Premium
Sci‐Fri PM Imaging‐05: Prediction of multiple scatter distributions using an iterative Monte Carlo technique for X‐ray Scatter Mammography
Author(s) -
Van Uytven E,
Pistorius S,
Gordon R
Publication year - 2006
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.2244680
Subject(s) - imaging phantom , monte carlo method , mammography , optics , compton scattering , collimated light , physics , detector , photon , projection (relational algebra) , computed radiography , medical imaging , image quality , computer science , mathematics , artificial intelligence , algorithm , image (mathematics) , medicine , statistics , laser , cancer , breast cancer
Compton and Rayleigh x‐ray scatter have been shown by several authors to be a source of information for electron density imaging. However, as these techniques rely on single scatter information only, the presence of second order and higher scatter typically requires collimation of the detector, reducing the usefulness of many scatter technique for in‐vivo imaging. Compton scatter radiography (CRSR) is a single scatter imaging technique that uses a scanning pencil beam and a mammographic geometry to extract 3D atomic structure information of the breast from a single projection. CRSR uses an uncollimated scatter geometry, and the detector system records both single and multiple scatter signal, the latter component being as large as 40% of the total scatter signal at some locations. The error tolerances of CSR (∼1%) require accurate prediction of second‐order and higher scattered photons. Under a mammographic setting, this work describes an iterative method of multiple scatter prediction using Monte Carlo simulation and CRSR imaging. Simulations were performed with a mammographic phantom to test the validity of this approach. A 5×5×5 cm 3 phantom was constructed containing a small calcification 3 cm deep and 1 cm from each beam‐parallel phantom edge. The multiple scatter component accounts for 26.2% of the total scatter, and CSR image reconstruction shows significant errors and a failure to localize the calcification. After multiple scatter prediction, 0.82% of the multiple scatter remains unaccounted for, and the calcification was successfully localized with a 1×1×2.5 mm 3 resolution.