EPI2k8 - Electronic Portal Imaging & Positioning Devices


Wednesday, May 21

7:00–8:00 am

Concurrent Refresher Courses

9:00 – 9:30 Invited Speaker: Clinical integration of kV, MV and CBCT technologies for patient positioning
Jan Verstraete CMD, Dosimetrist, University of Leuven, Belgium
9:30– 10:30 Proffered Papers, Positioning with EPID
Session Facilitator: Kay Hatherly RTT
10:50–11:20 Invited Speaker: Flat Panel Technology
Larry Antonuk Ph.D., Professor in Radiation Oncology, University of Michigan, US
11:20–12:15 Proffered Papers, Image quality, performance evaluation and system QA
Session Facilitator: Phil Evans Ph.D.
13:30–14:00 Invited Speaker: Treatment dosimetric verification using exit dose with EPID: Clinical Experience
Bas Nijsten PhD, Medical Physicist, MAASTRO clinic, NLD

14:00–15:00 Proffered Papers, Dosimetric properties and applications
Session Facilitator: James Balter Ph.D.
15:20–15:50 Invited Speaker: 4D in-room Imaging

Paul Keal Ph.D., Chief Physicists, Stanford University, California, US

15:50–17:00 Proffered Papers, Dosimetric properties and applications
Session Facilitator: Marcel van Herk Ph.D.

Thursday, May 22

7:00–8:00 am

Concurrent Refresher Courses

8:30 – 9:00 Invited Speaker: Image Guided Adaptive Radiotherapy: New Tools and New Directions
Katja Langen PhD, Medical Physicist , MD Anderson Cancer Center, Orlando, US
9:00– 10:00 Proffered Papers, New technologies for patient positioning
Session Facilitator: Lei Xing Ph.D.
10:30–11:00 Invited Speaker: Planning without PTV
Marcel van Herk PhD, Physics Project Leader, Netherlands Cancer Institute
11:00–12:00 Proffered Papers, New developments with CT on board
Session Facilitator: Paul Keall Ph.D.
13:30–14:00 Invited Speaker: Dose re-calculation and the DGRT Process

Jean Pouliot Ph.D., Professor, University of California San Francisco, US

14:00–15:00 Proffered Papers, Dosimetric properties and applications
Session Facilitator: Sebastien Nijsten Ph.D.
15:20–15:50 Invited Speaker: Patient modeling, sampling for imaging versus change estimation

James Balter Ph.D., Associate Professor, University of Michigan, Ann Arbor

15:50–16:50 Proffered Papers, Miscellaneous
Session Facilitator: Jean Pouliot Ph.D.

Sessions Details

Concurrent Refresher Courses

Wednesday, May 21, 2008, 7 – 8:00 am

Sources of uncertainties and margins, in the 3D world, the IMRT world and the IGRT world
Marcel van Herk PhD, Physics Project Leader, Netherlands Cancer Institute

Description TBD

Wednesday, May 21, 2008, 7 – 8:00 am

A review of the "GOLD" standard; implanted markers
Frank van den Heuvel PhD, Professor, University of Leuven, Belgium

Invited Speaker

Wednesday, May 21, 2008, 9:00 – 9:30 am

Clinical integration of kV, MV and CBCT technologies for patient positioning
Jan Verstraete CMD, Dosimetrist, University Hospital Leuven, Belgium

Nowadays we know that EPIDs offer significant advantages over film and are generally more convenient to use. But at the time we moved to our new department in 1994 none of the 3 accelerators and cobalt unit had portal imaging. Till 2006 we used ready packed verification or localization films with a cassette in a mobile film holder. Almost all image evaluation was done by visually comparing simulation with the treatment film.

Photo Bas EPI2K8.jpg

Proffered Papers - Session 1

Wednesday, May 21, 9:30 – 10:30

Positioning with EPID
Session Facilitator: Kay Hatherly RTT

  • Evaluation Of 3D Offline Setup Correction Protocols Using Clinical Data Of 155 Prostate Patients With Fiducial Markers
    Kasper Pasma
    Position correction protocols are conventionally evaluated on efficacy and workload. A novel system for automated remote table control (TCSA: Theraview Couch Setup Assistant) necessitated the re-evaluation of offline setup protocols.

  • Characteristaion Of Inter-Fractional Variation In Shape Of Breast Tumour Cavity Using Portal Imaging And Implanted Fiducial Markers
    Emma Harris
    Our centre has recently taken part in a multi-centre study that was designed to evaluate the efficacy of using fiducial markers to visualise the breast tumour cavity and aid patient set-up for the delivery of a boost dose. Using the dayto-day changes in the centre-of-mass of the fiducial markers we can evaluate the statistical distributions of treatment set-up uncertainties and corresponding treatment margins for the tumour cavity. However, by only considering this data we neglect the possibility that the shape and size of the volume may change without any noticeable change in the centre-of-mass of the markers. Change in size or shape (deformation) may be due to response to radiation, healing or the migration of fiducial markers. Consequently, we may not allow adequate margins and part of the breast cavity may be subject to under-dosing. The aim of this study is to identify an index or indices with which we can characterise changes in cavity shape. Using these indices we aim to evaluate the evolution, throughout the treatment, of the shape of the tumour cavities of the 11 patient treated at our centre and determine if the margins calculated using the centre-of-mass are adequate.

  • Setup Uncertainties And Margins For Localization With Kv Imaging
    Thomas Niedermayr
    The success of radiation therapy depends on the delivery of the prescription dose to the intended target while minimizing dose to the surrounding normal tissue. Higher and more conformal target doses demand improved accuracy in target positioning. Because target positioning depends both on the technology and on clinic specific factors such as implemented procedures and processes, it is necessary to quantify margins specifically for each individual institution and treatment site. This work presents an analysis of setup uncertainties and margins for one such clinical practice.

  • The Relationship Between Rectal Size At The Planning CT Scan And Prostate Movement In Patients Receiving Radical Radiotherapy To The Prostate
    Helen McNair
    Accuracy of radiotherapy treatment depends on the reproducibility of the patient’s position and anatomy from planning CT during treatment. It has been shown that rectal volume changes can alter the prostate position and a large rectal volume at planning has since been related to poor treatment outcome. The hypothesis being that patients with a large rectal volume will experience greater rectal volume changes during treatment and hence greater prostate movement, resulting in geographical misses. We examined the relationship between rectal size at planning CT with prostate movement at treatment.

  • Tracking Of Multiple Moving Internal Fiducial Markers Based On Simultaneous kV And MV Imaging With The Presence Of A MLC
    Weihua Mao
    Intra-fraction organ motion can limit the advantage of highly conformal dose techniques as IMRT due to target position uncertainty. A new algorithm has been developed to track fiducial markers in nearly real-time based on simultaneous kV and MV imaging in a 3D CRT or IMRT treatment with the presence of a MLC.

Invited Speaker

Wednesday, May 21, 2008, 10:50 – 11:20 am

Flat Panel Technology
Larry Antonuk Ph.D., Professor in Radiation Oncology, University of MIchigan

Active matrix, flat-panel imagers (AMFPIs) have become one of the dominant x-ray detector technologies.  Such devices are based upon two-dimensional, pixelated arrays employing an “active matrix” of addressing, thin-film transistors (TFTs) – with each pixel consisting of a single TFT coupled to a pixel storage capacitor.  Their conceptual development in the late 1980’s was the direct result of a convergence of several technological trends.

In this presentation, a review of the various strategies being pursued to overcome restrictions will be presented in the context of how such strategies address fundamental limitations in AMFPI design and operation.

Photo Bas EPI2K8.jpg

Proffered Papers - Session 2

Image quality, performance evaluation and system QA

Wednesday, May 21, 11:20 – 12:15

Session Facilitator: Phil Evans Ph.D.

  • Quantification Of Sources Of Fluence Variation Incident On An EPID
    Joe Gardner
    Currently, there are two common methods to achieve EPID-based dose verification: 1) comparing the measured fluence with a simulated expected fluence1,2, and 2) backprojecting the measured fluence to reconstruct the patient dose for comparison with the planned patient dose3,4. Disagreement between measured and expected fluence originates from variation in machine-related sources (accelerator output and MLC leaf positioning) or patient-related sources (set-up errors or deformations). The aim of this work is to quantify the machine-related sources of fluence delivery errors and to establish whether fluence backprojection is valid in light of these variations and patient geometry uncertainty.

  • Long Term Stability Of CBCT Scanners For Image-Guided Radiotherapy
    Carmen Panneman
    Kilo-voltage (kV) cone beam computer tomography (CBCT) allows non-invasive and fast localization of soft-tissue structures. The use of kV CBCT integrated with a linear accelerator for image guided radiotherapy has therefore increased significantly over the past few years. The coordinate system of the linear accelerator and the image guidance system, however, are not intrinsically linked and need to be calibrated. The accuracy of the image guidance system will thus be limited by the mechanical stability over time. In this study, the long term geometrical stability of 4 Elekta Synergy systems was quantified over a period of 1-4 years.

  • Quantifying The Effect Of Motion On EPID Image Quality For Respiratory Gated Radiotherapy
    Annie Hsu
    Motion degrades image quality for all imaging modalities including electronic portal imaging resulting in increased target uncertainty during radiotherapy. This study aims to quantify the effects of respiratory motion on x-ray image quality for respiratory gated and non-gated radiotherapy and determine optimal parameters.

  • Optimisation Of A Low Z Imaging System And Flat-Panel Imagers
    David Roberts
    Portal imaging is an important stage in the treatment process as it allows verification that the patient is in the same position as when dose planning data was acquired. It is essential to enable accurate treatment of the target volume and avoidance of critical structures. Unfortunately current portal images, taken using a megavoltage beam suffer from inherently low contrast and thus set-up accuracy is compromised. In this work we explore an alternative to the current in room kV systems by modifying the bremsstrahlung target in a standard linac.

  • Portal Sensor Based On Thin-Film CDTE: Feasibility Study For Clinical High Energy X-Ray Detection
    Jun Kang
    In radiation therapy the size of a portal sensor is dictated by the largest available treatment field size, typically 40x40 cm2, not easily implemental with crystalline semiconductors. Currently most popular EPIDs are manufactured from hydrogenated amorphous silicon (a-Si:H). As a material with low atomic number and electron density, it is not the best choice for detection of high energy x-rays and charged particles, generally exhibits low quality images, and exhibits poor radiation hardness. Only in the recent years semiconductor deposition techniques has been developed to a stage where materials other than silicon can be manufactured with quality adequate for use in large-area devices such as in solar cells. We propose new design for a portal imager based on thin-film Cadmium Telluride (CdTe), which is expected to improve the device in both imaging and dosimetric applications in radiation therapy. While crystalline CdTe material is widely used in small-scale radiation detectors and has superior radiation hardness, thin film devices have not been considered from this prospective. We present the result of our investigation of thin film CdTe as a possible large area detector in comparison with a-Si system of the same design.

  • Physical Performance Of Megavoltage Cone-Beam CT
    Olivier Morin
    To evaluate the physical performance of Megavoltage Cone-Beam CT (MVCBCT) and to optimize system and reconstruction settings for image quality.

Invited Speaker

Wednesday, May 21, 13:30 – 14:00

Treatment dosimetric verification using exit dose with EPID: Clinical Experience
Bas Nijsten Ph.D., Medical Physicist, MAASTRO clinic, NDL

    EPIDs are usually applied for patient set-up verification and detection of organ motion in clinical routine. Another application is the usage of EPIDs for dosimetric verification of a treatment. In the literature, several papers describe methods to use the EPID for dosimetry. However, only few departments use dosimetric EPID methods in clinical routine for large scale patient treatment verification. We have implemented 2-D EPID dosimetry clinically based on in-house developed methods and apply these procedures to all our patients treated with a curative intent.

    The goal of this presentation is to provide an overview of our verification procedures, the applied physical methods, clinical examples and the advantages of implementing 2-D EPID dosimetry for routine in vivo dosimetry.

    Photo Bas EPI2K8.jpg

    Proffered Papers - Session 3

    Dosimetric properties and applications
    Wednesday, May 21, 14:00 – 15:00

    Session Facilitator: James Balter Ph.D.

    • The Effect Of Non-Uniform Backscatter On EPID Dosimetry
      Peter Greer
      Backscattered radiation from the support arm structures of the Varian amorphous silicon EPID (Varian, Palo Alto, CA) has been found to affect dose by up to 5% for large field sizes, particularly at the superior end of the detector where the support arm components lie.1, 2 A potential solution would be to place a lead sheet between the support arm and the EPID detector, however to date this solution is not available2. The backscattered component (BSC) is largest in fields that irradiate the maximum amount of the support arm eg the flood-field (FF). The FF calibration image therefore only compensates for BSC for an acquired image of the same field size. For smaller fields the BSC in the FF correction image will be larger than in the acquired image introducing dosimetric artifacts. In this work we characterize the BSC with field size, and investigate whether removing the BSC of the FF will improve the accuracy of EPID dosimetry for smaller fields.

    • A Comparison Of a-Si EPID Characteristics For Flat And Flattening Filter Free Photon Beams
      Elaine Tyner
      The use of amorphous silicon (a-Si) EPIDs for portal dosimetry has generated much interest. The characteristics of commercial a-Si EPIDs have been investigated and a number of groups are using a-Si EPIDs for Intensity Modulated Radiation Therapy (IMRT) dosimetry. Recently several investigators have looked at the impact of removing the flattening filter for IMRT treatment deliveries. This group is examining the possibility of using a-Si EPIDs as two-dimensional dosimeters for flattening filter free (F.F.F.) treatments. The response of an a-Si EPID to a F.F.F. beam is determined and compared with previously measured EPID characteristics for a flat beam.

    • Scatter-Primary Ratio In An a-Si EPID, By MC Simulation
      Tanya Kairn
      The use of a-Si electronic portal imaging devices (EPIDs) for dosimetry is complicated by the effects of scattered radiation. In photon radiotherapy, primary signal at the detector can be accompanied by photons scattered from linear accelerator components, detector materials, intervening air, treatment room surfaces
      (floor, walls, etc) and from the patient/phantom being irradiated. Consequently, EPID measurements are highly sensitive to these different contributions. One example of this susceptibility is the process of calibrating an EPID for use as a gauge of (radiological) thickness, where specific allowance must be made for the effect of phantom-scatter on the fluence of radiation measured by the EPID. This is usually done via a theoretical calculation which assumes that phantom scatter is linearly related to thickness and field-size. We have, however, undertaken a more detailed study of the scattering effects of fields of different dimensions when applied to phantoms of various thicknesses in order to derive scattered-primary ratios (SPRs) directly from simulation results. This allows us to make a more-accurate calibration of the EPID, and to qualify the appositeness of the theoretical SPR calculations.

    • Monte Carlo Based Investigation Of Dose Rate Dependency In IMRT Portal Dosimetry
      Rebecca Cufflin
      Sophisticated radiotherapy techniques, such as Intensity Modulated Radiotherapy (IMRT), require extensive verification measurements to ensure that the treatments are delivered correctly. Monte Carlo (MC) models of electronic portal imaging devices (EPIDs) enable the dose to the detector to be accurately predicted and directly compared with the corresponding electronic portal image (EPI). The process intrinsically involves simulating the dose within the patient as well as at the detector. This gives the potential for significantly improved accuracy in the verification of patient IMRT treatments.

    • Investigation Of An Amorphous Silicon EPID To Measure X-Ray Field Match-Lines And Define MLC Calibration
      Matthew Clarke
      Film match-line measurements have been used both to check abutting field treatments and as an MLC calibration method1. In this study the use of an Elekta iViewGT EPID to replace these film measurements was investigated. The consistency of the match-line method for MLC calibration was also investigated for a range of photon beam energies and depths.

    Invited Speaker

    Wednesday, May 21, 2008, 15:20 – 15:50 am

    4D in-room Imaging
    Paul Keal Ph.D. Chief Physicists, Stanford University, California, US,

    The goal of image guided radiation therapy is to have knowledge, or a good estimate of, the anatomy of the target and normal tissues intersected by the beam during radiation delivery, and appropriately adjust the radiation delivery for the anatomy.  4D in-room imaging facilitates the measurement or estimate of the anatomy with time.  The number of available technologies for 4D in-room imaging has dramatically increased in recent years.  Imaging prior to radiation delivery is useful as this gives an estimate of the patient anatomy during treatment, however this should be augmented by monitoring during treatment to enable the building of 4D models of the patient anatomy during treatment (see James Balter’s invited presentation).  Examples of the limitations of pre-treatment imaging alone will be shown. 

    Photo Bas EPI2K8.jpg

    Proffered Papers - Session 4

    Clinical applications with CT on board
    Wednesday, May 21, 15:50 – 17:00

    Session Facilitator: Marcel van Herk Ph.D.

    • Monte Carlo Investigation Of Dose From Cone Beam CT To Patients Receiving Stereotactic Radiosurgery For Lung Cancer Located Upper Lobe Region
      Sung Ho Park
      The linear accelerator equipped with kilovoltage cone beam on-board imager (OBI) has been designed for image guided radiotherapy (IGRT) and used for stereotactic radiosurgery (SRS) for lung cancer patients at Asan Medical Center. Cone beam CT (CBCT) images of the patient on treatment position are used for daily patient setup before SRS (48 Gy in 4 fractions). However the additional dose derived from a series of use for daily setup is one of particular concerns with IGRT. The purpose of this study is to estimate the absorbed dose accumulated from CBCT by using Monte Carlo method.

    • Cone-Beam CT Guided Radiotherapy Of Bladder Cancer
      Peter Remeijer
      Variations in bladder filling are the predominant cause of geometrical uncertainties in the radiotherapy treatment of bladder cancer. Online corrections can be a solution, but need to be reliable and fast, with regard to urinary inflow and efficiency. We propose a simple correction method based on cone-beam CT (CBCT) image guidance using lipiodol as a surrogate for the tumor position. Lipiodol is a fatty radio-opaque substance, which can be inserted with relative ease during cystoscopy. Since the correction needs to be done online, only translations are corrected in the proposed protocol. The accuracy of this method will therefore depend on the magnitude of non-translational geometrical errors, residual errors of the matching procedure and changes caused by urinary inflow. The purpose of this study was to determine the attainable margin reduction when using this strategy, given a worst case cranially located bladder tumor for which in a conventional scheme a margin of 2-4 cm would be required.

    • What Effect Does The Volume Of The Alignment Clipbox Have On Automatic Registration When Using Cone-Beam CT Guided Radiotherapy For Patients With Oesophageal Cancer?
      Alexandra Aitken
      Traditionally portal imaging and the use of stable bony landmarks have been used to verify radiotherapy set-up accuracy for patients with oesophageal cancer. Kilo-voltage cone-beam CT (CBCT) provides detailed 3D soft-tissue and bone information. This could provide more accurate treatment delivery and enable a reduction in planning target volume (PTV) margins and the possibility of dose escalation. The Elekta cone-beam software offers automatic alignment algorithms for performing registration using ‘bone’ and ‘grey’ value inside an operator defined 3D volume referred to as the alignment clipbox. The aim of this study was to assess the effect of clipbox size and position on automatic CBCT image registrations.

    • Stereographic Targeting: Automated On-Line Prostate Positioning And Monitoring Using Crossfire Kilo- And Megavoltage Portal Imaging
      Hans de Boer
      Gantry-mounted kilovoltage (kV) imaging systems are increasingly available. Hence new applications based on combined kV and megavoltage (MV) portal imaging are emerging. In the XVI system (Elekta AB), the kV beam is perpendicular to the MV treatment beam. If the kV and MV beams are imaged in rapid succession, the planar portal images (kVI respectively MVI) allow for fast imaging and 3D position reconstruction at a low imaging dose. For 3 or more implanted fiducial markers, this reconstruction is exact. We therefore developed an automated on-line positioning procedure based on kV/MV-crossfire imaging, designated stereographic targeting (SGT). We present methods and routine clinical application of SGT in prostate patients.

    • Application Of Dynamic Margins To Account For Respiration-Induced Lung Tumour Motion And Its Variability
      Catherine Coolens
      In order to reduce the sensitivity of radiotherapy treatments to organ motion, compensation methods are being investigated such as gating, tracking of tumour position and 4D scanning and treatment planning. An outstanding problem that would occur with all methods is the unrealistic assumption that breathing motion is reproducible throughout the planning and delivery process. It has been shown that the respiration-induced motion of lung and liver tumours can display a strong hysteresis effect 1 and that this effect can be exploited to define ITV margins that follow the tumour trajectory (using polar coordinates) 2. This new margin model concept 2 has been extended and applied to 3D clinical lung tumour data to (i) illustrate the effect on margin size and reduction in normal tissue irradiation and (ii) explain how such a dynamic margin model can work in practice for providing safe tumour tracking or gating in the presence of breathing variability.

    Concurrent Refresher Courses

    Thursday, May 22, 2008, 7 – 8:00 am

    Quality Assurance of EPID and IGRT technology
    Luis Fong PhD, Medical Physics Resident-Fellow, Mayo Clinic, Rochester

    Scott Hadley PhD, Associate Professor, University of Michigan, Ann Arbor
    Description TBD

    Thursday, May 22, 2008, 7 – 8:00 am

    Practical experience/guidance with implementation of IGRT technologies
    Lei Xing Ph.D., Associate Professor, Chief of Medical Physics Research, Stanford University School of Medicine, US
    Olivier Morin PhD, Physics Resident, University of California San Francisco, US


    Invited Speaker

    Thursday, May 22, 2008, 8:30 – 9:00 am

    Image Guided Adaptive Radiotherapy: New Tools and New Directions
    Katja Langen PhD, Medical Physicist , MD Anderson Cancer Center, Orlando, US

    The merging of imaging and treatment capabilities in one unit allows the smooth integration of image-guidance on the TomoTherapy Hi*ART II unit. On a day-to-day basis these images are used for patient alignment. In addition, these images can be used in an off-line process to modify subsequent treatments based on feedback from previous images. This process is generally referred to as adaptive radiation therapy (ART).
    The daily acquisition of volumetric CT data offers the opportunity to quantify the effect of patient deformation on the patient dosimetry. Once this information is gathered (at any point during the course of treatment) a clinician can judge if subsequent treatments should be modified. This dose-guided form of adaptive radiation therapy requires several steps.
    The acquisition of daily images is necessary to assemble a complete record of patient deformation during the course of treatment. Next, the dose needs to be recalculated on each image. A deformable image registration algorithm is required to assemble all dosimetric information in a common image. The difference between the original plan and the delivered dose is visualized and this information can be incorporated in the modification of subsequent treatment plans.

    Proffered Papers - Session 5

    Thursday, May 22, 9:00 – 10:00

    New technologies for patient positioning
    Session Facilitator: Lei Xing Ph.D.

    • Tumor Drift During Stereotactic Body Radiotherapy: A Note Of Caution For Pre-Treatment Imaging As The Sole Guidance Strategy
      Paul Keall
      Literature shows that if the systematic error component of motion can be removed, according to margin formulas the residual random error is not significantly deleterious to the dose delivered to the patient. The problem of the above conclusions is that the systematic error component correction, i.e. mean target position of the intrafraction motion cannot be known a priori before commencing each fraction of treatment. Thus, the most accurate static beam delivery scenario of online pre-treatment target-beam alignment may still result in systematic errors. The aim of this work is to quantify the difference in mean position and motion pattern
      between pre-treatment imaging and treatment delivery for abdominal and thoracic tumors

    • Image-Guided Radiotherapy Using Active Pixel Technology
      John Osmond
      Complementary metal-oxide-semiconductor (CMOS) active pixel sensors (APS) are commonly used in consumer products such as digital cameras, mobile phones and webcams, and recent developments have produced sensors suitable for scientific applications (e.g. Bohndiek et al., 2008). Properties such as the high read-out speed, radiation hardness, possibility of constructing low cost, large area sensors and incorporation of intelligence into the sensor itself gives these devices a potential advantage over current EPIDs. The APS architecture, together with a field programmable gate array (FPGA) situated adjacent to the sensor, can process and verify image data without the need for a computer. Such technology may form the basis of an on-line verification system capable of high speed treatment verification, monitoring patient motion and controlling the treatment where necessary. This work investigates the feasibility of using APS technology to image the megavoltage treatment beam produced by a Linac. A range of phantoms and dynamic beam deliveries were imaged and used to determine whether an APS provides sufficient information to evaluate the treatment. Two algorithms able to verify IMRT and compatible with FPGA are demonstrated, and further investigation into the use of a large area sensor is discussed.

    • Analysis Of The Effect Of Prostate Rotation On Target Coverage Using Real-Time Tracking Technology
      Camille Noel
      Electromagnetic tracking has recently been introduced for radiation therapy prostate localization by Calypso Medical Systems. Data from a clinical trial has shown that intrafraction prostate motion is more prevalent then previously thought and needs continuous tracking for its detection (Kupelian, IJROBP, 2007 & Noel, ASTRO, 2007). In addition to continuously tracking the isocenter location the system reports the initial rotation as measured during the localization. The simple interface allows quick patient set-up by the therapist, but does not facilitate more complicated analyses. Our goal was to develop the infrastructure to 1) determine appropriate rotational thresholds for patients, pre-treatment, 2) analyze dosimetric target coverage utilizing tracked positions of real patient data, 2) and visualize three-dimensional rotations of the prostate during treatment with respect to the target volume planned from simulation.

    • Intra-modality prostate positioning with 3D US and its dosimetric impact
      Danielle Fraser
      Image guided radiation therapy often depends on a reference CT image defined at the time of treatment planning, while efficient daily information is often gathered using a different imaging modality. Multi-modality registration of images acquired at different times, and including organ motion, is not straight-forward. A 3D ultrasound patient positioning system capable of both multi-modality and intra-modality image fusion is described and evaluated for prostate cancer. Image acquisition, positioning precision, and dose distributions are assessed throughout the course of radiation therapy. A ceiling-mounted optical tracking camera recorded the position and orientation of an ultrasound transducer equipped with active infrared emitters in order to synchronize the transducer’s position with respect to the room’s coordinate system. Ultrasound images were collected at the time of both treatment planning in the CT-simulator room (the reference image) and at every subsequent treatment fraction in the delivery room. Organ misalignment at the time of treatment delivery was determined by matching the position of prostate contours to that drawn on the reference image. Daily contour volumes and positions were used in Monte Carlo dose calculations to assess the dose to changing anatomy due to organ motion, and to assess the dose to the target when the patient was realigned on the treatment couch to relocate the target (prostate) at the isocentre.

    • Comparison Of Transabdominal Ultrasound And Electromagnetic Transponders For Prostate Localization: Initial Clinical Experience
      Ryan Foster
      Daily prostate localization using transabdominal ultrasound (NOMOS BAT) has been performed at our institution since 2000. For localization, an axial and a sagittal image are acquired and contours from the treatment planning system are aligned with the ultrasound images. A shift from the initial patient position to the planned position is calculated. More recently, a technology using electromagnetic transponders implanted within the prostate was introduced into our clinic. The Calypso 4D localization and tracking system (Calypso Medical Technologies, Inc., Seattle, WA) uses an array of AC magnetic coils to generate a resonant response in the transponders, which is subsequently detected using a separate array of receiver coils. The array position relative to the linac isocenter is determined by three infrared cameras mounted on the ceiling in the room. The Beacons (8 mm in length and 2 mm in diameter) are implanted in the right and left base and the apex of the prostate. With each technology, patients were localized initially using three skin marks. Localization error distributions were determined from offsets between the initial setup positions and those determined by BAT or Calypso. BAT localization data was summarized from 16236 imaging sessions spanning over 7 years; Calypso localization data consists of 1527 fractions in 41 prostate patients treated in the course of a clinical trial at five institutions and separately, the first 16 patients treated with our clinical system.

    Invited Speaker

    Thursday, May 22, 2008, 10:30 – 11:00 am

    Planning without PTV
    Marcel van Herk Ph.D., Physics Project Leader, Netherlands Cancer Institute, NDL

    Even with image guided radiotherapy, there are residual geometrical uncertainties that need to be taken into account a-priory during treatment planning, such as delineation variation and intra-fraction motion. The conventional solution is to expand the clinical target volume (CTV) with a margin to the planning target volume (PTV) such that the CTV receives a high dose even if it is not exactly in the planned location or deviates from the planned shape. Mostly, one aims at delivering a homogeneous dose to the PTV. With intensity modulated radiotherapy (IMRT), margin reduction is often applied to spare organs at risk close to the CTV.

    In addition, the homogeneous dose constraint is being loosened. It is, however, unclear whether the potential clinical benefit of these changes outweighs the risk of missing the tumor. The objective of this work is therefore to explicitly incorporate geometrical uncertainties in all steps of treatment planning for IMRT, without using a PTV, to construct better and safer treatment plans. Such a new, ‘probabilistic’, planning system has more degrees of freedom when optimizing to compromise between target coverage and sparing of organs at risk.

    Photo Bas EPI2K8.jpg

    Proffered Papers - Session 6

    New developments with CT on board

    Thursday, May 22, 11:00 – 12:00

    Session Facilitator: Paul Keall Ph.D.

    • Megavoltage Cone-Beam Digital Tomosynthesis: Imaging Characteristics And Potential Clinical Applications
      Martina Descovich
      Megavoltage Cone-Beam Digital Tomosynthesis (MVCB DT) is a recently developed imaging technique, which enables the reconstruction of several 2D slices of the patient in the treatment position moments before the radiation delivery. As for MVCB CT, 2D projection images are obtained using a medical linear accelerator equipped with a flatpanel detector. The acquisition is isocentric, one projection is acquired per degree, and the Feldkamp backprojection algorithm is used for image reconstruction. In MVCB DT, a limited set of projection images (in the range of 20o-40o of gantry arc) is used for image reconstruction. The limited angular range provides the advantage of reducing the acquisition time and the dose delivered to the patient, but affects the image quality of DT tomograms. For small arcs, the amount of out-of-plane signals, reconstructed on the plane of interest (tomographic noise) increases. In this paper, phantom images are used to study MVCB DT imaging characteristics as a function of the angular range, and patient images are used to investigate its potential clinical applications.

    • Scatter Correction For The OBI Using A Kernel Model
      Mingshan Sun
      Adaptive radiotherapy demands CBCT images with accurate Hounsfield Units (HU). To achieve this goal, the scatter component of the raw data signal must be estimated and subtracted before log normalization. This is a challenging task since scatter-to-primary ratios (SPR) can be very high (>2) [1]. The goal of this study was to develop an efficient, analytical scatter estimation and correction method that applies to full-fan and half-fan (offset detector) acquisitions, the latter which are used for body acquisitions and are particularly challenging due to complicated scatter profiles

    • Removal Of Artifacts Due To Metal Fiducial Markers In Cone Beam CT Scans
      Jaco Barnhoorn
      Implanted metal fiducial markers are commonly used in portal imaging based correction protocols, in particular for prostate. For instance, in our hospital stereographic targeting (SGT) is applied. In SGT, gold fiducials are imaged by crossfire of the MV treatment beam and orthogonal kV beam, with subsequent automated analysis and couch correction. SGT reduces systematic and random errors of the prostate center to < 1 mm (SD) in < 1 min [Mutanga et al, IJROBP 2008 (in press)]. In addition, for accurate reconstruction of delivered dose and for adaptive treatment planning, volumetric imaging such as CBCT can be useful. The high density markers, however, yield streak artifacts and shadows in the CBCT reconstruction. Identification of the prostate and rectum can be hampered by these artifacts (fig. 1). Also, automatic registration of CBCT scans, either soft-tissue or marker-based, may become less accurate. We therefore present a method for robust and easy CBCT reconstruction free from marker artifacts. The method also allows replacing the original high density markers by 'software-created' low density markers ('soft markers') with negligible artifacts.

    • Post-Reconstruction Correction Of Cone-Beam CT Scatter-Related Shading Artifacts For Improvement Of Dose Calculation Accuracy
      Tom Depuydt
      Cone beam CT (CBCT) images contain electron density information required for radiation therapy (RT) dose calculation. However, the images suffer from scatter-related shading artifacts which lower the dose calculation accuracy compared to high quality fan-beam CT (FBCT). In this study a method is presented to correct the density information, expressed in Hounsfield units (HU), of the CBCT images after tomographic reconstruction based on a priori FBCT information to improve accuracy of CBCT-based dose calculations. The algorithm was designed for application in an online replanning adaptive radiotherapy procedure.

    • Metal Artifact Correction Using Hybrid kV And MV Imaging
      Lei Zhu
      On-board cone-beam computed tomography (CBCT) performed at kV energies is increasingly being used to guide radiotherapy. While kV CT images have higher contrast-to-noise ratios than MV CT images, reconstructions of soft tissue regions near highly-attenuating metal objects often contain severe streak artifacts due to missing projection data. Over the past 25 years, a significant amount of research has been put into the development of correction algorithms. However, an effective metal artifact correction still remains a challenge. An alternative approach is to combine kV and MV data [1]. Here, we propose to compensate for the missing kV projection data using MV projections. The additional dose is minimized by irradiating only the metal object(s) with MV rays and the partial MV data are combined with a full kV data set for reconstruction. The concept of this method is described and simulation results are shown.

    Invited Speaker

    Thursday, May 22, 2008, 13:30 – 14:00 am

    Dose re-calculation and the DGRT Process
    Jean Pouliot Ph.D., Professor, University of California San Francisco, US

    For patients undergoing radiation therapy, the delivered dose may be different from the planned dose due to a variety of reasons ranging from geometrical uncertainties to anatomical changes of the patient.

    Adaptive Radiation Therapy (ART) strategies use increasingly sophisticated imaging capabilities at the treatment time to improve treatment delivery accuracy and patient dose distribution by introducing feedback into the treatment process. Although performing accurate daily positioning of the patient was the main motivation to develop imaging in the treatment room, the availability of a calibrated image depicting the patient anatomy daily offers the possibility to re-calculate the dose distribution of the initial plan on the current anatomy of the patient. The focus is thus on the true end-point of radiation therapy: the delivered dose.

    Photo Bas EPI2K8.jpg

    Proffered Papers - Session 7

    Dosimetric properties and applications
    Thursday, May 22, 14:00 – 15:00

    Session Facilitator: Sebastien Nijsten Ph.D.

    • Errors Detected By Clinical Use Of EPID Dosimetry
      Anton Mans

    • Radiological Thickness As A Measure Of Dose In An EPID
      Tanya Kairn
      Monte Carlo (MC) simulations are highly amenable to predicting the dose deposition in the active layer of an a-Si EPID arising from the transmission of the radiotherapy beam through some intervening attenuating medium. However, reliably comparing these simulation data with experimental results (other than in relative terms) remains a challenge. Firstly, the experimental images may be automatically flood-field corrected by their acquisition system, making them qualitatively as well as quantitatively different from the MC images and, secondly, the difficulty of relating relative EPID-image pixel values to absolute dose is well known. The purpose of this work is to suggest a clinically viable method of obviating both of these difficulties and generating quantitatively comparable data from experiment and MC simulation, using radiological thickness as a surrogate for delivered dose. This work is unique in comparing experimental thickness measurements with commensurate MC results, in investigating the utility of the a-Si flat panel EPID as an indicator of radiological thickness, and in contending that such indications of thickness can be used to monitor the fundamental properties of the treatment.

    • In Aqua Vivo EPID Dosimetry - A Method For In Situ Dose Verification For Lung Cancer Treatments By EPID Transmission Measurements
      Markus Wendling
      With the development of high dose/high precision radiotherapy for cancer patients, dose verification becomes more and more important. Legislative requirements may underline this necessity in the future. In our hospital we now perform dose verification for IMRT treatments of all curatively treated cancer patients using EPID dosimetry. This is done preferably in vivo or otherwise prior to treatment using a phantom. Our EPID dosimetry method is a measurement-based approach combined with a fast and simple back-projection algorithm. Its major drawback is its poor ability to model (large) tissue inhomogeneities, such as lung. In these cases, pre-treatment dose verification is an option: the patient is replaced by a homogeneous slab-geometry phantom on which the plan is re-calculated. However, such verification cannot be done during actual patient treatment and it involves extra work. In this study we present a new method for in vivo dosimetry that allows dose verification during the actual patient treatment but avoids the inhomogeneity problem.

    • In-Vivo Portal Dose Measurements For Reconstruction Of The Dose Of The Day
      Marcel Breuers
      As part of our patient-specific QA program, pre-treatment fluence delivery verification has routinely been performed for each IMRT field for over 6 years using a CCD-based EPID (Theraview NT, Cablon Medical) (1). The required dosimetric calibration of the EPID, including correction for cross-talk, has entirely been based on a limited number of EPID measurements (no ion-chamber involved) (2). A 3D dose reconstruction based on measured portal dose images and the planning CT scan was used to evaluate the impact of observed pre-treatment fluence deviations (3). However, in-vivo verification of the delivered 3D dose distribution, based on fluences measured behind a patient combined with the actual patient anatomy in each fraction, i.e., the dose of the day, is the most complete verification method. Therefore, we have developed a method for accurate 3D in-vivo dosimetry, fully based on EPID measurements for calibration.

    • 3D In-Vivo Dose Reconstruction Using EPID Measurements And Mega Voltage Cone-Beam CT Scans
      Wouter van Elmpt
      Verification of the dose delivered to cancer patients during external beam radiotherapy is important to guarantee accurate treatment delivery. We have developed a method to verify the 3D dose distribution delivered to the patient during treatment: 3D in-vivo dosimetry. By combining EPID measurements behind the patient with on-line imaging using a Mega Voltage cone-beam CT-scan (MV CBCT), it is possible to obtain the delivered 3D dose distribution incorporating possible changes in patient anatomy between treatment planning and prior to treatment delivery.

    Invited Speaker

    Thursday, May 22, 2008, 15:20 – 15:50 am

    Patient modeling, sampling for imaging versus change estimation
    James Balter Ph.D., Associate Professor, University of Michigan, Ann Arbor, US

    The wealth of new technology provided in-room for setup, monitoring and plan modification presents interesting challenges, and simultaneously opens a new wealth of opportunities for optimizing the processes of precise treatment delivery.  Highly detailed radiographs and tomographic images can be produced in the treatment room.  Potentially very precise measurements can be extracted from these images, although confounded by temporal effects and at potentially high costs in terms of imaging-related body doses.

    Photo Bas EPI2K8.jpg

    Proffered Papers - Session 8

    Thursday, May 22, 15:50 – 17:00

    Session Facilitator: Jean Pouliot Ph.D.

    • Improving Tracking Of Implanted Radio-Opaque Markers On MV EPID Imaging With Support Vector Machines And Morphological Analysis
      Nadeem Riaz
      Highly conformal radiotherapy techniques provide radiation doses that closely conform to tumor volume while sparing sensitive structures. However, these techniques only crudely account for potential tumor motion during a treatment session. Several groups have investigated implanting gold markers into tumors and tracking these markers with kV imaging during a radiotherapy treatment session. The disadvantages of kV imaging for tumor tracking are increased radiation dose to the patient and the need for additional equipment. Using MV EPID imaging can reduce the radiation dose and does not require additional equipment. However, MV EPID images have significantly reduced contrast compared to their kV counterparts which significantly increases the difficulty of marker detection. Our group has worked on an algorithm (recently submitted, see figure 1) robust enough to detect moving markers in implanted tumors on MV EPID images. However our prior work is limited to tumors moving slower than 1.6 cm/s. Here we expand on our prior work to detect faster moving fiducials.

    • Grey-Value Prostate Registration After Cone Beam CT Motion Artefact Reduction
      Jasper Nijkamp
      Organ motion limits the accuracy of radiotherapy. In case of prostate cancer patients treated in supine position the position of the prostate is mainly influenced by the rectum. Cone beam CT (CBCT) integrated with a linear accelerator is frequently used to image soft-tissue structures just prior to treatment. Besides visualizing the prostate with these scans it is also possible to visualize the position of the seminal vesicles and shape of the rectum. During acquisition of a CBCT scan (1 - 2 minutes), however, intra-scanning motion induces artefacts. Moving gas pockets at the level of the prostate therefore make it difficult to visualize and localize the prostate (with an automatic soft-tissue registration technique [1]). Ignoring scans because registration was not successful might lead to an underestimation of the organ motion. The purpose of this study was to develop an artefact reduction algorithm, called Flatex (Flatus exit), and to quantify its effect on the success rate of automatic prostate localization.

    • Prostate Imaging With A High Contrast Megavoltage Cone Beam CT Beam Line
      Vincent Wu
      Patient dose and contrast-to-noise ratio (CNR) are concerns associated with megavoltage cone beam CT. A novel imaging beam line has been developed that employs a target with low-Z material to generate low energy (diagnostic) photons. In addition, the flattening filter is removed to prevent the absorption of low energy photons that play a key role in high contrast imaging. This imaging beam has been used to generate cone beam CT on a number of patients undergoing radiotherapy for prostate cancer. The purpose was twofold, (1) to determine how well soft tissue such as the prostate, rectum and bladder could be visualized with this new beam, and (2) to determine how much the dose could be reduced for the same image quality. The minimum dose required for patient setup using the gold seeds was also determined.

    • MEGAVOLTAGE CONE-BEAM CT (MVCBCT): Dose Calculation And Comparative Analysis Of Dose Delivered To Target And OAR
      Cecilia Haddad
      During these last few years there has been great progress in improving precision and accuracy in patient positioning and tumor localization in Radiation Oncology. These goals have been achieved through image acquisition before and/or during treatment delivery, new strategies pertaining Image Guided Radiotherapy (IGRT). Hospital Sírio Libanês (HSL) has selected Megavoltage Cone-Beam CT (MVCBCT) for IGRT, using the linear accelerator.

    • A Monoscopic Method For Real-Time Target Position Estimation Combining X-Ray Imaging And Respiratory Monitoring
      Byungchul Cho
      Three major linear accelerator vendors offer gantry-mounted single x-ray imagers. The use of monoscopic imaging to estimate three-dimensional (3D) target positions is limited by the unresolved motion that is parallel to the imaging beam axis. In this study, we propose a monoscopic method for real-time tumor position estimation, combining occasional internal target monitoring and continuous external respiratory signals, where a correlation model between the unresolved coordinate and the two other coordinates on the projection plane is determined by a prior 3D target trajectory and updated during the treatment. The estimation performance of the novel method was demonstrated and compared with that of the stereoscopic method.