New developments in radiotherapy quality assurance Tuesday 7th November 2017, County Hotel, Newcastle-upon-Tyne

FINAL PROGRAMME

09:00 – 09:30 Coffee and registration

Session 1: New approaches to QA Chair: Steve Weston

09:30 – 09:45 Introduction

09:45 – 10:30 IPEM 81 Update – Physics aspects of QC in RT

Invited Speaker: Imran Patel, The Christie NHS Foundation Trust

10:30 – 10:45 Novel QA in the era of frameless intracranial stereotactic radiosurgery with the new Icon™-model Gamma Knife® Gavin Wright, St. James’s University Hospital, Leeds

10:45 – 11:00 Clinical Impact of Treatment Machine Daily Output Variation

Matthew Bolt, Royal Surrey County Hospital

11:00 – 11:15 Quality Assurance of FFF beams: are Flattened Beam Methods and Equipment Transferrable?

Natalie Thorp, Lancashire Teaching Hospitals NHS Foundation Trust

11:15 – 11:45 Coffee

Session 2: VMAT QA Chair: Linda Carruthers

11.45 – 12.00 In-vivo EPID dosimetry for IMRT treatments based on Varian through-air predicted portal dose algorithm at fixed source to imager distance Maan Najem, The Royal Marsden NHS Foundation Trust

12.00 – 12.15 Statistical analysis of VMAT plan complexity using MCS for three anatomical sites

Polly E Darby, Aberdeen Royal Infirmary

12.15 – 12.30 Developing a new Patient Specific QA (PSQA) technique for the verification of Stereotactic Radiosurgery (SRS) treatments Anna Bangiri, Nottingham University Hospitals NHS Trust

12.30 – 12.45 Questions and Discussion

12.45 – 13.45 Lunch

Session 3: QA Management Chair: John Byrne

13.45 – 14.30 TG100 – Risk based RT Quality Management

Invited speaker: Prof Peter Dunscombe, Alberta Health Services, Calgary

14.30 – 14.45 The Benefits of a Network Wide Approach to Linac Quality Control Rob Brackenridge, Carlisle Infirmary

14.45 – 15.00 A review of the implementation and use of QATrack+, an open source QC database for Radiotherapy Clinics

Thomas Burrows, HCA Healthcare UK

15.00 – 15.15 QATrack+ implementation as evolution from paper-based to paper-light

Rollo Moore, Royal Marsden NHS Foundation Trust

15.15 – 15.30 Comparison of Mobius Doselab Pro with Elekta ACAL for Linac QA using EPID images

Dr Ruth Harding, Singleton Hospital, Swansea

15.30 – 16.00 Coffee

Session 4: MR & EPID Use Chair: Graham Freestone

16.00 – 16.15 Improving the efficiency and accuracy of radiotherapy MR quality assurance in preparation for MR-only radiotherapy planning Jonathan Wyatt, Newcastle upon Tyne Hospitals NHS Foundation Trust

16.15 – 16.45 Developing a Quality Assurance Program for a High-Field MR Linac

Dr James Agnew, The Christie NHS Foundation Trust Dr Ian Hanson, Royal Marsden NHS Foundation Trust

16.45 – 17.00 EPID based In-vivo dosimetry - The ultimate QA tool?

Dr Steve Weston, Leeds Teaching Hospitals NHS Trust

17.00 – 17.15 Commissioning and experience of using Truebeam’s Machine Performance Check application

Gary Barfield, United Lincolnshire Hospitals NHS Trust

17.15 – 17.30 Prize-giving and closing remarks

Organised by IPEM’s Radiotherapy Special Interest Group

New developments in radiotherapy quality assurance Tuesday 7th November 2017, County Hotel, Newcastle-upon-Tyne

POSTERS

1 Management of Radiotherapy QA Data: Our First Experience in the Implementation of a Commercial Paperless QA System Dr Ashraf Esmail, Leeds Teaching Hospitals NHS Trust

2 Questioning the Validity of Pre-Treatment Patient Specific QA Will Holmes-Smith, Norfolk & Norwich University Hospital Foundation Trust

3 Quick assessment of mechanical isocentre Daniel Johnson, James Cook University Hospital, Middlesbrough

4 A Paperless Dosimetry QA System: three years of experience Dr Mohamed Metwaly, United Lincolnshire Hospitals NHS Trust

5 An analysis of a variety of Patient Specific QA equipment to determine which is the most suitable approach to PSQA in a new Radiotherapy Department Eoin O’Hare, Altnagalvin Hospital, Western Health & Social Care Trust

6 Presenting an in-house, automated and efficient film analysis technique for the pre-treatment verification of stereotactic radiosurgery (SRS) brain treatments Jonathan Sutton, Nottingham University Hospitals NHS Trust

7 Improvement of off-axis SABR plan verification results by using adapted dose reconstruction algorithms for the Octavius4D system Prakash Jeevanandam, Belfast Health and Social Care Trust

Organised by IPEM’s Radiotherapy Special Interest Group

Revision of IPEM Guidance on Quality Control of Radiotherapy Equipment 1Patel I, 2Weston S, 3Palmer A L, 4Mayles W P M, 5Whittard P, 4Clements R, 6Reilly A and 7Jordan T J.

1 Christie Medical Physics and Engineering, The Christie NHS Foundation Trust, Manchester, UK 2 Department of Medical Physics, St. James’s University Hospital, Leeds, UK 3 Medical Physics Department, Queen Alexandra Hospital, Portsmouth Hospitals NHS Trust, Portsmouth, UK

4 The Clatterbridge Cancer Centre NHS Foundation Trust, Bebington, Wirral, UK 5 The Beacon Centre, Musgrove Park Hospital, Taunton, UK 6 North West Cancer Centre, Western Health and Social Care Trust, Altnagelvin Hospital, Londonderry, Northern Ireland, UK.

7 Department of Medical Physics, St Luke’s Cancer Centre, Royal Surrey County Hospital NHS Foundation Trust, Guildford, Surrey, UK

Fundamental to the aim of optimum, high quality, safe radiotherapy is robust and appropriate control of all physical aspects that may influence treatment. As the professional body for medical physics in UK, the Institute of Physics and Engineering in Medicine (IPEM) are revising their basic recommendations report for physics aspects of quality in radiotherapy. To ensure treatment delivery is as proposed a quality assurance system in radiotherapy is essential. A quality control (QC) measurement program is a crucial component within a quality assurance system. Guidance on QC of radiotherapy equipment was previously published by IPEM in 1999 in the form of Report 81. IPEM Report 81 was primarily based on results of a survey of UK QC practice that was undertaken in 1991. Since this publication dramatic technological advances in radiotherapy equipment and increasingly complex clinical use required this guidance report to be reviewed and updated. To this end a working party was approved by IPEM and setup in June 2012 with eight expert members from the UK chosen to be on the editorial board. Several leading individuals within the field were approached to update report 81 content and to submit new subject matter that could be included in the publication of the revised guidance. The progress made by the working party and the proposed changes to the updated report will be considered.

Novel QA in the era of frameless intracranial stereotactic radiosurgery with the new Icon™-model Gamma Knife®. 1Wright G, 1Harrold N, 1Nix M, 1Bownes P 1Dept. Medical Physics & Engineering, St. James’s University Hospital, Leeds, UK.

Background The new Icon™-model Gamma Knife® (GK) allows for the first time GK stereotactic radiosurgery (SRS) to be performed in a mask as an alternative to conventional frame-based immobilisation, thus facilitating fractionated delivery [1,2]. Icon™’s novel integral cone-beam CT (CBCT) provides stereotactic localisation and verification of position (previously guaranteed with frame immobilisation) and subsequent coregistration-based plan adaptation for each fraction. Intrafraction motion (previously eliminated with frame immobilisation) is managed through treatment gating, based upon tracking of patient markers by an integral stereoscopic camera known as the HDMM (high-definition motion management) system. The novel workflows facilitated by Icon™’s CBCT and HDMM systems necessitate equally-novel QA procedures [2,3]. As the fourth centre in the world to adopt Icon™, we discuss the QA protocol accordingly developed at our centre, and present results from our first 21-months experience of Icon™ QA.

Methods CBCT stereotactic (STX) localisation is verified pre-treatment using a vendor-supplied tool. An in-house manufactured known-target phantom used bi-monthly provides more complete assessment of localisation accuracy. Accuracy of STX coordinates imposed onto MR images via co-registration against Icon™ CBCT was verified at commissioning using the commercial PTGR known target phantom. Geometric accuracy of plan adaptation is verified bi-monthly via a full system test using pin-marked radiochromic film, initially inside the vendor’s calibration phantom and subsequently inside an in-house substitute slice for the commercial Rando phantom. Functioning of the HDMM gating trigger is verified weekly using a simplistic hydraulic tool allowing remote displacement of an HDMM marker from outside the treatment room. Accuracy of HDMM-reported displacements is verified bi-monthly using an in-house tool to induce known sub-millimetre displacements along three orthogonal axes.

Results The vendor’s test shows the maximum STX error in the CBCT volume is 0.10(0.04)mm mean(SD) (n=397). In-house known target phantom shows max(mean) STX error of -0.05(0.40), -0.05(-0.55) and -0.27(-0.70) [mm] in x, y and z, respectively. CBCT coregistration-based STX definition of MR images is accurate to within 0.83(0.47)mm max(mean), as compared to 0.91(0.50)mm typically achieved with conventional fiducial-based MRI STX definition. Geometric accuracy of CBCT coregistration-based plan adaptation is within 0.31(0.26)mm max(mean), compared to inherent couch positional accuracy of <0.4mm. HDMM-indicated marker displacements agree within 0.02(0.05), 0.01(0.05) and 0.01(0.05)mm mean(SD) of known displacements in LR, AP and SI directions, respectively, and HDMM-triggered gating functions without any failures.

Discussion Geometric accuracy is crucial to the safety of intracranial SRS. Novel QA of Icon™ can ensure confidence that the geometric accuracy of GK SRS - well-established with frame immobilisation - is maintained with mask immobilisation. Our results show the accuracy of STX localisation of Icon™ CBCT is well within greater uncertainties inherent to SRS as a whole (e.g. voxel size of planning MR images, typically ~1mm), and that imposition of STX coordinates onto MR image via coregistraion against Icon™ CBCT is of an accuracy comparable to that conventionally achieved with the fiducial markers of frame-based SRS. Plan adaption with CBCT co-registration is accurate within vendor-specified tolerance for couch positional accuracy. The HDMM provides reliable tracking of patient position with sub-millimetre accuracy, and associated HDMM-triggered treatment gating operates safely and reliably.

Conclusion Novel QA of Icon™’s integral CBCT and HDMM systems provide confidence that mask-based GK SRS achieves a geometric accuracy comparable to that typical of the well-established frame-based GK SRS.

References [1] Wright G, Harrold N, et al. (2017), Jour. of Radiosurgery and SBRT 4(4) 289-301 [2] Blake S, Winch L, Appleby H. (2016), Radiotherapy and Oncology, 119: S921. [3] Zeverino M, Jaccard M et al. (2017) Med. Phys. 44 (2) 355-363

Clinical Impact of Treatment Machine Daily Output Variation 1,3,4Bolt M, 1,2Nisbet A, 1,4Clark C, 3Chen T 1Department of Medical Physics, Royal Surrey County Hospital, Guildford, UK. 2Department of Physics, University of Surrey, UK. 3Department of Chemical and Process Engineering, University of Surrey, UK 4National Physical Laboratory, Teddington, UK

Background Within the UK dose delivered by Megavoltage radiotherapy treatment machines is directly traceable to the NPL primary standard through use of the published codes of practice [3] and audits are regularly performed to ensure tight control on beam calibration [2]. Beam output measurements are taken routinely (usually daily) on all treatment machines to ensure the variation in patient dose due to changes in beam output are maintained within a locally set tolerance [4,5]. However these routine measurements are seldom considered within the accuracy of radiotherapy treatments, either routinely or as part of a clinical trial, and yet there may be significant variation in delivered dose between treatment machines [1].

Methods A UK wide request was made for 6MV beam output data covering the period January 2015 – June 2015 through the IPEM interdepartmental audit groups. A statistical analysis of this data was performed including assessing the mean beam output for each machine and the measurement variation over time. Measurement variation nationally and locally has also been quantified. The dose variation due to beam output alone has been incorporated in the clinical outcome modelling using the linear quadratic model for prostate and head and neck cancers. Using this model an estimate of the impact of beam output variation on clinical outcomes is given. Analysis of DVHs from the PARSPORT trial has been used to put the output dose variation in context with dose variation from different treatment planning techniques.

Results Data was received from 204 treatment machines across 52 sites. The mean beam output varied between -2.1% and +1.6% between individual machines. Within a single site the maximum range between treatment machines was 2.2% (-0.7% to 1.5%). Radiobiological modelling indicates that for individual patients this variation may lead to a variation in TCP of up to 10% for head and neck cancers, and 2.5% for prostate cancer. The change from conformal to IMRT planning techniques resulted in a significant reduction in dose variation to the PTV with the standard deviation in mean dose reducing by 90%.

Discussion With the increase in IMRT over recent years, the importance of maintaining machine beam output within strict limits has become more significant within the overall uncertainty. The most commonly used tolerance on beam output is +/-2% and it was surprising to find a treatment machine for which the mean measured beam output was outside this tolerance over a 6 month period. With treatment planning techniques only becoming more conformal and precise, the accuracy of the beam output becomes increasingly significant, with the possibility of significant variation in outcomes associated with beam output alone as demonstrated.

Conclusion The control of beam output is critical within the overall uncertainty in radiotherapy dose delivery and is becoming increasingly important as technology and treatment techniques advance. This is of particular importance in cancers which exhibit a steep dose response gradient.

Key references

[1] Bolt, M. et al. Radiotherapy and Oncology, 2017; 123: S399. [2] Clark, C. et al. Br. J. Radiol., 2015; 88: 1055. [3] Lillicrap, S. et al. Phys. Med. Biol., 1990; 35:1355-60. [4] Mayles, W. et al. Physics aspects of Quality control in radiotherapy, 1999, IPEM. [5] Palmer, A, et al. British Journal of Radiology, 2012; 85: e1067-73.

Quality Assurance of FFF beams: are Flattened Beam Methods and Equipment Transferrable? Natalie Thorp, Lancashire Teaching Hospitals NHS Foundation Trust

Aims and Background: The clinical merits of unflattened high dose rate (FFF) beams relative to flattened beams have been well documented by many authors. The aim of this study was to measure the responsiveness of a range of dosimetric equipment to FFF beams, especially under sub-optimum beam conditions. Conventional tests to ensure dosimetric accuracy of new treatment techniques, such as variable dose rate VMAT and small field high dose SABR, were also assessed to establish their validity for the QA of unflattened high dose rate beams. Method: The department’s dosimetry equipment was assessed for suitability of use in a 6MV FFF beam on an Elekta Versa linac. The equipment included ion chambers with a range of volumes, PTW’s Linacheck, Quickcheck (with in-house bespoke MCP alloy filter) and Starcheck (with FFF normalisation as proposed by Fogliata), Scandidos’ Delta 4, IBA’s blue water phantom and an EPID imager. Subtle changes were made to the beam characteristics (such as output, dose rate, symmetry and beam energy) were changed in service mode and the corresponding responsiveness of the dosimetry equipment was measured. Linac performance was assessed by measuring dose linearity with MU, symmetry variability with dose rate, and isocentre accuracy with gantry angle, using standard QA methods. Results: Ion chambers picked up small changes in beam output and were used to show that the quality index was not a reliable measure of beam energy change. A CC13 exhibited a 3% change in Pion over a range Dmax to D30 for an 80SSD depth dose. Profiles taken in a large water tank showed that changes in symmetry and energy were constant with depth. A ±4% change in local dose at the 12pts corresponded to a ±2.5mm shift in depth dose at 10cm deep at 100SSD. Profile measurements with the Starcheck 2D array and the Quickcheck device showed that they were linear, dose rate independent and that they correctly indicated changes to symmetry and energy. The Starcheck array allowed the user to assess the central axis position with gantry angle, whereas the Winston Lutz test, analysed using Elekta kV flexmap software, significantly overestimated the locus of the isocentre proving itself unsuitable for FFF beams. Conclusion: A sliding Pion correction factor must be applied if large chambers are used for output and depth dose measurements. The Starcheck can be used to measure beam symmetry and energy and can also be used to assess changes in beam profile with dose rate and servo performance. Measurements of output with MU, and profiles with dose rate, can enable the user to identify minimum limits on MU and dose rate to ensure optimum beam performance in the clinical setting.

Independent three-dimensional dose calculation software for volumetrically modulated arc therapy treatment plans Christopher S. Rose, Barry Evans and Simon J. Thomas Medical Physics Department, Addenbrooke’s Hospital, Cambridge CB2 0QQ, United Kingdom

Volumetric intensity modulated arc therapy (VMAT) is delivered under continuous variation of multi-leaf collimator (MLC) segments, Dose rate and gantry speed.(1) The dynamical freedom of VMAT delivery increases the complexity of treatment planning; and reinforces the need for machine quality assurance and independent plan verification.(2) Independent plan verification can include EPID dosimetry and Monte Carlo phase space predictions(3), phantom measurements(4) and dose verification using other planning systems. Regardless of the verification method used, the results are usually evaluated against the planned distribution using dose-distance difference indices such as the Gamma, Kappa or Box index.(5) We describe a means of performing three dimensional independent dose calculation using DICOM-RT plan objects and patient CT data. The resulting distribution is then evaluated against the DICOM dose cube exported from the planning system. The software described here uses the average dose to the 50% isodose and a dose-distance box analysis as metrics for plan comparison. Here we describe the commissioning and implementation of an independent dose calculation tool written using MATLAB (Mathworks, inc.). The tool imports a patient CT, DICOM-RT plan, and DICOM dose cube, as exported from a treatment planning system, and re-calculates an independent three-dimensional dose distribution based on the patient CT and plan information. The results of the calculation are compared to the planned dose distribution using both an average dose difference analysis (to the 50% isosdose line) and a box analysis. If the box analysis and average dose difference a fall within certain tolerances the plan is found to be suitable for treatment. If a plan fails either type of analysis it will be sent for a DQA. For the purposes of commissioning we have analysed plans for previously treated prostate and prostate bed IMRT VMAT plans, with dose per fractions of 2 -3Gy. Each of the plans had been previously treated clinically and passed quality assurance measurements on treatment machines. Each prostate plan was recalculated within the verification software, and the resulting dose distribution compared to the plan DICOM dose cube. The doses were compared in terms of percentage dose difference, as well as a dose-distance Box analysis. Several dose-distance tolerances, ranging from 2%/2mm to 4%/4mm, were employed in an effort to quantify plan analysis pass criteria. Robustness testing was also performed using plans with deliberately introduced errors. Two such plans were modified to simulate a ±5 mm variation in planned PTV margin and analysed by the software, in both cases the analysis highlighted significant differences between the planned and re-calculated distributions. The appropriate use of quality assurance software can provide significant time savings for a department, in terms of workload, and staffing requirements. Since the implementation of our in-house solution we have seen a 40% reduction in the total amount of DQA’s being performed in our department. The process of commissioning has highlighted the importance of rigorous software testing, transparent development processes, and risks identification. Several European and international standards provide recommendations relating to the development of medical software; such as IEC 62304, ISO 90003 and ISO 9001, and should be consulted where possible.(6)

1. Otto K. Volumetric modulated arc therapy: IMRT in a single gantry arc. Med Phys. 2008;35(1):310–17

2. Ezzell GA, Galvin JM, Low D, et al. Guidance on delivery, treatment planning,and clinical implementation of IMRT: report of the IMRT Subcommittee of the AAPM Radiation Therapy Committee. Med Phys. 2003;30(8):2089–115.

3. Popescu I A, et al. Patient-specific QA using 4D Monte Carlo phase space predictions and EPID dosimetry. Journal of Physics: Conf. Ser.

2015;537

4. Bedford JL, Lee YK, Wai,P, et al. Evaluation of Delta4 phantom for IMRT and VMAT verification. Phys Med Biol. 2009 May 7;54(9):N167-76

5. Thomas SJ, Cowley IR. A comparison of four indices for combining distance and dose differences. Int J Radiat Oncol Biol Phys. 2012 Apr

1;82(5):e717-23 6. Cosgriff, Philip, et al. In House Development of Medical Software: Regulations and Standards. (IPEM, York, 2016).

In-vivo EPID dosimetry for IMRT treatments based on Varian through-air predicted portal dose algorithm at fixed source to imager distance 1Najem M A , 2Tedder M, 1King D, 1Bernstein D, 1Trouncer R and 1Meehan C 1 Physics Department, The Royal Marsden NHS Foundation Trust, London, UK. 2 Medical Physics Department, Guy's and St Thomas' NHS Foundation Trust, London, UK

Background: Berry et al (2012) introduced a method to perform transit EPID dosimetry for IMRT treatment for a fixed air gap between the patient exit and EPID, based on the Varian through-air portal dose image prediction algorithm.

Purpose: To adapt the methodology Berry et al (2012) for use with IMRT treatments at a fixed source to imager distance (SID) in order to reduce the total treatment time.

Methods: A correction factor was introduced to account for the change in air gap between patient and imager in order for the SID to remain fixed. Commissioning data, consisting of multiple field sizes, solid water thicknesses and air gaps, were acquired at 150 cm SID for two different EPID systems: aS500 and aS1200.The method was verified using six IMRT prostate and seminal vesicles plans (120 fields in total) on up to three different phantoms: 19 cm thick solid water blocks, RT01 (Moore et al, 2006) and BrainLab pelvis phantoms. The predicted portal images through phantom were compared to the measured portal images using a global 3%/3mm gamma criteria. Plans and predicted portal images through air were created in the Varian Eclipse treatment planning system (TPS).

Results: 117 out 120 fields in the six IMRT plans passed the 3%/3mm gamma criteria by more than 95% on the three phantoms and both portal imagers. The average gamma pass rate for IMRT plans was 99.9% ± 0.2%(1SD), 100.0% ± 0.1% (1SD) and 99.1% ± 1.5% (1SD) on the 19 cm solid water blocks, RT01 and BrainLab pelvis phantoms, respectively.

Discussion: Using a fixed SID instead of a fixed air gap reduces the overall treatment time as the EPID does not need to be moved for each IMRT beam to maintain the same air gap. In addition, the fixed air gap approach requires a separate verification plan for each field to be created in the TPS. Our method of introducing an air gap correction factor means that only a single verification plan is needed to produce the predicted images.

Conclusion: Berry et al (2012) transit EPID dosimetry method was adapted to be used for IMRT treatments at fixed SID. A correction factor was introduced to account for the change in the air gap between the patient exit and EPID. Using a fixed SID requires a single verification plan to be created in the TPS for the transit portal images calculation in contrast with the fixed air gap approach. The current method was verified successfully for several IMRT plans on three different phantoms and two EPID systems and could be extended for in-vivo verification of VMAT treatments.

Key references: Berry, S.L., Sheu, R.D., Polvorosa, C.S. and Wuu, C.S., 2012. Implementation of EPID transit dosimetry based on a through‐air dosimetry algorithm. Medical physics, 39(1), pp.87-98. Moore, A.R., Warrington, A.J., Aird, E.G., Bidmead, A.M. and Dearnaley, D.P., 2006. A versatile phantom for quality assurance in the UK Medical Research Council (MRC) RT01 trial (ISRCTN47772397) in conformal radiotherapy for prostate cancer. Radiotherapy and oncology, 80(1), pp.82-85. Van Esch, A., Depuydt, T. and Huyskens, D.P., 2004. The use of an aSi-based EPID for routine absolute dosimetric pre-treatment verification of dynamic IMRT fields. Radiotherapy and oncology, 71(2), pp.223-234.

Statistical analysis of VMAT plan complexity using MCS for three anatomical sites 1Darby P E, 1Carnegie D J 1Radiotherapy Physics, Aberdeen Royal Infirmary, UK.

Background- Patient-specific QA for VMAT treatment plans is used to ensure that plans are deliverable under the limitations of the system. It has been suggested that an automatic electronic assessment of plan complexity may provide an alternative to the current dosimetric verification to assess deliverability [5]. The identification of less complex plans may provide a path by which VMAT quality assurance can be optimised to only include plans that have a high probability of producing inaccurate dosimetric results [1].Complexity metrics correlated with the accuracy of VMAT plans have been suggested as a possible means by which VMAT QA optimisation may take place [1-3,5]. The most promising metric is the modulation complexity score (MCS) which takes account of the leaf position and aperture variability within the plan [4]. The aim of this study is to assess the ability of MCS to inform on the deliverability of VMAT treatment plans and to verify work from previous studies. Methods- 118 VMAT plans were created and optimised for three anatomical sites; head and neck (H&N), rectum and prostate, including prostate only, prostate bed and prostate with nodes. Descriptive statistics were performed for all plan parameters (MU/field, ALPO/field, MCS), and gamma pass rates, using IBM SPSS Statistics v 16.0 (Portsmouth, Hampshire, UK). Correlations were also calculated between all plan parameters and all gamma pass rates using Pearson’s correlation. For the purposes of this report, a weak correlation was defined as r <0.4, a moderate correlation as 0.4≤ r ≤0.7 and a strong correlation as r >0.7. In addition, a statistical significance was defined as p≤0.05.

Results- The results from the current study show a weak correlation but high significance between

gamma pass rate, at all tolerance criterion, and MCS (0.330 p=0.000, 0.390 p=0.000 and 0.358 p=0.000 respectively). Discussion- The results indicate that, although MCS goes some way to predict a low gamma pass rate for VMAT plans, additional factors are involved. VMAT plan deliverability is a multi-faceted problem with a complex mix of factors. MCS takes into account some of these, however, MCS also has a number of limitations. For most plans, the gantry has a near constant rotation speed, delivering radiation during its rotation. However, when a high number of MU is required at a specific set of control points the gantry slows down in order to deliver the dose. Although MCS is weighted according to the number of MU per control point it does not take into account the limitations of the equipment. Conclusion- We found a weak correlation and high significance between MCS and gamma pass rate for VMAT treatment plans over three anatomical sites including, to date, the largest analysis specifically for head and neck sites. Our study did not find the high correlation between MCS and gamma pass rate found in similar studies. Key references (1) Agnew CE, Irvine DM, McGarry CK. Correlation of phantom-based and log file patient-specific

QA with complexity scores for VMAT. Journal of Applied Clinical Medical Physics. 2014;15(6):205-216

(2) Coselmon M, Moran JM, Radawski J, Fraass B. Improving IMRT delivery efficiency using intensity limits during inverse planning. Med Phys. 2005;32(5):1234–45.

(3) Masi L, Doro R, Favuzza V, Cipressi S. Impact of plan parameters on the dosimetric accuracy of volumetric modulated arc therapy. Med. Phys.2013;40(7):071718

(4) McGarry CK, Chinneck C, O’Toole M, Sullivan J, Prise K and Hounsell A. Assessing software upgrades, plan properties and patient geometry using intensity modulated radiation therapy (IMRT) complexity metrics. Med Phys. 2011;38(4):2027–34

(5) McNiven A, Sharpe M, Purdie G. A new metric for assessing IMRT modulation complexity and plan deliverability. Med Phys. 2010;37(2):505–15

Developing a new Patient Specific QA (PSQA) technique for the verification of Stereotactic Radiosurgery (SRS) treatments 1Bangiri Anna, 1Sutton Jonathan, 1Littler Jonathan, 1Gnutzmann Ekaterina, 1Langmack Keith 1Radiotherapy Physics, MPCE, Nottingham University Hospitals NHS Trust, UK.

Background. After a successful bid for an NHS contract, a linac-based SRS technique had to be developed, using our existing equipment, to treat patients within the East Midlands. A new single isocentre VMAT planning technique was developed using Monaco TPS. A new technique for testing the deliverability of the treatment plans as well as the accuracy of the calculations of the planning system had to be developed. Previous methods for checking the deliverability of other IMRT plans could not be applied. Methods. An anthropomorphic phantom was purchased that would specifically be used for the verification of the SRS treatments. The PSQA process that was developed included the use of a new small volume ionisation chamber (A26) as well as of a new type of Gaf-Chromic film (EBT-XD) specifically for SRS. Three measurements are collected for each PTV. One of absolute dose, a coronal film and a sagittal film with 2D dose distributions through the centre of the PTV. An in-house ImageJ script was developed to extract the 2D dose distributions from the films. An in-house Matlab script was also developed to carry out 2D gamma analysis using a 95% gamma pass rate (5%/2mm criteria). A semi-automated treatment verification process was developed for time efficiency and to ensure the accurate positioning of the phantom, in order to minimise human errors. A total of 28 patients (45 PTVs) have been treated so far.

Results. The mean±SD gamma pass rate was 97.1±2.5% and 97.5±2.3% for the coronal and the sagittal films respectively. The values ranged between 89.2% and 99.9% for coronal and 88.9% and 99.9% for sagittal films. The absolute doses measured for all PTVs had a mean±SD of +1.6%±3.5% with a range of (-6.5 and 9.5%). Four patients were not treated due to QA failure.

Discussion. The PSQA technique that was developed incorporated both absolute dose measurements as well as 2D dose distributions. This has provided us with the confidence that the SRS treatments are delivered with sub-millimetre accuracy (image on the right). Furthermore, treatment plans whose delivery was substandard were identified and the patients were not treated (image on the left). The development of in-house software as well as the automation of the PSQA delivery process have assisted greatly in identifying and minimising treatment delivery errors and have made the process more time efficient.

Conclusion. We have been successful in implementing a new frameless, single isocentre SRS technique for the East Midlands. A comprehensive PSQA process has been developed that ensures the treatment delivery is accurate and safe.

TG 100 – Risk Based RT Quality Management 1Dunscombe P. 1Department of Oncology, University of Calgary, Calgary, Canada.

Particularly in recent years there has been an increasing awareness of the actual and potential risks to patients receiving radiotherapy. However, not only might patients be injured as a result of errors made in the planning and delivery of therapeutic radiation but poor quality care could lead to compromised clinical outcomes including unnecessary morbidity and shorter than expected survival following treatment. Medical physicists have always made their best efforts to ensure that radiation-emitting equipment is functioning correctly and that treatment-planning computers calculate doses accurately. Even a superficial review of reported radiotherapy incidents will confirm, however, that failure to deliver quality care has many more causes and contributing factors than can be attributed solely to equipment. The American Association of Physicists in Medicine established a task group, TG 100, to explore and recommend on issues surrounding the safety and quality of radiotherapy in the broadest sense. This presentation will start with the rationale for adopting the prospective risk assessment approach employed by TG 100. There will be a brief overview of the 4 components of the TG 100 approach, Process Mapping, Failure Modes and Effects Analysis (FMEA), Fault Tree Analysis and Quality Management, followed by a more in-depth discussion of one of the tools, viz. FMEA. Prospective risk assessment does consume not insignificant resources and questions have been raised about its effectiveness. Some recent literature addressing such issues will be reviewed.

The Benefits of a Network Wide Approach to Linac Quality Control 1Hawthorn K,

2Brackenridge R, 1Allen V, 1Clark A. 1Radiotherapy Physics, The Northern Centre for Cancer Care, The Freeman Hospital, Newcastle, UK. 2Radiotherapy Physics, Carlisle Infirmary, Carlisle, UK.

Background. Radiotherapy services are provided within Cumbria by North Cumbria University Hospitals NHS Trust. Historically, the radiotherapy service in Carlisle has suffered from a lack of investment and a shortage of experienced staff, due in part to its modest population size and geographical context. As part of a reorganisation of healthcare within Cumbria, strong ties have developed with the Northern Centre for Cancer Care operated by Newcastle upon Tyne Hospitals, including Newcastle’s procurement of a new Varian Truebeam linac installed at Carlisle. Small centres such as Carlisle may suffer from both a lack of workforce capacity and appropriate skill mix, therefore closer working relationships between departments - benefitting both – have been employed. Our centres have collaborated on the development and implementation of a joint, networked linac QA program, bringing appropriate, modern QA methods to Carlisle in an efficient, timely manner. This collaboration has obvious advantages from the perspective of both centres; development work overheads are shared, small centres can benefit from the greater resources of others, and consistent QA protocols permit flexibility of working. Furthermore, from a governance perspective, the day-to-day performance of off-site equipment can be unobtrusively monitored. Methods. Daily dosimetry checks are carried out using Sun Nuclear QA3 devices and software which stores the data in a SQL database. An in-house software system ‘Machine Log’ is used to record the daily activity of each machine; treatment time, downtime, QA time and when control of the linac is passed between staff groups. This software also facilitates reporting of uptime data and other useful metrics easily, as well as recording the nature of machine faults and delays. Linac QA is based upon TG-142 with the addition of Ling tests and other in-house developed tests, including the dynamic reporting of flatness and symmetry with a Sun Nuclear Profiler 2. This latter test characterises energy and symmetry changes as a function of gantry rotation and dose rate. Mobius Doselab software is used to record these QA results in a SQL database; this software also allows time-saving automation of the imaging tests. Finally Qlikview Business Intelligence software is used to create a web-based front-end, providing a read-only, interactive analysis of the results from Doselab, QA3 and Machine Log from both Carlisle and Newcastle in a single place, accessible to all relevant members of staff. Discussion. It’s imperative that the automation of some tests and remote monitoring of data is used effectively and staff performing QA are not inadvertently de-skilled. As always, a deeper understanding of the tests beyond that final “Pass” or “Fail” result is important. Connectivity between centres relies on effective collaboration between Trust IT departments. Both centres have effectively increased their pool of expertise and capacity for development. With a limited number of physics staff, this collaborative approach has worked well for Carlisle to develop the service at a faster rate than would have been possible otherwise. The network wide solution is also a sensible option to a unique situation; it permits the real time performance oversight of equipment that is located remotely. In terms of the Cancer Alliance, the benefits from the harmonisation of best practice are an important step on the way to reduce healthcare inequalities. Conclusions. With the recent IPEM Position Statement on the Radiotherapy Workforce, reporting vacancy rates of 9% in the radiotherapy scientific staff group1, the opportunity for closer inter-departmental working within geographically similar (or similar equipment profiles) presents a natural efficiency to redress some of current staffing problems. This workflow is also consistent with the direction of travel from NHS England in terms of the modernisation of radiotherapy services in England.

1. Institute of Physics and Engineering in Medicine. 2016. Position statement on the radiotherapy physics workforce [Online]. IPEM. Available at: https://www.ipem.ac.uk/Portals/0/UPDATED%20POSITION%20STATEMENT%20on%20the%20Radiotherapy%20Physics%20Workforce%20FINAL.pdf [Accessed 21 August 2017].

A review of the implementation and use of QATrack+, an open source QC database for Radiotherapy Clinics. 1Burrows T, 2Greener A, 1Robinson A 1Medical Physics, The Harley Street Clinic, HCA Healthcare UK. 2Medical Physics Department, Guy’s and St. Thomas’ NHS Foundation Trust, London, UK.

Background:

In our experience the recording of quality control (QC) data is typically spread across many paper and paperless formats, most commonly maintained through excel spreadsheets. Examination of QC data in isolation makes poor use of the vast information being collected. While dedicated software does exist to maintain QC programmes it is typically proprietary, designed to work with the devices supplied by the vendor and at considerable cost to a department.

According to a 2015 global survey the use of open source software is on the rise[1]. QATrack+[3] is a highly configurable open source software originally developed in the The Ottawa Hospital Cancer Centre, Canada. As it is vendor independent it is an attractive and cost effective option for reliably recording, trending and reviewing a centres QC test results. We discuss the implementation of QATrack+ in two centres, Guy’s and St. Thomas’ NHS Foundation Trust (GSTT) and The Harley Street Clinic (HSC).

Methods:

In both centres a dedicated server was obtained, Ubuntu Server 12.04 LTS[4] was installed in GSTT and 16.04 LTS[4] was installed in HSC. QATrack+ was implemented and customised to the host centres requirements. The software is written in Python and uses a Django web framework. Tests are defined by input type, frequency and category. While basic testing can be set up with little to no programming experience additional scripting in Python allows more advanced analysis. QATrack+ is accessible from any modern web browser and allows multiple users simultaneous access.

Results:

In GSTT QATrack+ is in clinical use for 8 Linacs, 3 CT scanners, a PET-CT scanner and an orthovoltage machine across multiple sites. The system is the sole QC system used to enter, trend and review data. At the time of writing this abstract QATrack+ is about to go live in HSC and is expected to be in full use across the department by September 2017 serving 2 Linacs, a Cyberknife, a Gamma Knife and a CT scanner.

Discussion:

QATrack+ has a simple browser interface meaning an ease of use for staff, both in training and daily use. Physicists have the ability to trend data over time, compare matched machines and oversee an entire departments QC requirements and compliancy with ease. While the initial install and system set up has a large associated work load it is for the long term benefit of the department.

Conclusion:

QATrack+ provides a centralised and robust QC database. It is a highly configurable and cost effective solution to QC management. With an ever increasing global user group and planned developments such as the inclusion of PyLinac[2] on the horizon the system has great potential and is highly recommended.

References:

[1] North Bridge Venture Partners and Black Duck, “The future of Open Source” Survey, 2015: https://www.slideshare.net/North_Bridge/2015-future-of-open-source-study [2] PyLinac from PyPI - the Python Package Index, https://pypi.python.org/pypi/pylinac [3] QATrack+ © 2014 The Ottawa Hospital Cancer Centre, Canada. http://www.qatrackplus.com [4] Ubuntu © 2017 Canonical Ltd. Ubuntu and Canonical are registered trademarks of Canonical Ltd., https://www.ubuntu.com/

QATrack+ implementation as evolution from paper-based to paper-light 1Moore AR, 1Connolly W, 1Brench D, 1Meehan C, 1Trouncer R, 1Humphreys M, 1Corsini L, 1Lamb C, 1Backshall A, 1Banks C, 1Bidmead M, 1radiography team 1Medical Physics and Radiotherapy, Royal Marsden NHS FT, Fulham Road, London, SW3 6JJ

Background. At the RMH the Radiotherapy Department operates within the ISO9001:2008 quality system. In recent years the technology of our QA platform has evolved from paper-based to electronic storage (with paper back up), moving towards paperless. Replacement of one of our linacs provided the opportunity for review and reduction of machine QC, coupled with the adoption of QAtrack+[1] as a means by which we could move towards paperless QA. We have found that QATrack+ provides ease of use, unifies document format, enables consistent terminology, and handles data in a manner that provides clinically relevant information.

Methods. Our skill and evidence base provide the basis for risk estimation. The QC regimen helps us perform risk estimation on the current state of a treatment (machine, protocol) and the scope of commissioning tests. QATrack+ helps provide data security and audit trails (to facilitate external audit teams e.g. CQC). We have used our paper-light system to form a link between information platforms which provide human friendly protocols (e.g. FOSwiki) and those that are designed to record data (e.g. QATrack+). Our approach to creating an effective link addressed the following considerations:

human communication

fit for purpose and safe

management of training and authorisation (user groups / access control lists)

reference data (baselines)

unification of tolerances

ad hoc tests (breakdowns and repairs)

scheduling development and transitioning to use

involvement of relevant expertise at all times

QATrack+ was implemented as paper-light (at present some paper-based records remain) but paper remains used as a notice to users of the critical safety status of machine. A crucial and critical code review and testing phase was undertaken as part of QATrack+ local implementation.

Results. As an exemplar of the data collection and display facility, the couch vertical readout consistency assessment is detailed below: fig 1a (left) QATrack+ chart display page within data review; 1b (right) derived sample statistics from csv file saved in Excel and R.

Franklin - [C-RT-077] Daily Machine Status and Quality Assurance :: couch VRT readout at 100cm

SDD on jig (expected reading 1.0cm)

Statistic [cm]

Stddev 0.05

Mean 1.02

Max 1.13

Min 0.91

Discussion. The systems we are implementing will facilitate a move towards risk-based prioritisation for QC with the aim of increased safety and number of patient treatments. Furthermore, statistical evidence for these assessments forms a key part of the decision making process. As implementation progresses models of risk will be derived from the statistical data via methods in decision and control theory.

Conclusion. QAtrack+ complements our current paper-light systems providing a unified platform whereby therories of risk, uncertainty and probability combine with human expertise to help the operators and patients in radiotherapy.

References. [1] http://qatrackplus.com A Free Machine QC Database for Radiation Therapy Clinics

Comparison of Mobius Doselab Pro with Elekta ACAL for Linac QA using EPID images R.Harding, O. Williams, D.W.Thomas, A.Selby, M.A.Edwards. J.Williams, R Thomas. Department of Medical Physics and Clinical Engineering, Singleton Hospital, Swansea, UK.

Background. The aging population is creating pressure on cancer services during a time when standards and expectations of treatment are increasing1. The Welsh government is promoting Prudent Healthcare2 and is encouraging digital solutions to manage medical physics services3. The Doselab Pro comprehensive Linac QA system4 is a software package designed for routine QA. The aim in implementing this software is in saving time and money during QA sessions since time on Linacs is at a premium, and Gafchromic film is expensive. This method also allows online feedback, thus improving QA efficiency and patient safety. This method is at least 2 x faster than our previous technique using Elekta ACAL (not designed for routine QA).

Methods. The automatic analysis packages in Doselab for flatness and symmetry (field size) and MLC strip test were used for these tests. Tolerances need to be established because EPID images do not measure dose and are a constancy check. Baseline images were acquired for Doselab on each Elekta Linac after optimising MLC and diaphragm positions using AutoCal (Elekta software). Radiographic films were taken at the same time to establish continuity with legacy QC methods. Fields containing deliberate (e.g. MLC position) errors just within, on and over tolerance were delivered and analysed within Doselab to test the system’s sensitivity and help determine tolerance values for the prescribed QC tests. Many normal test images were taken on Linacs deemed within clinical tolerance by ACAL, to generate and test Doselab tolerances statistically.

Results. Total time required for Doselab QC per Linac including online analysis is 45 minutes, compared to approximately 1.5-2 hrs with AutoCaL. It was found that baseline images did not need to be repeated after replacing an EPID panel. Doselab picked up all errors that were deliberately introduced and these helped tolerances to be set. Tolerances were also derived statistically by analysing all images taken for the Linacs known to be within tolerance. Baseline field size values for 10 x 10 cm fields measured with the EPID and analysed using Doselab were found to be the same as the expected values (10 x 10 cm). Identical baselines and tolerances can be used for every Linac although small differences were found.

Discussion. Doselab tolerances and baseline values have been established and quality system documents have been created for the safe introduction of Doselab for routine Linac QC. It is initially being introduced for the MLC strip test and for flatness and symmetry modules. These will be used to measure field size, light radiation field co-incidence, penumbrae, diaphragm field size and return to a small diaphragm field from a large field as well as individual MLC positions, leaf bank positions and the angle between MLC bank to diaphragms. Currently the MLC strip test is not sensitive to intra/interleaf leakage with the fields currently used and this is future work.

Conclusion The use of DoseLab will be more efficient in the time taken to perform QA and obtain results as well as in terms of resource use. Time saving on the Linear accelerators will have a large benefit to patients by reducing linac down-time. Over one year it is estimated that up to 60 hrs could be saved over four linacs allowing up to 240 extra treatment slots, helping to reduce waiting times, with the associated benefits for patient survival. Time saving will also free medical physics staff from routine QC and allow them to devote time to R&D and other clinical work.

Key references. 1 http://www.cancernetwork.com/review-article/aging-and-cancer-1 2 http://gov.wales/topics/health/nhswales/prudent-healthcare/?lang=en 3 Science, technology and healthcare Dr. Rob Orford, Welsh Government, All Wales Meeting 2017 4 http://mobiusmed.com/doselab/

Improving the efficiency and accuracy of radiotherapy MR quality assurance in preparation for MR-only radiotherapy planning 1Wyatt J, 1Hedley S, 2Johnstone E, 3Speight R, 1McCallum H 1Northern Centre for Cancer Care, Newcastle upon Tyne Hospitals NHS Foundation Trust, UK. 2Leeds Institute of Cancer and Pathology, University of Leeds, UK. 3Leeds Cancer Centre, Leeds Teaching Hospital NHS Trust, UK

Background

Magnetic Resonance (MR) imaging is increasingly being used within radiotherapy because of the improvements in the reproducibility of target and organ at risk delineation due to MR’s superb soft-tissue contrast.1 An MR-only planning approach will remove registration uncertainties inherent in a CT-MR registration.2 MR-only planning will require a robust Quality Assurance (QA) programme,3 which must include geometric distortion assessment over the whole scanner Field of View (FoV).4

The Northern Centre for Cancer Care installed a dedicated radiotherapy MR scanner and has been using MR-CT fusion for radiotherapy planning since 2008. Monthly QA testing is carried out using the MagNet phantom suite and an in-house geometric distortion phantom. However these tests are time-consuming to acquire and analyse, which is costly due to the scanner’s high caseload (9am-8pm, 7 days a week) and they do not assess distortion over the scanner’s full FoV. The aim of this project was to develop a new, comprehensive set of QA tests which were efficient to deliver and analyse, and assessed geometric distortion over the full FoV in preparation for MR-only planning.

Methods

An American College of Radiologists (ACR) phantom and a Spectronic Medical GRADE phantom were used. The ACR phantom is widely used for diagnostic MR QA testing and includes small FoV geometric accuracy, high-contrast spatial resolution, slice thickness and position accuracy, image intensity uniformity, percent-signal ghosting, low-contrast object detectability and signal-to-noise ratio assessment. Open source semi-automatic analysis software5 was substantially modified and improved to make it more robust and accurate. The GRADE phantom consists of ~1200 MR signal generating markers embedded in a grid pattern with 5 cm spacing which covers the full FoV of the scanner. The included automatic analysis software was used to determine large FoV geometric distortion and further software written to summarise the results and compare to QA baselines.

Results

The new QA tests are significantly more efficient to deliver with equivalent functionality testing taking 1.5 hours rather than 3 hours. The ACR phantom semi-automatic analysis takes 2 minutes rather than 45 minutes for the manual analysis, and the results differ negligibly. The GRADE phantom analysis is entirely automatic and the results have been demonstrated to be repeatable.6

Conclusion

The new QA tests are twice as fast to deliver and the analysis is nearly entirely automatic making them significantly more efficient. This maximises the clinical availability of the scanner as well as reducing costs by saving physicist time. The assessment of distortion over the full FoV is an essential pre-requisite for the scanner being used for MR-only radiotherapy planning. Key references [1] V Khoo and D Joon. New developments in MRI for target volume delineation in radiotherapy. The British journal of radiology, 2006. [2] T Nyholm et al. Systematisation of spatial uncertainties for comparison between a MR and a CT-based radiotherapy workow for prostate treatments. Radiation Oncology, 4(1):1, 2009. [3] G Liney and M Moerland. Magnetic resonance imaging acquisition techniques for radiotherapy planning. Seminars in Radiation Oncology, 2014. [4] E Paulson et al. Comprehensive MRI simulation methodology using a dedicated MRI scanner in radiation oncology for external beam radiation treatment planning. Medical Physics, 42(1), 2015. [5] J Sun et al. An open source automatic quality assurance (OSAQA) tool for the ACR MRI phantom. Australasian Physical & Engineering Sciences in Medicine, 38(1), 2015. [6] J Wyatt et al. Investigating the reproducibility of geometric distortion measurements for MR-only radiotherapy. Poster presented at: ESTRO 36th Conference; 2017.

Developing a Quality Assurance Program for a High-Field MR Linac 1Agnew J, 1Berresford J, 1Carson P, 1Gohil P, 1O’Grady F and 1Budgell G 2,Nill S, 2Hanson I M, 2Mitchell A, 2Wetscherek A, 2Oelfke U

1Christie Medical Physics and Engineering, The Christie NHS Foundation Trust, Manchester, UK 2 The Joint Department of Physics, The Institute of Cancer Research and The Royal Marsden NHS Foundation Trust, London, UK

Background: Elekta (Stockholm, Sweden) has developed a new treatment device (named ‘Unity’) which

combines the superior soft tissue imaging capabilities of a high-field MR imaging system (1.5T) and a state of the art linear accelerator (7MV, 160 Multi Leaf Collimator). The Christie and The Institute of Cancer Research (ICR) / Royal Marsden Hospital (RMH) are members of the international Elekta MR-Linac consortium working on the clinical implementation of the MR-Linac (MRL). In the past year prototype versions of the Elekta Unity system have been installed at each consortium site.

There are several challenges in performing quality assurance (QA) measurements on an MRL due to the permanent presence of the static magnetic field of the MRI scanner. The magnetic field influences the fundamental dosimetry of the radiation beam via the Lorentz force acting on secondary electrons. This affects both the general characteristics of the radiation beam in water in terms of percentage depth dose (PDD) and profiles (Raaymakers et al 2004, Raaijmakers et al 2007) as well as the response of standard measurement equipment like ionisation chambers (Meijsing et al 2009). The Lorentz force also affects measurements in solid water, where changes in air gaps around the chamber alter the local secondary electron fluence and therefore result in large measurement uncertainties (O’Brien et al 2015, Hackett et al 2016, Agnew et al 2017).

In the Elekta Unity the linear accelerator is mounted onto a ring gantry that can rotate continuously around a bespoke magnet. Since the treatment head is not directly accessible from within the treatment room, some tests proposed by IPEM 81 and AAPM-TG145 for standard C-Arm linacs need to be modified, while others are not applicable due to the design of the system. One example is the missing light field and therefore no surrogate for the radiation field is available. The removal of the light field and the non-existing setup lasers also make the positioning of equipment more challenging.

Image guidance guidelines, originally derived for linacs with on-board kV Cone Beam CT devices, need to be redesigned to reflect the different challenges and possibilities of hybrid MR-radiotherapy units.

In its clinical phase the MRL will adapt the planned dose daily to the patient anatomy as measured at the time of treatment. Therefore patient specific pre-treatment QA measurements will need to be replaced with alternative methods. In addition to the challenges arising from the introduction of a new treatment device most radiotherapy physicists are not used to work in an MR environment; this therefore presents practical challenges regarding training aspects and safety issues, including limiting the variety of equipment that can be used or personnel who can carry out the tasks. One obvious example of this is the requirement that all QA devices must be suitable for use in the MR environment.

Within this presentation the current status of the QA program at The Christie and ICR/RMH will be presented with the focus on relative and absolute dosimetry.

Methods:

The Elekta Unity system includes a MV electronic portal imager device (EPID) panel rigidly mounted to the ring gantry. The EPID can been used to aid in the positioning of QA devices, as well as a direct method of measuring primary beam characteristics, such as output consistency, beam profiles and multi leaf collimator positions.

Absolute dose measurements have been taken in the MRL, in water, without the magnetic field present for a range of PTW ion chambers and other dosimeters, including a waterproof Farmer chamber (PTW30013) and Semiflex 3D plotting tank chamber (PTW31021) on loan from PTW. The dosimeters were positioned at isocentre using orthogonal EPID images from GA0 and GA90 and then identifying the central pixel. These measurements were repeated with the magnetic field present (B=1.5T) and a ratio taken to establish chamber correction factors for measurements made in a strong magnetic field.

MRL-specific QA platforms have been developed to ensure the reproducible positioning of equipment, such as MR compatible 2D ionisation chamber arrays, water-tanks and imaging test phantoms. These QA platforms are fixed onto the MRL couch using patient localisation bars. The

couch is then automatically positioned within the MRL bore with a sub-mm translational accuracy. The setup of the devices can then be confirmed via EPID imaging.

For the MR scanner established procedures like the MRI ACR phantom tests were performed to characterise the imaging system. Additional tests focusing on the geometric fidelity, geometric orientations, alignment of MR and radiation isocentre and online imaging have been developed within the consortium. Results:

Farmer chamber (PTW30013) measurements from the Christie were 1.4% ± 0.3% higher when measured with the chamber stem parallel to the 1.5T magnetic field in the MRL compared to the equivalent measurement in 0T field. When perpendicular to the 1.5T magnetic field, the measurement was 4.4% ± 0.3% higher than the equivalent 0T measurement.

Initial work from the RMH showed that the EPID device can be used to check output and field size consistency. Daily EPID images acquired at cardinal gantry angles show a consistency in output of ±1.5%. Water-tank, 2D array and 3D gel dosimetry measurements have highlighted the effect of the magnetic field on deposited dose, specifically the electron-return-effect at air boundaries, as well as shifts in beam profiles and PDDs.

Initial phantom imaging performed on the system with static gantry suggested that the image quality of the MRI is not affected by the operation of the linear accelerator and similar in performance to a conventional 1.5T scanner for standard imaging sequences. Discussion:

MRLs present a novel hybrid treatment device and challenging environment for QA in radiotherapy. Both the Christie and the ICR/RMH are developing methods to address these challenges as part of an international consortium. Initial results demonstrate good performance of both the radiation delivery and MR scanning components of the MRL. .

Both UK sites are working with the National Physical Laboratory to establish an independent absolute dose calibration method for the MRL to give confidence in the definite dose calibration before patients are treated on the clinical system.

Conclusion:

Procedures for performing radiation and MR QA measurements are under development and initial experience with the MRL are promising.

References: Agnew et al 2017 Quantification of static magnetic field effects on radiotherapy ionization chambers Phys. Med. Biol. 62 1731–43

Hackett et al 2016 Consequences of air around an ionization chamber: are existing solid phantoms suitable for reference dosimetry on an MR-linac? Med. Phys. 43 3961–8

Meijsing et al 2009 Dosimetry for the MRI accelerator: the impact of a magnetic field on the response of a Farmer NE2571 ionization chamber Phys. Med. Biol. 54 2993–3002

O’Brien et al 2015 TH-CD-304–08: Small air-gaps affect the response of ionization chambers in the presence of a 1.5 T magnetic field Med. Phys. 42 3724

Raaijmakers et al 2007 Experimental verification of magnetic field dose effects for the MRI-accelerator Phys. Med. Biol. 52 4283–91

Raaymakers et al 2004 Integrating a MRI scanner with a 6 MV radiotherapy accelerator: dose deposition in a transverse magnetic field Phys. Med. Biol. 49 4109–18

EPID based In-vivo dosimetry - The ultimate QA tool? Weston SJ1, Rixham PA1, Paynter D1, Fryer A1, Dawoud S1, Allen S1, Enever P2, Pyett A2, Holmes C2

(1) Medical Physics and Engineering, Leeds Teaching Hospitals NHS Trust (2) Radiotherapy, Leeds Teaching Hospitals NHS Trust

The idea of using verification images for dosimetry purposes was first evaluated in the 1980’s, and effective epid based systems have been developed in several research centres. However first commercial solutions only become available in the last five years. Efficient epid based patient dosimetry tools offer the prospect of performing in-vivo dosimetry for all patients. The effectiveness of epid-based in-vivo dosimetry as part of an overall quality process is explored. The commercial market is reviewed and initial results from a pilot of IviewDose will be presented

Commissioning and experience of using Truebeam’s Machine Performance Check application 1Barfield G, 2Carnegie DJ, 1Metwaly M, 1Cawley MG 1Radiotherapy Physics, United Lincolnshire Hospitals NHS Trust, UK. 2Radiotherapy Physics, NHS Grampian, UK.

Background. Machine Performance Check (MPC) is an application available with Varian Truebeam 2.0 platform which verifies and monitors the geometry of the LINAC as well as the properties of the beams by utilising the ISOCAL phantom and images acquired with the on board EPID. This is a fully automated application which can test both mechanical (including radiation isocentre size, MV imager, kV imager, MLC, gantry, collimator and couch positioning) as well as beam parameters (centre, uniformity and output) in under 3 minutes. Results to every MPC “check” are presented at the end of the test and allow the user to trend every result recorded by MPC to spot for drifts or sudden changes in the Truebeam which trigger further investigation.

Methods. Since the first of the department’s Truebeams were installed in late 2014 the department has performed MPC on a daily basis alongside the department’s quality control procedures allowing for a comparison to be made on those MPC “checks” and quality control procedures which are the same or similar. In tandem with this comparison the accuracy and sensitivity of MPC was also ascertained by deliberately miscalibrating the systems of the LINAC associated with some MPC “checks”. For those MPC “checks” where no comparison could be made between MPC and the department’s QA procedures then the XIM images recorded by MPC were converted to PNG files and were analysed using in-house Python scripts allowing for comparisons to be made.

Results. A sample of results are given here. Dose output as measured between MPC and daily QA3 over a period of 12 months had an average discrepancy of 0.2% (σ=0.17). The accuracy of MPC’s beam centre shift “check” was assessed by converting the 18cm x 18cm field XIM image acquired by MPC to measure the beam centre and converting it to a PNG image. A Python script was implemented to independently measure the beam centre shift on the same acquired image over an 11 month period (see figure 1).

Figure 1- plot of MPC and in house Python script measured beam centre shift (red and blue respectively)

Discussion. For the majority of “checks”, MPC has been proven to be accurate and in some cases more precise than equipment currently used in the departments quality control procedures. The use of MPC over a long period of time has given evidence of instances where MPC has aided/failed to clinically identify issues with the department’s Truebeams (such as ion chamber fault, Truebeam stand misalignment, couch rotation encoder). These instances shall be discussed.

Conclusion. With the commissioning data and experience of implementing MPC alongside our current QC checks, the department has confidence that MPC can perform accurate and precise measurements in a more time effective manner than some of our current quality control tests. The department is currently investigating what part MPC could play in the quality assurance system at ULHT.

Key references. Machine Performance Check Reference Guide v1.1, Varian Medical Systems, August 2014

0

0.1

0.2

0.3

0.4

0.5

0.6

22/12/2015 31/03/2016 09/07/2016 17/10/2016

Be

am c

en

tre

sh

ift

POSTER: Management of Radiotherapy QA Data: Our First Experience in the Implementation of a Commercial Paperless QA System. 1Esmail AA, 1Rixham P, 1Weston S, 1Cosgrove V, 2Liebl J 1Medical Physics and Engineering, Leeds Cancer Centre, Leeds Teaching Hospitals NHS Trust, Leeds, LS9 7TF. 2PTW, Lörracher Strasse 7, 79115 Freiburg, Germany.

Background In the last two decades, radiotherapy has evolved from three dimensional conformal therapy (3D-CRT) to complex IMRT and VMAT treatments. As a result of these changes, radiotherapy Quality Assurance (QA) programmes have had to adapt to the growing needs of complex treatment requirements. QA of Radiotherapy equipment is critical for achieving consistent clinical outcomes as well ensuring patient safety. Method and Results We have developed a paperless QA programme using the commercially available PTW Track-it V1.0 software (PTW -Freiburg, Germany) 1. The software was installed at the Leeds Cancer Centre in early 2016 and underwent commissioning and QA protocol development to suit the needs of our radiotherapy physics service. Protocols have been carefully developed with QA tests divided into Daily, Radiation, XVI-Imaging, 6-Monthly and Output tasks2. Each of these tasks has been assigned with acceptable QA tolerances and Pass or Fail criterion when either entering data manually via a tablet computer or importing beam profile data using PTW MultiCheck software. Each data point is recorded via Track-it software and migrated to a remote server, allowing network access to interrogate the QA data and recognise deviations and outliers between our radiotherapy machines. All protocols underwent independent clinical testing using a single radiotherapy machine before it was introduced into our QA programme. In addition to this a clinical audit log was kept of the all versions of protocols created as way of keeping a track of all versions made available for clinical use. We have now successfully developed a QA platform on PTW Track-it to perform our weekly and monthly QA programme for our twelve clinical Elekta linacs. Since its implementation we have collected 6 months of QA data for both our MLC and Agility machines. Flatness and symmetry beam profiles have been collected and analysed using PTW Multicheck Software with a two-dimensional Starcheck Array before importing into Track-it. Managing QA data efficiently also allows us to perform trending which increases our ability to predict and correct drifts. We are constantly monitoring the overall performance of our machines by comparing gamma pass rates measured using a Delta4 phantom (Scandidos, Sweden) for VMAT patient QA to evaluate whether further investigation is needed. Conclusions We share our first experience in implementing and developing a successful clinical paperless QA platform using PTW Track-it V1.0. We intend to extend this facility by developing its use for annual Linac QA. We continue to work closely with PTW in developing Track-it for future software releases.

References 1) PTW Product and Solutions, Radiation Therapy, Track-it (http://www.ptw.de/3468.html) 2) IPEM Report 81 “Physics Aspects of Quality Control in Radiotherapy”

POSTER: Questioning the Validity of Pre-Treatment Patient Specific QA 1Holmes-Smith W, 1Rudrum A 1Radiotherapy Physics, Norfolk & Norwich University Hospital Foundation Trust, Norwich, Norfolk, NR4 7UY

Background. At the NNUH we are about to start our second Managed Equipment Service (MES) contract. With the contract there is a fund dedicated to purchasing new QA equipment. The current patient QA phantom is the PTW Octavius 729 Array. This is either used on the couch for IMRT deliveries or attached to the gantry head using a gantry mount, for RapidArc deliveries. Neither of these produces a 3D dose distribution to compare to a calculated dose cube from the Treatment Planning System (TPS).

So the initial question was “which phantom do we buy?” We undertook testing of the 3 devices (SunNuclear ArcCheck, PTW Octavius 4D and Scandidos Delta 4) to compare to the 2D array.

Methods. Device testing was performed using 20 patient plans (10 simple single RapidArc plans, and 10 complex dual arc RapidArc plans. All 20 plans were delivered to all devices and compared to Eclipse TPS doses. All devices were setup using the manufacturers’ guidelines and all plans were compared using gamma index1 (3%/3mm local gamma, threshold of 20%).

Results.

Discussion. Using the standard software comparison and, in some cases, using the built in DVH comparison software, the results show that it is possible for drastic differences to occur between the phantoms when comparing the same plan. This only occurs in some instances but was more frequently seen for complex dual arc plans. This shifted the focus of our discussion towards how clinically relevant patient QA actually is. What are we trying to achieve with it?

Questions were raised about what we would do with an out of tolerance result. Is the result showing an actual error, or are the errors from phantom inaccuracies? What level of failure would cause concern to the clinician? Given the mechanical and dosimetric accuracy of machines do we need to test if the machine is capable of delivery? These questions will be explored further in the presentation.

Conclusion. With the current mechanical and dosimetric stability of linacs, there is no need to perform patient specific QA if your standard linac QA covers all aspects of the techniques you are treating and your treatment techniques have been properly commissioned. Unless you have a defined process for plan improvement using the results from the QA, then more resources should be invested in monitoring the delivered dose to patients and the effect that patient movement has on this, both internal and external. This is especially true where the results acquired vary drastically between differing current phantom options.

Key references.

1 Low DA, Harms WB, Mutic S, Purdy JA, A technique for the quantitative evaluation of dose distributions. Medical Physics 1998; 25:656-661.

55

75

95

SingleArc 1

SingleArc 2

SingleArc 3

SingleArc 4

SingleArc 5

SingleArc 6

SingleArc 7

SingleArc 8

SingleArc 9

SingleArc 10

Gam

ma

Pas

s R

ate

(%

)

ArcCheck

Octavius 4D

Delta 4

2D Array

55

75

95

DualArc 1

DualArc 2

DualArc 3

DualArc 4

DualArc 5

DualArc 6

DualArc 7

DualArc 8

DualArc 9

DualArc 10

Gam

ma

Pas

s R

ate

(%

)

ArcCheck

Octavius 4D

Delta 4

2D Array

POSTER: Quick assessment of mechanical isocentre Daniel Johnson, Joshua Kirby Department: Medical Physics, James Cook University Hospital, Middlesbrough

Aims and background Coincidence between the mechanical isocentre, the radiation isocentre and the patient-positioning lasers is important for the accurate delivery of radiotherapy. Elekta offer a phantom that, used in conjunction with the EPID, can move a ball bearing to the radiation isocentre – the lasers can then be aligned to this point. Assessing the convergence of the mechanical isocentre with this point is more difficult and often relies on locally-made bespoke equipment that can be difficult and time consuming to use. The aim of this work was to develop a quick and cheap method of assessing the convergence between the patient-positioning lasers and the mechanical isocentre as well as defining the volume that the mechanical isocentre occupies to ensure that it falls within recommended values.

Method A spot was attached to the front pointer and aligned so that the centre of the spot converged with the isocentre defined by the patient-positioning lasers. A Raspberry Pi, with camera module was programmed to sequentially take images of the spot as the gantry rotated. Using in-house software, written in python and utilising the opencv module, the spot’s size and position of its centre could be determined. Using a secondary spot on the front pointer allowed the x,y and z coordinates derived from these readings to be correlated with gantry angle.

Figure 1:from left to right: Image of the experimental setup, schematic of the setup and an image of the spot-phantom with the red spot being the one that is tracked and the green spot allowing determination of the gantry angle.

Results The agreement between the mechanical isocentre and the lasers was seen to agree to within 2mm and the volume of the mechanical isocentre was seen to have no diameter greater than 2.5mm

Discussion The analysis for the current work was done offline, however it is hoped that the analysis software can be compiled and kept on the Raspberry Pi so that a sealed unit, including the camera, could be assemble that would be compact and capable of giving a result almost immediately. It is estimated that the total cost of all the equipment necessary to make the unit would be <£200 and the total time to setup and asses the mechanical position and volume of the mechanical isocentre would be <5 minutes.

Conclusion Using the Raspberry Pi and camera module a quick and simple assessment of the linac’s mechanical isocentre can be made saving considerable time.

POSTER: Launching in-house software to improve workflow in Radiotherapy and Diagnostic settings 1Jordan K, 1 Galal M, 1 Agnew A 1 Medical Physics Department, Hermitage Medical Clinic, Lucan, Co.Dublin.

Aims: Create and commission in-house software that will better utilise Physics time and resources by reducing the time taken to complete a specific task and improving service accuracy and/or gaining additional information from the task being completed. Background: The physics department are a crucial aspect to providing scientific support to Radiotherapy and Diagnostic Radiology services. Physics staff complete tasks on a daily, weekly and monthly basis and although these tasks are paramount to ensure the safe delivery of such services, the repetitive and time consuming nature of these would lend themselves to automation, either through commercial software or in-house software. Commercial software could have less initial setup time but they tend to be expensive and may not allow customisation to produce the desired results. By purchasing programming software, bespoke software can be created, customised and integrated into current software to support a dynamic department. Methods: Tasks were identified as possible ‘automated’ or ‘semi-automated’. Programs were created using Matlab and stress tests were performed to remove bugs within the programs. Commissioning of each program was carried out and an effective QC program was created to ensure the programs continuous integrity. Results: A number of programs were created and are now clinical. These include Linac QC (e.g. field size testing and picket fence), treatment planning (e.g. manual calculations, treatment planning 3rd checks and resizing CBCT’s for possible adaptive radiotherapy) and within diagnostic radiology (e.g. daily CT QC). Using in-house software to analyse field size and picket fence tests allow quick and accurate results of MLC and jaw positioning while daily CT QC enables fast and effective analysis of uniformity and HU accuracy. Utilising programmable software for treatment planning, 3rd checks can be completed automatically and manual calculations can be completed accurately and efficiently. In-house software allows CBCT’s to be resized and imported into the treatment planning system to look at dosimetric changes due to patient contour or anatomy. Conclusion: In-house software has greatly reduced physics time and improved resources within the radiotherapy and diagnostic radiology departments in routine tasks enabling more time to be used to develop services.

POSTER: A Paperless Dosimetry QA System: three years of experience. 1Barfield G, 2Carnegie D, 1Metwaly M 1Radiotherapy Physics, United Lincolnshire Hospitals NHS Trust, UK. 2Radiotherapy Physics, Aberdeen Royal Infirmary, UK.

Background. The driver towards this project was to replace an old “mostly pen and paper” system by a completely paperless one with data being collected, stored and analysed entirely electronically. The implemented system was the Sun Nuclear Systems (SNC) ATLAS™ platform which is designed as a complete software solution to consolidate QA data and integrate with associated QA equipment [SNC QA3 for daily QA and the two dimensional ion chamber array (IC Profiler) for monthly QA]. The system was implemented around April 2015 and running smoothly with three TrueBeams. An overview of the system and the experience from the collected data are found to be worth demonstrating to encourage adequate data management nationally. Methods. Both of the QA3 and IC profiler were commissioned within the expected practical range of MU values and dose rates for a range of Photon [6MV, 10MV,10FFF] and Electron [from 6MeV to 18 MeV] energies. Each one of our four treatment rooms was individually defined in ATLAS software and equipped by a QA3 device with gantry mount, an ion chamber, an electrometer and a water Perspex phantom. This allows us to connect the dosimetry hardware with its relevant machine in ATLAS software. Once the machine name is selected in ATLAS all the related correction and calibration factors of its dosimetry hardware are automatically called from the ATLAS database. Thereby eliminating the risk of an operator selecting the wrong machine/hardware for a given QA task. The local QA3 devices are used daily, while the ion chamber measurements are performed after the QA3 once every week. Within the ATLAS system a “test builder test” is created so the ion chamber, temperature and pressure readings are entered to calculate the output and compare it with the same day QA3 reading. Thereby checking the QA3 dosimetry calibration. On a monthly basis, an output measurement is performed with an independent “Tertiary Standard” dosimetry system. This allows the comparison between the ion chamber from the treatment room and the Tertiary Standard (and hence the QA3 system for that treatment room). The Tertiary Standard system is then intercompared with the secondary standard chamber on a biannual basis. The test builder tests module allows us to store these intercomparisons, compare it with the weekly output measurements and consequently with the QA3 data for the individual machines. This hierarchical chain has also permitted the retirement of Strontium stability tests. The symmetry measurement chain is designed similarly. The IC profiler is used monthly to check the QA3 and is compared with the water tank on a biannual basis across the 4 treatment machines - all the data are stored in ATLAS. A set of baselines of open and enhanced dynamic wedged [EDW] fields were taken for the QA3 devices at zero gantry rotation while the actual tests are delivered with different gantry rotations and arcs arranged over the weekdays. For every week day a treatment plan was created to be delivered in the clinical mode. The field sequence of the individual plans was synchronized with the scheduled tests in the ATLAS software so that the expected fields in the ATLAS side always matches the delivered fields in the machine side. In fact this way of delivery offered three benefits: the first one is increasing the efficiency of the delivery and eliminating the human error of the manual setting up. The second is testing the functionality of such techniques in clinical mode, and, therefore, retiring the inherited daily test plans for this purpose. The third is performing symmetry, energy and wedge angle checks versus gantry rotations during the week days, which allows us to remove them from the biannually and the annual checks.

Results. The average time of the QA3 fields delivery is 20 minutes a day including the weekly ion chamber measurements. The trends of the output and symmetry measurements are quite steady, which gives a narrow and realistic forecast of the drifts date range. This allows us to plan in advance any machine, and prevents unpredictable adjustment. It has also proved to be an effective tool to spot a machine ion chamber leak, for incident investigation and for human error tracking.

Conclusion. The implementation of the new paperless dosimetry system has provided real time and secure data recording with no human intervention, higher frequency of tests without affecting the actual QA time on the machine, testing the functionality of special techniques (arcs and EDW) in a quantitative manner, reliable predication of the likelihood of essential dosimetry parameter drifts, and a decent tool for troubleshooting and error tracking.

Key references.

IPEM54 Commissioning and Quality Assurance of Linear Accelerators, IPEM Report 54, York, IPEM (1988)

IPEM81 Physics Aspects of Quality Control in Radiotherapy, IPEM Report 81, York, IPEM (1999).

TG 142 report: Quality assurance of medical accelerators, AAPM (2009).

90919293949596979899

100101

% P

as

s R

ate

Treatment plan

Graph 1: Comparison of PSQA gamma analysis results using three instruments for prostate plans

ArcCheck 2%/2mm

PerFraction 2%/2mm

Portal Dosimetry 2%/2mm

Pass rate for 10% plans to fail = 97.2%

POSTER: An analysis of a variety of Patient Specific QA equipment to determine which is the most suitable approach to PSQA in a new Radiotherapy Department 1O’Hare E, 1Haughey A 1North West Cancer Centre, Altnagalvin Hospital, Western Health & Social Care Trust

Introduction

PSQA in radiotherapy comprises a set of procedures and measurements to ensure that the dose to the target region is delivered accurately and as planned. An investigation was performed to determine which PSQA equipment is the most suitable for PSQA in a new radiotherapy centre. The aim of the study was to establish which of the PSQA instruments analysed (if any) is the most appropriate for regular use in a busy radiotherapy department, and what should the passing criteria for the particular equipment be for a given treatment plan.

The ideal solution will comprise PSQA equipment which is entirely independent of the Linac, whilst boasts high levels of resolution and sensitivity and is not time consuming to assemble and use, thus minimising any adverse effects to the normal clinical schedule. Method Data on 80 Breast treatment plans was gathered with comparison between the planned dose distribution and the actual measured dose distribution. Gamma analysis was performed at 2%/2mm and 3%/3mm in order to evaluate suitably sensitive tolerances. Similarly, the data for 80 prostate treatment plans was gathered.

The measured dose versus the expected TPS dose percentage correlation rates were calculated and presented using each of the ArcCheck, PerFraction, and Portal Dosimetry system solutions.

To determine the pass fail criteria it was determined that a 10% failure rate should be expected.

Results

It was found that 10% of all prostate plans will fail when the pass rate was set to 97.2% for gamma analysis at 2%/2mm, and 10% of all breast plans were found to fail when the accepted pass rate was set to 91% at 3%/3mm gamma analysis.

Discussion & Conclusion

Portal Dosimetry provides the highest resolution out of each of the PSQA instruments investigated and measurements can be made relatively quickly, however as this method uses and relies on the imaging panel of the Linac, it is not an independent instrument of the machine. In contrast to this, ArcCheck is a completely independent device to the Linac and is highly sensitive. However, ArcCheck has the lowest resolution of each instrument investigated and must also be physically placed and positioned on the treatment bed of the Linac therefore performing a measurement takes a longer time. This may have a clinical impact; requiring longer time on the treatment machine.

PerFraction measurements take less time than an ArcCheck measurement and show a higher level of resolution. Like Portal Dosimetry, the PerFraction method relies on the Linac imaging panel and is not a completely independent device, however, it makes use of independent software and employs its own unique algorithm to calculate dose measurements.

In conclusion none of the various PSQA instruments analysed where superior to the others in all aspects, aspects such as; instrument resolution and sensitivity, instrument independence from accelerator, total assembly and measurement time for the equipment. Each set of PSQA equipment has its individual advantages and disadvantages.

POSTER: Presenting an in-house, automated and efficient film analysis technique for the pre-treatment verification of stereotactic radiosurgery (SRS) brain treatments Sutton J, Bangiri A, Littler J, Gnutzmann K Radiotherapy Physics, Nottingham University Hospitals NHS Trust

Background

SRS is a hypofractionated radiotherapy technique commonly used to treat brain metastases. Since August 2016 we have been delivering SRS using Elekta linacs [3]. Prescribed doses of around 20 Gy are typical, often delivered in a single fraction. The high doses and small field sizes heighten the need for thorough pre-treatment verification. We have strived to streamline our verification workflow whilst ensuring it is capable of detecting errors that could significantly impact the treatment’s accuracy.

Method

Our pre-treatment verification process includes using EBT-XD Gafchromic film [5] to measure 2D dose distributions in the coronal and sagittal planes of an anthropomorphic phantom, STEEV [1]. We previously analysed these films by visually assessing 1D profiles taken through the centre of each film, a task which was extremely laborious and used only a small portion of the available film data. Furthermore the results were subject to interoperator variability. The rigorousness and efficiency of the film analysis have been improved through the introduction of a highly automated workflow which compares measured and planned data using a 2D gamma analysis written with MATLAB 2013 [4]. Our tolerance is for 95% of the points to pass a global gamma analysis with dose / distance-to-agreement criteria of 5%/2mm [2], [6].

Results

The analysis workflow can be performed in less than 10 minutes per PTV, compared with approximately 30 minutes to an hour using the previous method. After 31 patients with a total of 51 PTVs, the mean gamma pass rate (5%/2mm) was 97.3% (range: 88.9% to 99.9%, SD: 2.4%). Four patients were not treated following unacceptable patient specific QA results (i.e. gamma pass rates of <95%). The offending linac was taken out of use for SRS treatments pending investigation into the poor results.

Discussion

The implementation of an automated film analysis workflow has greatly reduced time spent by physicists on this task. It is simple to perform and less prone to error than the previous method which involved a high amount of manual data manipulation. Performing a 2D gamma analysis in 2 planes of STEEV gives us an objective measure which can be used to quantitatively judge a treatment’s accuracy. The technique is robust and has identified errors that may not have been detected with a simpler method. This has prevented patients from receiving poor quality treatments.

Conclusion

Our work highlights the importance of effective pre-treatment verification and could be of use to other centres performing SRS.

References [1] CIRS, Norfolk, United States. [2] D. A. Low, W.B. Harms, S. Mutic and J. A. Purdy, “A technique for the quantitative evaluation of dose distributions,” Med. Phys., vol. 25, no. 5, pp 656-661, 1998. [3] Elekta, Stockholm, Sweden. [4] MathWorks, Natick, United States. [5] MediTron, Frauenfeld, Switzerland. [6] T. Ju, T. Simpson, J. O. Deasy and D. A. Low, “Geometric interpretation of the γ dose distribution comparison technique: Interpolation-free calculation,” Med. Phys., vol. 35, no. 3, pp. 879-887, 2008.

Improvement of off-axis SABR plan verification results by using adapted dose reconstruction algorithms for the Octavius4D system 1Jeevanandam P, 1Agnew CE, 1Irvine DM, 1,2McGarry C 1Radiotherapy Physics, Belfast Health and Social Care Trust, Belfast. 2Centre for Cancer Research & Cell Biology, Queen’s University, Belfast, UK.

Background. Lung SABR patients can be treated with VMAT plans using off-axis target geometry to allow treatment in their CBCT verified position1,5-6. For patient specific quality assurance measurements using the PTW Octavius 4D phantom (PTW, Freiburg, Germany) (OCT4D) in conjunction with an Octavius array, repositioning the phantom off-axis is required to ensure the Octavius 1000SRS array (OCT1000) (PTW, Freiburg, Germany) high-resolution measurement area coincides with the tumour position 2-4. However, work to commission this setup geometry revealed larger gamma fail rates in plans delivered off-axis compared to the same plan delivered at isocentre. The aim of this work is to quantify delivery errors using an array repositioned off axis and evaluate new software that incorporates corrections for off-axis phantom measurements.

Methods. Dynamic conformal arcs and 25 lung SABR plans were created with the isocentre at the patient midline and the target volume off-axis. Measurements were acquired with an OCT729 placed at isocentre. These plans were recalculated and delivered to the both the OCT729 and OCT1000 arrays repositioned off-axis so that the high dose region was at the centre of the phantom. Comparisons were made using VeriSoft v7.0 (PTW, Freiburg, Germany) and the newly implemented version 7.1 with 2%/2mm gamma criterion (10% and 50% threshold) and results correlated with off-axis distance to the tumour.

Results. Average gamma pass rates for VeriSoft v7.0 significantly reduced from 92.7±2.4% to 84.9±4.1% with the OCT729 at isocentre and off-axis respectively. Furthermore, the average gamma pass rate was significantly correlated with off-axis tumour distance. As expected due to the improved resolution of the OCT1000, significantly higher pass rates of 95.7±3.6% were observed at isocentre but again a significant decrease in gamma pass rate with off-axis phantom distance was observed. These results demonstrate the level of errors expected due to off-axis phantom position. In contrast, even with phantom repositioning, the pass rates for analysis with VeriSoft v7.1 were 93.7±2.1% and 99.4%±1.1% for OCT729 and OCT1000 respectively. No significant difference in gamma pass rates were observed with off-axis phantom position irrespective of array type with the new software.

Discussion. The increase in errors with off-axis phantom position results from no information about phantom position in the dose reconstruction algorithm. The new VeriSoft v7.1 incorporates the phantom position information in the dose reconstruction algorithm and accounts for variations in percent depth dose (PDD) and beam divergence due to off-axis location.

Conclusion. Plans delivered with the phantom repositioned from the isocentre resulted in significant errors in dose measured due to limitations in the software to account for PDD and beam divergence at off-axis treatment geometry. We have demonstrated that it is essential that these errors are quantified and novel solutions implemented to allow accurate pass-rates to be established at off-axis target geometry. This solution will allow the use of the high resolution OCT1000 array for quality assurance of off-axis plans.

Key references.

1 A. Smith et al (2014). Progress in Med Phys. 25 288-297.

2 B. Poppe et al (2013) Med. Phys. 40(8) 082106

3 B. Poppe et al (2007) Phys Med Biol. 52(10) 2921-35

4 C. McGarry et al (2013) Med Phys. 40(9) 091707

5 H. Liu et al (2015) Med Dosim. 40(1) 76-81.

6 K. Quan et al (2015) Front Oncol. 5 213.