A Monte Carlo simulation toolkit for optimization studies of digital mammography

1. A Monte Carlo simulation toolkit for optimization studies of digital mammography Sakellaris T1*, Pascoal A1 and Koutalonis M2 1Faculty of Engineering, Catholic University of Portugal, Lisbon, Portugal 2Bart's Health NHS Trust, Clinical Physics Department, London, United Kingdom * Corresponding author: msakellar@gmail.com 1. PURPOSE • The Graphical User Interface (samples) Development of a Monte Carlo simulation toolkit dedicated to the design of optimization studies of digital mammography. Form which summarizes the irradiation conditions Form for breast phantom design 2. METHOD A validated Monte Carlo model for conventional mammography[1], which considers a mathematical breast phantom with various types and shapes of clinical relevant lesions (microcalcifications and masses), was extended to include an a-Se detector, a linear parallel anti-scatter grid (stationary and moving) as well as a partially isocentric moving focal spot for oblique irradiations. Avalidated Monte Carlo dose calculation algorithm was integrated in the model to enable 3-D glandular dose estimations[2]. The code is currently being associated with a graphical user interface (GUI). The detector model simulates x ray-matter interactions, and produces images considering the energy absorbed inside a-Se and Poisson noise. Its validation was made comparing the simulation pre-sampling MTF (MTFpre) with theory[3]. The grid simulation is based on a compartmental approach of photon transport and was validated comparing the signal difference to noise ratio improvement factor (SIF) with published data[4]. The simulation of oblique irradiation uses Euler transformations and it was validated comparing the simulation MTF due to oblique x-ray incidence (MTFobl) with theoretical predictions[5]. • Characteristic simulation output Simulation images of a designed breast phantom Image from Primary & Scattered radiation Image from Scattered radiation only 3. VALIDATION RESULTS 20 keV x-rays, 2 cm thick breast with a 2 cm radius & 50% glandularity The MTFpre as well as MTFobl differed approximately by 1.2% (< 0.5 lp/mm) from theoretical, and the SIF factor differed 4.7% from the published values. Simulation images showing the effects of a moving anti-scatter grid Images from Primary & Scattered radiation Images from Scattered radiation only No anti-scatter grid With anti-scatter grid No anti-scatter grid With anti-scatter grid 4. THE SIMULATION TOOLKIT Air Calcium Oxalate Calcium Oxide • Overview of simulation capabilities 20 keV x-rays, 2 x 2 cm2 irradiation field, 2 cm thick breast with 50% glandularity Simulation images of oblique irradiations with a 2 x 2 cm2 field size • Moving Focal spot • Partial isocentric motion • User defined focus-center of rotation (COR) distance and irradiation angle Focal spot: • Point or • With finite dimensions and 3 intensity distributions: Images from Primary & Scattered radiation -30o -10o 0o +10o +30o • X-ray energies: • Mammographic spectra (From tabulated data) or • Monoenergetic spectra Uniform Distribution Gaussian Distribution Double Edge Distribution Images from Scattered radiation only Collimator -30o -10o 0o +10o +30o Mathematical breast phantom: • Semi-cylindrical slice • Varying radius, thickness & composition • Homogeneous or • With various types & shapes of clinical relevant lesions (microcalcifications & masses) and/or • Test-object patterns (e.g. bar patterns) Varying Focus-Breast Distance (Magnification mammography) 20 keV x-rays, 2 x 2 cm2 irradiation field, 2 cm thick breast with 50% glandularity Edge-test device (tiltable): • Parallel or • Vertical to chest wall • User defined: • Focus-edge distance • Edge dimensions, material, thickness and angle Simulation image of an edge-object placed on top of a breast phantom of 4 cm thickness and 50% glandularity 2-D AGD distribution inside a 2 cm thick breast of 50% glandularity, irradiated with 20 keV x-rays at 30o angle, with a field size of 2 x 2 cm2 Final Report Glandular dose (mGy) • Glandular & Adipose dose calculations • 3-D dose distributions • User defines voxel size Compressor paddle Edge-object (Tungsten) Linear anti-scatter grid: • Stationary or Moving • User defined: • Strip material, thickness and height • Interspace material and width a-Se digital detector: • User defined: • Detector dimensions • Number of pixels • Detector thickness The user selects which components to irradiate e.g. edge and/or breast phantom and/or grid and detector Voxel size: 100 um Chest Wall side • The Graphical User Interface (samples) 5. CONCLUSION Form for digital detector specification The validated model, associated with a GUI, could provide a user-friendly simulation toolkit for multi-parametric image and dose optimization studies of digital mammography, with perspectives of its use in digital breast tomosynthesis. • The code produces linearizable simulation images which can be calibrated using the Signal Transfer Property of a real a-Se system 6. REFERENCES • [1] Spyrou G, Panayiotakis G and Tzanakos G 2000 MASTOS: Mammography Simulation Tool for design Optimization Studies Med. Inform.25 275-93 • [2] Delis H, Spyrou G, Panayiotakis G and Tzanakos G 2005a DOSIS: a Monte Carlo simulation program for dose related studies in mammography Eur. J. Radiol. 54 371–6 • [3] Zhao W, Ji W G and Rowlands J A 2001 Effects of characteristic x rays on the noise power spectra and detective quantum efficiency of photoconductive x-ray detectors Med. Phys.28 2039-49 • [4] Cunha D M, Tomal A and Poletti M E 2010 Evaluation of scatter-to-primary ratio, grid performance and normalized average glandular dose in mammography by Monte Carlo simulation including interference and energy broadening effects Phys. Med. Biol.55 4335-59 • [5] Que W and Rowlands J A 1995 X-ray imaging using amorphous selenium: inherent spatial resolution Med. Phys.22 365-74 • It gives the ability to the user to calculate objective image quality metrics such as pre-sampling MTF, NNPS and DQE • It gives the ability to the user to perform complex simulations in order to calculate more specialized objective image quality metrics such as generalized (G) GMTF, GNNPS and GDQE or effective DQE 7. ACKNOWLEDGEMENTS We would like to thank Prof. George Spyrou for his valuable contribution in this work. This work was supported by national funding through the Foundation for Science and Technology (FCT) of Portugal within the framework of the project with reference number: PTDC/SAU-BEB/100745/2008.