent dosimeters (OSLDs), silicon diodes, metal-oxide semi-conductor field effect transistors (MOSFETs), and plastic scintillators are used throughout the world in this role, although not commonly for lung SBRT unless legally required. Potential new point dosimeters for in vivo dosimetry are still being actively investigated, for example, recently gel patch dosimeters were applied to lung SBRT cases (in phantom) in a proof-of-concept demonstration [21]. In general, the point detector is placed at a specific location on the patient’s skin surface at a beam entrance (or exit) location, and the measured dose is compared to the dose predicted at the patient’s skin surface at that point by the treatment planning system. However there are significant practical challenges using point detectors including reproducibility of positioning of the detector at the intended entrance/exit surface location, lower accuracy of the treatment planning system’s patient dose calculation algorithms at the patient entrance surface (and to a lesser extent at the exit surface), limited sampling (lung SBRT delivery is typically 8–12 static conformal/IMRT fields or several VMAT arc deliveries), and furthermore their use is very resource intensive.

3.2. EPID detectors (transit dosimetry) Electronic portal imaging devices (EPIDs) are imaging systems that provide digital images formed using the megavoltage therapy beam. Originally developed to provide pre-treatment and on-treatment anatomical images to verify and guide patient setup and shielding placement (long before KV imaging on linacs became routine), they have recently been heavily investigated as a dosimetry tool, including for in vivo dosimetry. All major radiotherapy equipment manufacturers supply a state-of-the-art amorphous-silicon EPID on nearly every linear accelerator in modern radiotherapy facilities. A benefit of image measurement based approaches compared to point measurement based approaches is that they potentially provide far more information, since an entire plane of fluence/dose data is captured instead of only a single point (although this area/volume is ultimately limited by the size of the imager with respect to the beam, since information from the edges of the treatment field that fall outside the imager will be lost). Furthermore, the data capture and storage process is mostly automated, facilitating more efficient routine use. In general, the measurement of the 2D dose or signal in the EPID during a treatment delivery is used to infer information about the dose actually delivered to the patient. This can be done by comparing a measured transmission image to a predicted transmission image or, alternatively, by analysing the EPID image(s) to estimate the dose delivered within the patient. The techniques available in the literature, which have been specifically applied to lung cancer patients or lung SBRT patients, will be described further below. First however, these approaches require an understanding of the many technical factors involved in the radiation transport through

3

the patient and the EPID imaging system, as well as how the imaging system behaves. 3.2.1. Technical challenges While EPIDs are widely available and easily used for routine setup imaging, there are several technical challenges to using them for in vivo dosimetry applications. These have been documented elsewhere [22,23] and will be only briefly reviewed here. The current generation of EPIDs utilize a metal plate backed by a thin phosphor scintillating screen, which makes the system overly sensitive to low-energy photons as compared to a water-equivalent detector. This has implications for modeling linacs with and without flattening filters, where off-axis energy spectra are different. Furthermore, the EPID signal is inherently a dose-to-phosphor measurement whereas many researchers prefer to work with dose-to-water. The imagers also exhibit pixel-to-pixel sensitivity variations, unique to each individual imager, which need to be accounted for. Also, EPIDs are susceptible to image ghosting and image lag effects. Optical photons scatter and spread within the translucent phosphor scintillator, introducing a small amount of optical blur. Photons from the primary beam that are scattered within the patient but still reach the EPID also need to be accounted for, and this patient scatter signal depends on patient geometry, field size, and the proximity of the EPID to the patient. The mechanical deflection of the gantry and EPID mount arm introduce small geometric errors that should be corrected. The mounting arm design on some EPIDs, notably the aS500/1000 on the E-arm, (Varian Medical Systems, Palo Alto, CA) leads to asymmetric backscatter effects in the acquired images. Flattening filter free beams may experience imager saturation. All of these effects have been investigated and well-described in the literature and are correctable or at least able to be modeled using various techniques. The application of in vivo dosimetry to lung SBRT in particular emphasizes one of these technical challenges. Lung patients present the most heterogeneous patient geometries and therefore strain the patient scatter estimation aspect of these types of models. Another technical challenge, specific to patient dose reconstruction methods, is the calculation of patient dose. Many methods are available in the literature and vary from simple pencil beam algorithms to full Monte Carlo approaches. Consideration of the algorithm used to make the patient dose estimate is important since, if it is different than the treatment planning system algorithm (used for comparison as the reference dose), then further uncertainty will be introduced. 3.2.2. 2D transmission image comparisons (measurement-topredicted) Many groups explored this method in the late 1990’s and early 2000’s using earlier generation technology EPIDs (e.g. liquid ionization matrix and charged-couple device or CCD camera systems) and a summary of those works is provided in a comprehensive 2008 EPID dosimetry review article [24]. This approach, which compares a measured transmission image with a predicted transmission image, can reveal differences between the delivered and planned dosimetry, but does not provide much detailed information about the dosimetric impact to the patient due to this difference. For example, Kroonwijk et al. used measured-to-predicted portal dose image comparisons to detect gas pockets in some prostate patient treatments [25]. The group at the MAASTRO centre used a semi-empirical method to convert measured electronic portal images to portal dose images, and require a set of non-transmission EPID images as part of the conversion process [26]. The portal dose images are an estimate of the water-equivalent dose that would be measured at the EPID plane. The method is based on that of Pasma et al. [27,28]. This technique has been applied to a 3D conformal

Please cite this article in press as: McCurdy , McCowan . In vivo dosimetry for lung radiotherapy including SBRT. Phys. Med. (2017), http://dx.doi.org/ 10.1016/j.ejmp.2017.05.065

4

B.M.C. McCurdy, P.M. McCowan / Physica Medica xxx (2017) xxx–xxx

radiotherapy lung cancer patient (and three other non-lung patients) to study trend analysis throughout a treatment course [29]. They have also presented data for a larger number of lung cancer patients (460 non-SBRT) treated with volumetric modulated arc therapy (VMAT), using measured integrated portal transit images (ie. a summation of all EPID images obtained over an entire VMAT arc) and categorized causes of differences between planned and measured images [30]. It was concluded that integrated portal transit images were of limited value in detecting on treatment patient anatomy changes, and this was emphasized again in later work with two lung patients out of four case studies [31]. The 460 non-SBRT patients were recently re-analysed and there was found to be little correlation between integrated portal transit dose images and dose-volume histogram metrics [32]. The in-air, pre-treatment EPID dosimetry algorithm proposed by Van Esch et al. [33] was extended to transit dosimetry applications by Berry et al. [34]. This semi-empirical approach added a narrow-beam attenuation factor and also two measurement-based correction factors that accounted for: i) the patient scatter, via a buildup factor which was dependent on phantom thickness and field size and ii) off-axis beam energy changes. The method was applied on 57 treatment fractions over nine patients, including two lung patients [35]. There has been further recent interest in the image comparison technique for a real-time error detection system [36,37], since the technique is relatively computationally simple and therefore attractive for real-time applications. Data specific to lung SBRT has not yet been published but the method is general and can be applied to any external beam treatment technique. Their approach for image prediction utilizes the model-based technique of Chytyk-Praznik et al. [38]. The method employs a two-source physical model of the linear accelerator head fluence, combined with the patient scatter prediction model of McCurdy and Pistorius [39,40], and a fluence-to-dose conversion method that physically models the EPID’s energy response. 3.2.3. Dose reconstruction from transmission images (0D, 2D, 3D) 3.2.3.1. 0D reconstruction (dose to a point within the patient). These methods generally use semi-empirical models to relate EPID signal at the central axis point (or central region) of the measured transmission image of the field to a point within the patient. Piermattei et al. derived correlation functions between EPID signal and isocentre dose for a variety of geometries (i.e. phantom thicknesses, field sizes) and used these to relate signal at the centre of the EPID to dose within the patient at isocentre [41,42]. Francois et al. implemented a similar empirical approach to relate signal at the EPID centre (converted to dose-to-water) to an upstream point in the patient [43]. Nijsten et al. [44] also implemented a method to back-project the measured central field EPID signal (converted to dose-to-water) into the patient to a 5 cm depth below the entrance surface. The method also used a set of measured EPID data for a variety of phantom thicknesses and field sizes, in order to estimate scatter-to-primary ratios at the EPID. 3.2.3.2. 2D reconstruction (dose to a plane within the patient). The work of Boellaard [45–48] used a liquid ionization matrix style EPID and developed a technique to reconstruct patient dose either at the exit plane or mid-plane of the patient for 3D conformal radiotherapy fields. This technique was eventually adapted to amorphous-silicon style EPIDs and updated to include IMRT in the pre-treatment setting [49] and then implemented as an in vivo tool in 2007 for prostate patients [50]. The method is also semi-empirical but more complex than 0D methods, since additional effects such as off-axis EPID response and off-axis beam energy spectra need to be accounted for in a two-dimensional manner. This method was in clinical use at the Antoni van Leeuen-

hoek hospital (The Netherlands) for many years before being replaced with a full 3D dose reconstruction technique. The technique was unable to accurately handle dose estimates in lung patients however, due to the heterogeneity of the patient [51]. 3.2.3.3. 3D reconstruction (dose to the patient volume). Continuing on with the algorithm development at the Antoni van Leeuenhoek hospital, they extended their 2D patient reconstruction to 3D by applying the 2D reconstruction method to every plane within the patient [52] but did not include tissue heterogeniety corrections at that time. In 2012 Wendling et al. [51] introduced a modification of the 3D stacked planar dose reconstruction technique, specifically for lung treatments. They replaced the heterogeneous patient model (ie. CT simulation data set) with a homogenous, water-equivalent patient model that preserved the external patient contour but reset all internal density to water. The estimated reconstructed delivered dose in the water-equivalent patient model was then compared to the treatment planning system dose recalculated with a homogeneous density override (i.e. to emulate the water-equivalent patient model). This was done to overcome the limitations of the simple dose calculation technique employed, which did not perform as accurately as desired in heterogeneous lung patients. Since the technique does not actually estimate the true in vivo patient dose, they renamed it ‘in aqua vivo’ dosimetry. Although the in aqua vivo dosimetry estimate is made in a virtual homogeneous phantom (albeit patient-shaped) instead of the actual patient CT dataset, it is compared to the treatment plan prediction in the same virtual phantom, and therefore is sensitive to changes in patient anatomy, patient setup variations, and linac output changes, and therefore should be able to detect delivery problems. The group at CancerCare Manitoba (Canada) has clinically implemented an iterative method for estimating the primary fluence entering the EPID [53] based on the work of McNutt et al. [54], but employ their own model-based EPID prediction method. This approach provides a 2D estimate of the incident fluence on the patient as a function of energy (discretized into several energy bins). An in-house collapsed-cone convolution model based on Ahnesjö [55] is then used to calculate patient dose. This model has been validated specifically for stereotactic radiotherapy beams (1000 MU/minute, Trilogy units, Varian Medical Systems) for their clinical in vivo dosimetry program implemented for SBRT patients [56]. 3.3. Clinical experience to date Presently, there is very little published data reporting on the clinical experience of in vivo dosimetry programs for lung SBRT, thus both non-SBRT and SBRT clinical experience for lung patients are reviewed here. There are a few publications that apply in vivo dosimetry methods on only a small number of lung patients, typically in the context of illustrating a particular technique, and these will be briefly overviewed here. From the MAASTRO group, Persoon demonstrated 2D transit dose comparisons for one 3DCRT lung example [14] and two VMAT lung examples [31]. In New York, 2D transit dosimetry was demonstrated on two lung patients (9– 11 dynamic IMRT fields per patient), and a significant change in bilateral pleural effusion was discovered in one patient resulting in a re-plan partway through the treatment course [35]. The Antoni van Leeuonhoeck group presented a 3D dose reconstruction for one lung SBRT patient (VMAT) [57]. The CancerCare Manitoba group demonstrated their model-based 3D dose reconstruction technique on one lung SBRT patient treated with VMAT [56]. The forementioned papers are useful in that they illustrate the technical details of a particular technique used and provide some limited clinical feedback, however the remainder of this section focuses

Please cite this article in press as: McCurdy , McCowan . In vivo dosimetry for lung radiotherapy including SBRT. Phys. Med. (2017), http://dx.doi.org/ 10.1016/j.ejmp.2017.05.065

5

B.M.C. McCurdy, P.M. McCowan / Physica Medica xxx (2017) xxx–xxx Table 1 Summary of clinical experience published to date involving lung radiotherapy. Reference

Delivery

Dose estimate

Patients/fractions

Dosimeter

Non-SBRT lung

Nijsten et al. 2007 [44] Piermattei et al. 2009 [59] Mans et al. 2010 [62] Persoon et al. 2015 [30] Fidanzio et al. 2015 [61] Celi et al. 2016 [58]

3D-CRT 3D-CRT IMRT VMAT 3D-CRT 3D-CRT/ IMRT

Point D5cm Point Diso 2D mid-plane 2D integrated at EPID Point Diso Point

587 fx 20 pts 454 pts 460 pts 120 pts Unspecified

CCD imager (Theraview-NT) Mix of Varian EPID and ion chamber Elekta EPID Varian EPID Varian EPID Varian EPID

SBRT lung

Wendling et al. 2012 [51] Mijnheer et al. 2015 [63] McCowan et al. 2017 [64]

IMRT/VMAT VMAT VMAT

3D ‘in aqua vivo’ 3D ‘in aqua vivo’ 3D

751 pts non-SBRT/50 pt SBRT >160 pts 100 pts

Elekta EPID Elekta EPID Varian EPID

on publications that include a larger body of clinical data, from at least 20 lung patients (either non-SBRT or SBRT). The publications from this latter group are summarized in Table 1.

3.3.1. Non-SBRT lung Recently a large clinical implementation of the EPIGray software (Dosisoft, Paris, France) was reported by Celi et al. [58]. Data for a one or few-point dose reconstruction were presented for a two year period involving 3163 patients, including lung patients. Results for lung patients were mostly aggregated together with other disease sites, however it was reported that more detailed statistical analysis for treatments occurring in the first six months of the study revealed a mean point dose difference of 1.1% ±6.3% over 62 beams delivered to thorax patients. An action level of 7.5% was used for this disease site and the authors suggested that IMRT fields applied to thorax patients were problematic due to heterogeneity and complexity of setup. In 2007, Nijsten et al. presented the results of a large, clinical implementation of a single-point patient dose reconstruction using EPID transit images acquired by a CCD-based imager (Theraview-NT EPID, Cablon Medical, Leusden, The Netherlands) [44]. The patient dose at a depth of 5 cm (D5cm) from beam entrance was compared to the treatment planning system prediction. Of the 3701 patient fractions (over multiple treatment sites) where in vivo data were estimated, 587 fractions (2257 fields) were delivered to lung cancer patients via 3DCRT. Considering measured fields for the lung patient group, a mean dose difference of 5.4 ± 5.9% was found at D5cm. The action level for this patient group was determined to be +7/ 13% based on the observed data distribution. Non-patient related errors were found in 39.1% of fractions (including 2.9% acquisition errors, 12.3% user errors, 13.8% implementation errors, and 10.1% procedure limitations), while patient-related errors were found in 20.6% of fractions. The patient-related errors were attributed to lung motion (15%) and proximity of the calculation point to the lung/tissue interface (6%). In another study, integrated transit planar dosimetry was applied to 460 lung cancer patients treated with VMAT applied in one or two arcs [30]. The EPIDs used were the modern amorphous-silicon style. Anatomical changes observed in the daily on-treatment cone-beam CT data (CBCT) were used to flag treatments. These flagged treatments then had the treatment plan recalculated on the CBCT data set, and a clinical decision was made (i.e. whether to adapt the treatment plan or not). The integrated transit planar dosimetry results were compared to the clinical decision. The authors found that there was not a significant correlation between the integrated transit planar dosimetry and the DVH metrics. They did not expect different clinical action levels to influence the results. Piermattei et al. modified their previously developed single point dose reconstruction method to account for patient density heterogeneity in lung patients by introducing a ratio of TMRs to obtain an isocentre dose estimate at an equivalent water depth

[59]. The method was applied to twenty 3DCRT lung patients from five different treatment centres (720 fractions). In three of the five centres, a transit dose estimate within each field was obtained by positioning a small ion chamber distal to the patient, while in the remaining two centres this transit dose was estimated by averaging the 25 central pixels of the acquired EPID image. A tolerance of ±6% for each beam was applied to the reconstructed patient isocentre dose. If a fraction was found out of tolerance, the result was investigated for non-patient related errors. If those were ruled out, the patient received another CT scan, to confirm the dosimetry result was due to a change in patient anatomy. In 60% of patients, the in vivo dosimetry program indicated a problem that was confirmed by the follow-up CT scan. In 2016, this group explored adding a 2D gamma analysis of integrated EPID transmission images (comparing the current fraction to the first fraction) to supplement the isocentre in vivo dose estimate in detecting dosimetry changes, although applied to head and neck patients and not yet lung patients [60]. This group has recently presented results from several years of clinical experience with their single point dose reconstruction method at one centre, employing the EPID as the transit dosimeter [61]. In 2015 they presented data from 1287 3DCRT patients, including 120 lung patients (405 fractions). The in vivo dosimetry measurement was made for the first three fractions and then weekly thereafter. Again, the tolerance for lung treatments was ±6% for each beam. For the lung patients, approximately 10% of fractions (41/405) exceeded the tolerance and were investigated further. Of these, 30 fractions were found to have planning or delivery errors (i.e. patient setup or immobilization issues), 9 fractions had morphological patient changes (tumour regression or atelectasis changes), and 2 fractions flagged due to procedural limitations. Mans et al. [62] reported data from over four years (2005–2009) of a large-scale clinical implementation of an EPID-based in vivo dosimetry program at the Antoni van Leeuwenhoek hospital. Of the 4337 patients reported on, 454 were for lung cancer treatments (the program was clinically implemented for lung patients in January of 2008). The lung treatments were delivered via step-and-shoot IMRT (30–40 segments total). For one of the first three fractions of the treatment course, the mid-plane patient dose was estimated for each beam and compared to the treatment planning system dose in that plane. Two errors were detected over the 454 lung cancer patients, both attributed to atelectasis changes. The authors emphasized that these errors (together with several other errors in non-lung sites) would not have been detected using only pre-treatment quality assurance methods.

3.3.2. SBRT lung Wendling et al. [51] retrospectively analysed 751 static-field IMRT lung patients and 50 VMAT lung patients. Although not explicitly stated, it is assumed that the 50 VMAT lung patients were all hypofractionated lung SBRT. The ‘in aqua vivo’ method

Please cite this article in press as: McCurdy , McCowan . In vivo dosimetry for lung radiotherapy including SBRT. Phys. Med. (2017), http://dx.doi.org/ 10.1016/j.ejmp.2017.05.065

6

B.M.C. McCurdy, P.M. McCowan / Physica Medica xxx (2017) xxx–xxx

was used, thus the mean dose difference at isocentre between the in aqua vivo estimate and the treatment planning system homogeneous dose calculation was compared and found to be 0.3 ± 3.1% for the IMRT lung patients and 0.2 ± 3.0% for the VMAT lung patients. Since this paper focused mainly on introducing the in aqua vivo dosimetry method, a thorough analysis of the errors discovered was not presented, although an example was given that showed a change in atelectasis. In 2015, Mijnheer et al. [63] presented a summary of recent results from their large-scale clinically-implemented in vivo dosimetry program, covering three years (2012–2014) and 15,076 treatments. For all lung plans the in aqua vivo method of dose reconstruction was employed and compared to the treatment planning system dose recalculated in a homogeneous water-equivalent patient model. The alert criteria used dose to the reference point (>3% dose difference) as well as 3D gamma analysis (3%/3 mm) within the 50% isodose line, including: mean gamma >0.5, gamma pass rate <85%, and the maximum gamma value (i.e. for highest 1% of voxels) >2. Of the 1785 lung plans analysed in this manner, 610 (or 34%) generated an alert, which was similar to the overall average of 31% for all sites combined. Although the total number of lung SBRT patients analysed over the entire three year period was not provided in that publication, more detailed data for the year 2013 was given, which indicated 4879 plans were measured for all sites, of which 570 were for lung patients, and of those 160 were hypofractionated lung patients. For the 160 hypofractionated lung cases, 58 (or 36%) generated an alert. The causes of these alerts were identified as: 17 (11%) due to model issues, 31 (19%) due to patient anatomy changes or patient setup issues, none due to image acquisition or software misuse, and 10 (6%) due to unidentified issues. Considering all 1785 lung patients over the entire three year period, 9 serious issues were identified that resulted in corrective action. The authors concluded that a large-scale clinical implementation of a 3D in vivo dosimetry program was a powerful instrument to ensure quality and safety in radiation therapy. In 2017, the CancerCare Manitoba group presented clinical results for 3D in vivo dosimetry estimated on the patient CT simulation data set for all SBRT patients over a 2.5 year period (December 2013–July 2016) [64]. Of this patient group, 100 were lung SBRT patients. Out of 537 fractions delivered for this group, 484 were available to analyse (i.e. did not experience an image acquisition problem). Reconstructed in vivo 3D dose was compared to the treatment planning system dose calculated with Acuros XB (Varian Medical Systems) using the 85% gamma pass rate tolerance similar to Mijnheer et al. [63]. After the frame averaging for EPID image acquisition was optimized, ensuring <0.5% error contribution to the reconstructed dosimetry from that step, the number of out-of-tolerance fractions was 8.0% (20/250 fractions) for the most recent 54 lung SBRT patients. Of these, 12 (4.8%) were due to patient anatomy changes or setup issues, and 8 (3.2%) were due to patient dose algorithm differences.

4. Summary Currently there are very few publications providing large-scale clinical experiences from in vivo dosimetry programs specifically applied to lung SBRT. Single point dose measurements using externally placed dosimeters are impractical for lung SBRT, and therefore much development effort has been spent on using the EPID as a transit dosimeter due to its attractive dosimetric characteristics, real-time digital imaging capabilities, and setup convenience. Several groups have developed methods to predict EPID images or water-equivalent dose images at the EPID plane, or developed methods to relate the measured EPID transit signal to dose within

the patient (in 0D, 2D or 3D). Many of these rely on semi-empirical methods and one utilizes model-based methods. In vivo dosimetry is a powerful tool that is much more useful than pre-treatment quality assurance techniques alone, catching errors that would be otherwise missed. For complex, hypofractionated treatment methods such as lung SBRT, in vivo dosimetry becomes even more valuable. From the limited published clinical data specific to lung and SBRT lung treatments, in vivo dosimetry methods are shown to be sensitive to a variety of errors, useful for plan adaptation, and feasible to implement on a large-scale. 5. Outlook and future directions For routine, large-scale clinical implementation, effort must be made to automate as much of the data collection and analysis as possible, in order to reduce the impact on resources. Appropriate clinical action levels must be developed to flag those treatments that actually merit a corrective response. Clinical workflows that promote efficiency of data flow as well as decision-making, need to be refined. Similarly to adaptive radiotherapy developments, the frequency of performing in vivo dosimetry during a given treatment course can be optimized, weighing patient benefit against cost of implementation. The importance of EPID-based in vivo dosimetry will continue to increase as new and more complicated treatment technologies are more routinely implemented, such as non-coplanar arc geometries, gated beam delivery, and real-time MLC tracking methods. As in vivo dosimetry methods steadily improve and move from the research environment to widespread commercial availability, clinical experience with in vivo dosimetry applied to lung SBRT (and all disease sites) will grow rapidly. This enormous amount of new information could be used in a variety of innovative ways to improve patient treatment. For example, the ability to accurately verify delivered patient dose in three dimensions for hundreds of thousands of patients treated annually worldwide could lead to more reliable and detailed dose response data for tumours and organs at risk. Advanced machine learning techniques might be implemented to help correlate longitudinal diagnostic information with delivered dose and treatment outcomes. These approaches could offer further personalized treatment management options as well as improved patient outcomes. The medical physicist will continue to play a key role ensuring the safe and effective implementation of in vivo dosimetry methods into modern radiation oncology practice. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Acknowledgements The authors thank the many researchers in the area of EPID dosimetry and in vivo EPID dosimetry for numerous interesting discussions over the years. References [1] Radiotherapy Risk Profile Technical Manual, World Health Organization, WHO/ IER/PSP/2008.12 Geneva (Switzerland); 2008. [2] Development of procedures for in vivo dosimetry in radiotherapy. IAEA Human Health Report No. 8. International Atomic Energy Agency, IAEA, Vienna (Austria); 2013. [3] Hartford AC, Palisca MG, Eichler TJ, et al. American society for therapeutic radiology and oncology (ASTRO) and american college of radiology (ACR) practice guidelines for intensity-modulated radiation therapy (IMRT). Int J Radiat Oncol Biol Phys 2009;73:9–14.

Please cite this article in press as: McCurdy , McCowan . In vivo dosimetry for lung radiotherapy including SBRT. Phys. Med. (2017), http://dx.doi.org/ 10.1016/j.ejmp.2017.05.065

B.M.C. McCurdy, P.M. McCowan / Physica Medica xxx (2017) xxx–xxx [4] Derreumaux S, Etard C, Huet C, Trompier F, Clairand I, Bottollier-Depois J, et al. Lessons from recent accidents in radiation therapy in France. Radiat Prot Dosim 2008;131:130–5. [5] The Royal College of Radiologists, Society and College of Radiographers, Institute of Physics and Engineering in Medicine, British Institute of Radiology. Implementing In Vivo Dosimetry. London: Royal College of Radiologists; 2008. [6] SSI (The Swedish Radiation Protection Institute). Regulations on radiation therapy Report SSI FS 4 (Stockholm. Sweden: SSI); 2000. [7] Mijnheer B, Beddar S, Izewska J, Reft C. In vivo dosimetry in external beam radiotherapy. Med Phys 2013;40. 070903-1-20. [8] Ford EC, Terezakis S, Souranis A, et al. Quality control quantification (QCQ): A tool to measure the value of quality control checks in radiation oncology. Int J Radiat Oncol Biol Phys 2012;84:e263–9. [9] Kron T, Lehmann J, Greer PB. Dosimetry of ionising radiation in modern radiation oncology. Phys Med Biol 2016;61:R167–205. [10] van Elmpt WJ, Nijsten SM, Schiffeleers RF, Dekker AL, Mijnheer BJ, Lambin P, et al. A Monte Carlo based three-dimensional dose reconstruction method derived from portal dose images. Med Phys 2006;33(7):2426–34. [11] van Elmpt WJ, Nijsten SM, Dekker AL, Mijnheer BJ, Lambin P. Treatment verification in the presence of inhomogeneities using EPID-based threedimensional dose reconstruction. Med Phys 2007;34(7):2816–26. [12] Fippel M. Fast Monte Carlo dose calculation for photon beams based on the VMC electron algorithm. Med Phys 1999;26(8):1466–75. [13] van Elmpt W, Petit S, De Ruysscher D, Lambin P, Dekker A. 3D dose delivery verification using repeated cone-beam imaging and EPID dosimetry for stereotactic body radiotherapy of non-small cell lung cancer. Radiother Oncol 2010;94(2):188–94. [14] Persoon LCGG, Egelmeer AGTM, Ollers MC, et al. First clinical results of adaptive radiotherapy based on 3D portal dosimetry for lung cancer patients with atelectasis treated with volumetric-modulated arc therapy (VMAT). Acta Oncol 2013;52:1484–9. [15] Berbeco RI, Hacker F, Zatwarnicki C, Park SJ, Ionascu D, O’Farrell D, et al. A novel method for estimating SBRT delivered dose with beam’s-eye-view images. Med Phys 2008;35:3225–31. [16] Aristophanous M, Rottmann J, Court LE, Berbeco RI. EPID-guided 3D dose verification of lung SBRT. Med Phys 2011;38:495–503. [17] Cai W, Hurwitz MH, Williams CL, Dhou S, Berbeco RI, Seco J, et al. 3D delivered dose assessment using a 4DCT-based motion model. Med Phys 2015;42:2897–907. [18] Hurwitz M, Williams CL, Mishra P, Rottmann J, Dhou S, Wagar M, et al. Generation of fluoroscopic 3D images with a respiratory motion model based on an external surrogate signal. Phys Med Biol 2015;60(2):521–35. [19] Lin M-H, Li J, Wang L, Koren S, Fan J, Forkal E, et al. 4D patient dose reconstruction using online measured EPID cine images for lung SBRT treatment validation. Med Phys 2012;39:5949–58. [20] Gimeno J, Pujades MC, García T, Carmona V, Lliso F, Palomo R, et al. Commissioning and initial experience with a commercial software for in vivo volumetric dosimetry. Phys Med 2014;30:954–9. [21] Matrosic C, Culberson W, Rosen B, Madsen E, Frank G, Bednarz B. Initial characterization of a gel patch dosimeter for in vivo dosimetry. Phys Med Biol 2016;61:N240–8. [22] McCurdy BMC. Dosimetry in radiotherapy using a-Si EPIDs:Systems, methods, and applications focusing on 3D patient dose estimation. J Phys: Conf Ser 2013;444:012002. [23] McCurdy BMC, Greer P, Bedford J. Electronic Portal Imaging Device (EPID) dosimetry. Clinical 3D dosimetry in radiation therapy. In: Mijnheer B, editor. 1st ed. Abingden (England): Taylor & Francis Books Inc; 2017. [24] van Elmpt W, McDermott L, Nijsten S, et al. A literature review of electronic portal imaging for radiotherapy dosimetry. Radiother Oncol 2008;88:289–309. [25] Kroonwijk M, Pasma KL, Quint S, Koper PC, Visser AG, Heijmen BJ. In vivo dosimetry for prostate cancer patients using an electronic portal imaging device (EPID); demonstration of internal organ motion. Radiother Oncol 1998;49(2):125–32. [26] van Elmpt WJC, Nijsten SMJJG, Mijnheer BJ, Minken AWH. Experimental verification of a portal dose prediction model. Med Phys 2005;32(9):2805–18. [27] Pasma KL, Heijmen BJM, Kroonwijk M, Visser AG. Portal dose image (PDI) prediction for dosimetric treatment verification in radiotherapy I: An algorithm for open beams. Med Phys 1998;25:830–40. [28] Pasma KL, Kroonwijk M, van Dieren EB, Visser AG, Heijmen BJM. Verification of compensator thicknesses using a fluoroscopic electronic portal imaging device. Med Phys 1999;26:1524–9. [29] Persoon LCGG, Nijsten SMJJG, Wilbrink FJ, Podesta M, Snaith JAD, Lustberg T, et al. Interfractional trend analysis of dose differences based on 2D transit portal dosimetry. Phys Med Biol 2012;57:6445–58. [30] Persoon LCGG, Podesta M, Hoffmann L, Sanizadeh A, Schyns LEJR, de Ruiter BM, et al. Is integrated transit planar portal dosimetry able to detect geometric changes in lung cancer patients treated with volumetric modulated arc therapy? Acta Oncol 2015;54:1501–7. [31] Persoon LCGG, Podesta M, Nijsten SMJJG, Troost EGC, Verhaegen F, Timeresolved versus integrated transit planar dosimetry for volumetric modulated arc therapy: patient-specific dose differences during treatment, a proof of principle technology in cancer research & treatment, 2016; 15(6):NP79–NP87. [32] Schyns LE, Persoon LC, Podesta M, van Elmpt WJ, Verhaegen F. Time-resolved versus time-integrated portal dosimetry: the role of an object’s position with respect to the isocenter in volumetric modulated arc therapy. Phys Med Biol 2016;61(10):3969–84.

7

[33] van Esch A, Depuydt T, Huyskens DP. The use of an aSi-based EPID for routine absolute dosimetric pre-treatment verification of dynamic IMRT fields. Radiother Oncol 2004;71:223–34. [34] Berry SL, Sheu R-D, Polvorosa CS, Wuu C-S. Implementation of EPID transit dosimetry based on a through-air dosimetry algorithm. Med Phys 2012;39:87–98. [35] Berry SL, Polvorosa C, Cheng S, Deutsch I, Chao KSC, Wuu C-S. Initial clinical experience performing patient treatment verification with an electronic portal imaging device transit dosimeter. Int J Radiation Oncol Biol Phys 2014;88(1):204–9. [36] Fuangrod T, Woodruff HC, van Uytven E, McCurdy BM, Kuncic Z, O’Connor DJ, et al. A system for EPID-based real-time treatment delivery verification during dynamic IMRT treatment. Med Phys 2013;40(9):091907. [37] Woodruff HC, Fuangrod T, Van Uytven E, McCurdy BM, van Beek T, Bhatia S, et al. First experience with real-time EPID-based delivery verification during IMRT and VMAT sessions. Int J Radiat Oncol Biol Phys 2015;93(3): 516–22. [38] Chytyk-Praznik K, VanUytven E, vanBeek TA, Greer PB, McCurdy BM. Modelbased prediction of portal dose images during patient treatment. Med Phys 2013;40(3):031713. [39] McCurdy BM, Pistorius S. A two-step algorithm for predicting portal dose images in arbitrary detectors. Med Phys 2000;27(9):2109–16. [40] McCurdy BM, Luchka K, Pistorius S. Dosimetric investigation and portal dose image prediction using an amorphous silicon electronic portal imaging device. Med Phys 2001;28(6):911–24. [41] Piermattei A, Fidanzio A, Stimato G, Azario L, Grimaldi L, D’Onofrio G, et al. In vivo dosimetry by an aSi-based EPID. Med Phys 2006;33(11):4414–22. [42] Piermattei A, Fidanzio A, Azario L, Grimaldi L, D’Onofrio G, Cilla S, et al. Application of a practical method for the isocenter point in vivo dosimetry by a transit signal. Phys Med Biol 2007;52(16):5101–17. [43] François P, Boissard P, Berger L, Mazal A. In vivo dose verification from backprojection of a transit dose measurement on the central axis of photon beams. Phys Med 2011;27:1–10. [44] Nijsten SMJJG, Mijnheer BJ, Dekker ALAJ, et al. Routine individualised patient dosimetry using electronic portal imaging devices. Radiother Oncol 2007;83:65–75. [45] Boellaard R, van Herk M, Uiterwaal H, Mijnheer B. First clinical tests using a liquid-filled electronic portal imaging device and a convolution model for the verification of the midplane dose. Radiother Oncol 1998;47(3):303–12. [46] Boellaard R, Essers M, van Herk M, Mijnheer BJ. New method to obtain the midplane dose using portal in vivo dosimetry. Int J Radiat Oncol Biol Phys 1998;41(2):465–74. [47] Boellaard R, van Herk M, Uiterwaal H, Mijnheer B. Two-dimensional exit dosimetry using a liquid-filled electronic portal imaging device and a convolution model. Radiother Oncol 1997;44(2):149–57. [48] Boellaard R, van Herk M, Mijnheer BJ. A convolution model to convert transmission dose images to exit dose distributions. Med Phys 1997;24(2):189–99. [49] Wendling M, Louwe RJ, McDermott LN, Sonke JJ, van Herk M, Mijnheer BJ. Accurate two-dimensional IMRT verification using a back-projection EPID dosimetry method. Med Phys 2006;33(2):259–73. [50] McDermott LN, Wendling M, Sonke JJ, van Herk M, Mijnheer BJ. Replacing pretreatment verification with in vivo EPID dosimetry for prostate IMRT. Int J Radiat Oncol Biol Phys 2007;67(5):1568–77. [51] Wendling M, McDermott LN, Mans A, et al. In aqua vivo EPID dosimetry. Med Phys 2012;39:367–77. [52] Wendling M, McDermott LN, Mans A, et al. A simple backprojection algorithm for 3D in vivo EPID dosimetry of IMRT treatments. Med Phys 2009;36:3310–21. [53] van Uytven E, van Beek T, McCowan PM, Chytyk-Praznik K, Greer PB, McCurdy BM. Validation of a method for in vivo 3D dose reconstruction for IMRT and VMAT treatments using on-treatment EPID images and a model-based forward-calculation algorithm. Med Phys 2015;42(12):6945–54. [54] McNutt TR, Mackie TR, Reckwerdt P, Paliwal BR. Modeling dose distributions from portal dose images using the convolution/superposition method. Med Phys 1996;23(8):1381–92. [55] Ahnesjö A. Collapsed cone convolution of radiant energy for photon dose calculation in heterogeneous media. Med Phys 1989;16(4):577–92. [56] McCowan PM, van Uytven E, van Beek T, Asuni G, McCurdy BMC. An in vivo dose verification method for SBRT–VMAT delivery using the EPID. Med Phys 2015;42:6955–63. [57] Mans A, Remeijer P, Olaciregui-Ruiz I, et al. 3D Dosimetric verification of volumetric-modulated arc therapy by portal dosimetry. Radiother Oncol 2010;94:181–7. [58] Celi S, Costa E, Wessels C, Mazal A, Fourquet A, François P. EPID based in vivo dosimetry system: clinical experience and results. J App Clin Med Phys 2016;17(3):262–76. [59] Piermattei A, Cilla S, Grimaldi L, Sabatino D, Fidanzio A, Greco F, et al. Integration between in vivo dosimetry and image guided radiotherapy for lung tumors. Med Phys 2009;36:2206–14. [60] Meluccio D, Fidanzio A, Azario L, Ianiro A, Macchia G, Digesù C, et al. Initial clinical experience with Epid-based in-vivo dosimetry for VMAT treatments of head-and-neck tumors. Phys Med 2016;32(1):52–8. [61] Fidanzio A, Azario L, Greco F, Cilla S, Piermattei A. Routine EPID in-vivo dosimetry in a reference point for conformal radiotherapy treatments. Phys Med Biol 2015;60:N141–50.

Please cite this article in press as: McCurdy , McCowan . In vivo dosimetry for lung radiotherapy including SBRT. Phys. Med. (2017), http://dx.doi.org/ 10.1016/j.ejmp.2017.05.065

8

B.M.C. McCurdy, P.M. McCowan / Physica Medica xxx (2017) xxx–xxx

[62] Mans A, Wendling M, McDermott LN, Sonke J-J, Tielenburg R, Vijlbrief R, et al. Catching errors with in vivo EPID dosimetry. Med Phys 2010;37:2638–44. [63] Mijnheer BJ, González P, Olaciregui-Ruiz I, Rozendaal RA, van Herk M, Mans A. Overview of 3-year experience with large-scale electronic portal imaging device–based 3-dimensional transit dosimetry. Pract Radiat Oncol 2015;5:e679–87.

[64] McCowan PM, Asuni G, van Beek T, van Uytven E, McCurdy BMC, Loewen SK, et al. Clinical implementation of a model-based in vivo dose verification system for stereotactic body radiation therapyevolumetric modulated arc therapy treatments using the electronic portal imaging device. Int J Radiat Oncol Biol Phys. 2017;97(5):1077–84.

Please cite this article in press as: McCurdy , McCowan . In vivo dosimetry for lung radiotherapy including SBRT. Phys. Med. (2017), http://dx.doi.org/ 10.1016/j.ejmp.2017.05.065