Proton dose calculation algorithms and configuration data Barbara Schaffner PTCOG 46 Educational workshop in Wanjie, 20. May 2007 VARIAN Medical Systems
Agenda Broad beam algorithms Concept of pencil beam dose calculation Beam configuration data for Eclipse Proton CT calibration VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 2
Agenda Broad beam algorithms Concept of pencil beam dose calculation Beam configuration data for Eclipse Proton CT calibration VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 3
Broad beam algorithm - Concept Find the range and the modulation of the beam required to cover the target Find the depth and the distance to the aperture for each point of interest (POI) Use look-up tables for the dose in depth and apply a correction for the lateral penumbra VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 4
Broad beam algorithm - Concept Distance to aperture Depth POI Range Modulation VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 5
Broad beam algorithm - Applications First proton dose calculation algorithms Used today only for ocular applications Very fast algorithm Loss in accuracy is small due to homogeneous tissues VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 6
Agenda Broad beam algorithms Concept of pencil beam dose calculation Beam configuration data for Eclipse Proton CT calibration VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 7
Eclipse pencil beam algorithm - Concept Principle Fluence e Φ(x,y,z) Dose Beamlet kernel B(x,y,WeR) Convolution of 3D undisturbed proton fluence in air with a beamlet in water. In practice Superposition of inhomogeneity - corrected beamlets and multiplication with fluence at calculation position. VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 8
Fluence of a double scattering beam line 1st and 2nd scatterer Homogeneous fluence distribution Eff. Source size and position Absorber, Modulator Broadening of penumbra Fluence Compensator Broadening of penumbra Inhomogeneity of fluence Block Transversal shape of beam (Block scattering component) VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 9
Example of fluence distribution VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 10
Fluence of a scanning beam line Phase space of single beam spot (Divergence, beam size at reference position) Fluence Scan pattern and weights of individual beam spots VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 11
Example of scanning fluence distribution Fluence of single pencil beam z Total fluence x Scan pattern and weights 10 9 8 7 6 5 y x 4 3 2 1 0 0 1 2 3 4 5 6 7 8 9 10 VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 12
Beamlet in water Definition: Dose deposited by a proton beam, which hits a water surface without lateral extension and without angular divergence or confusion. Characterization: Depth dose distribution Transversal distribution Independant of technique! VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 13
Model of beamlet in water Model is based on analytical functions fitted to Monte Carlo calculated depth dose distributions Depth dose and scattering distributions are modeled for primary and secondary protons separately Measured depth dose curves are used to extract beam-line specific parameters for the model VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 14
Example of calculated depth dose distribution VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 15
Transversal distribution Beamlet transversal distribution at each position in depth is calculated from the sigma of one or several Gaussians. Transversal distribution of primary protons = sum of two Gaussians VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 16
Inhomogeneity correction Each beamlet is corrected for density variations before it is multiplied with the proton fluence in air VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 17
Total SOBP dose Contributions of all layers are added in a final step. VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 18
Agenda Broad beam algorithms Concept of pencil beam dose calculation Beam configuration data for Eclipse Proton CT calibration VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 19
Measurements the input for the dose model Fluence calculation Open field cross profiles -> Virtual SAD Fluence along Z axis -> Effective SAD Half beam block -> Lateral penumbra of fluence Cross profiles of spot fluence -> Phase space Beamlet calculation Depth dose curve -> Depth dose parameters VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 20
Virtual SAD Calculation The virtual SAD is the geometric SAD It is found by a linear fit to the field sizes measured at different positions along the beam axis Field Size Vitual SAD Block Position VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 21
Effective SAD Calculation The effective SAD is the dosimetric SAD It is found by a 1/r 2 fit to the measurements of the fluence along the beam direction Fluence in air Measurement Effective SAD VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 22
Effective Source Size Calculation It is assumed that the penumbra shape can be modelled by the effective source size concept introduced by Hong et al. The proton source is assumed to be located at the position of the nominal source and to have a Gaussian distribution. Proton distribution at source position Half-beam block Error function shaped penumbra Nominal SAD VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 23
Phase Space Parameters Phase space parameters are extracted from cross profile measurements of the spot fluence in air. The following formula is fitted to the spot size plotted as a function of the distance from isocenter: A C f z = + B z + 2 2 Parabola with A, B, C parameters 2 ( ) z Nominal SAD Spot fluence VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 24
Preprocessing of measured depth dose Measured depth dose gets shifted in depth by water equivalent thickness of the beamline Measured depth dose is corrected for intensity loss due to 1/r 2 effect 0.14 0.12 Dose [Gy/MU] 0.1 0.08 0.06 Measurement Shifted by NeT SAD corrected 0.04 0.02 0 0 2 4 6 8 10 12 14 16 Depth in water [cm] VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 25
Calculation of Depth Dose Parameters Resulting depth dose is used for a least square fit of analytic depth dose formula with free parameters Energy Spectral distribution Σ Normalization ~Energy Normalization MeV/cm Gy/MU Fraction of secondary protons (optional) Spectrum Σ VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 26
Parametrization vs. LUTs Parametrization Requires an analytical model for measured data. Failure of the model leads to bad dose calculation results Allows interpolation (extrapolation) for situations other than the measured one (e.g. intermediate energy) Smoothes or averages results from measurements LUT (Look up table) Measured curve can have an arbitrary shape Difficult to interpolate, almost impossible to extrapolate Needs some processing of measurements to smooth uncertainties VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 27
Parametrization vs. LUTs Example 1 Depth dose Extrapolation of dose calculation and better model for coarse measurements Interpolation of calibration factor for different energies VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 28
Parametrization vs. LUTs Example 2 A-Parameter (= 2*spot size at isocenter^2) for a scanning beam Note: User must know, whether observed behavior of spot size is due to measurement uncertainties or other effects. VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 29
Agenda Broad beam algorithms Concept of pencil beam dose calculation Beam configuration data for Eclipse Proton CT calibration VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 30
Why CT calibration? CT scanners measure Hounsfield units defined as μ HU =1000 water -μ μ water material Hounsfield units are a measure for the attenuation of an X-ray beam Hounsfield units cannot be used directly to calculate the behavior of proton or even high energy photon beams VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 31
Definitions Scaled HU: HU sc = HU + 1000 Absolute stopping power (SP) : -de/dx [MeV/cm] Absolute mass SP : -(1/ρ)dE/dx [MeV/(g/cm 2 )] Relative SP : Ratio between abs. SP for material and water Relative mass SP : Ratio between abs. mass SP for material and water CT calibration: LUT for conversion between HU and relative SP VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 32
Importance of accurate CT calibration A few percent error in the CT calibration causes a few percent shift of the dose in depth Due to the high distal gradient of the dose, this may lead to 100% difference in dose! Note: The calibration curve is sensitive to the X-ray spectrum of the CT scanner. It is important, that each scanner is calibrated individually. VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 33
The stoichiometric CT calibration method Scan tissue equivalent samples of well known composition individually in the center of a water phantom Obtain parameters for HU sc calculation by fitting results to HU-formula Calculate (HU sc,sp rel ) pairs for real tissue compositions Define calculation curve to follow calculated (HU sc,sp rel ) pairs for real tissue compositions VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 34
Properties of the stoichiometric CT calibration method Scanner specific due to the use of reference measurements Uses a simplified, empirical theory for the calculation of Hounsfield units Yields good results for commonly used scanner energies of approx. 80 kev and biological tissue VARIAN Medical Systems
Calculation of (HU sc, SP rel ) reference data pairs Calculation of HU sc and SP rel pair based on literature data for tissue composition { } ph 3.62 coh 1.86 KN HU sc = Cmat K + K + K Don t use for MV CT! Z ~ Z ~ K xy parameterize the response of the scanner. They are determined by fitting the above formula to HU sc - values measured on samples of well known chemical composition SP rel calculated with Bethe-Bloch formula VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 36
Plot of calculated (HU sc,sp rel ) pairs and linear fits 1.8 Small body calibration 1.6 1.4 1.2 Rel. Stopping Power 1.0 0.8 0.6 0.4 0.2 0.0 Organs and muscle Adipose Breast Bone marrow Cartilage Bone Fit to soft tissue Fit to bone Fit to adipose Calibration curve 0 500 1000 1500 2000 2500 Hounsfield Units Rel. Stopping Power 1.10 1.05 1.00 0.95 0.90 0.85 Organs and muscle Adipose Breast Bone marrow Cartilage Fit to soft tissue Fit to bone Fit to adipose Calibration curve 0.80 800 850 900 950 1000 1050 1100 1150 Hounsfield Units VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 37
Some references for further reading Schneider U, Pedroni E, Lomax A 1996 The calibration of CT Hounsfield units for radiotherapy treatment planning Phys. Med. Biol. 41 111-124 Schaffner B, Pedroni E 1998 The precision of proton range calculations in proton radiotherapy treatment planning: experimental verification of the relation between CT-HU and proton stopping power Phys. Med. Biol. 43 1579-92 Kanematsu N et al. 2003 A CT calibration method based on the polybinary tissue model for radiotherapy treatment planning Phys. Med. Biol. 48 1053-64 Schneider U et al. 2005 Patient specific optimization of the relation between CT-Hounsfield units and proton stopping power with proton radiography Med. Phys. 32(1) 195-9 VARIAN Medical Systems Barbara Schaffner, PTCOG 46 Slide 38
Thank you for your interest and attention VARIAN Medical Systems