TY - GEN
T1 - Development of a peripheral thickness estimation method for volumetric breast density measurements in mammography using a 3D finite element breast model
AU - Alonzo-Proulx, Olivier
AU - Mainprize, James G.
AU - Packard, Nathan J.
AU - Boone, John M.
AU - Al-Mayah, Adil
AU - Brock, Kristy
AU - Yaffe, Martin J.
N1 - Copyright:
Copyright 2010 Elsevier B.V., All rights reserved.
PY - 2010
Y1 - 2010
N2 - A method was developed to determine the area in a mammogram where the breast is not in contact with the compression paddle (the periphery), and to predict the breast thickness in that peripheral region. The periphery is determined by evaluating the variation of the signal intensity along radial lines, and the peripheral thickness is modeled assuming the breast has a semi-circular shape. The method was tested on 26 simulated mammograms for which the volumetric information was available. The mammograms were obtained using CT data that were deformed to simulate mammographic compression and then projected using a physical model. The method predicted the thickness in the periphery to within 3.3 mm of the actual value and the volumetric breast density within 4.3 percentage points. The method was also tested on 209 digital mammograms, and on average it was estimated that thickness errors occurred on 9% of the breast image, and the average absolute thickness error on those points was estimated to be approximately 2.0 mm in the periphery and central region of the breast but as large as 10.5 mm in the extreme periphery where the thickness is small.
AB - A method was developed to determine the area in a mammogram where the breast is not in contact with the compression paddle (the periphery), and to predict the breast thickness in that peripheral region. The periphery is determined by evaluating the variation of the signal intensity along radial lines, and the peripheral thickness is modeled assuming the breast has a semi-circular shape. The method was tested on 26 simulated mammograms for which the volumetric information was available. The mammograms were obtained using CT data that were deformed to simulate mammographic compression and then projected using a physical model. The method predicted the thickness in the periphery to within 3.3 mm of the actual value and the volumetric breast density within 4.3 percentage points. The method was also tested on 209 digital mammograms, and on average it was estimated that thickness errors occurred on 9% of the breast image, and the average absolute thickness error on those points was estimated to be approximately 2.0 mm in the periphery and central region of the breast but as large as 10.5 mm in the extreme periphery where the thickness is small.
KW - digital mammography
KW - periphery detection
KW - volumetric breast density
UR - http://www.scopus.com/inward/record.url?scp=77954651008&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=77954651008&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-13666-5_63
DO - 10.1007/978-3-642-13666-5_63
M3 - Conference contribution
AN - SCOPUS:77954651008
SN - 3642136656
SN - 9783642136658
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 467
EP - 473
BT - Digital Mammography - 10th International Workshop, IWDM 2010, Proceedings
T2 - 10th International Workshop on Digital Mammography, IWDM 2010
Y2 - 16 June 2010 through 18 June 2010
ER -