TY - GEN
T1 - Size-adapted segmentation of individualmammographie microcalcifications
AU - Arikidis, Nikolaos S.
AU - Karahaliou, Anna
AU - Skiadopoulos, Spiros
AU - Korfiatis, Panayiotis
AU - Likaki, Eleni
AU - Panayiotakis, George
AU - Costaridou, Lena
PY - 2008
Y1 - 2008
N2 - Accurate Microcalcification (MC) segmentation is acrucial first step in morphology based Computer Aided Diagnosissystems for microcalcifications in mammography. In this articlewe present an automated segmentation method of individualMCs adaptive to both size and shape variations. Size is estimatedby active rays (polar-transformed active contours) on continuouswavelet representation while shape adaptivity is achieved by asubsequent region growing step. Following MC seed pointannotation, contour point estimates are obtained byimplementing active rays on an analytic scale-spacerepresentation in a coarse-to-fine strategy. Initial coarsest scale iautomatically defined by analyzing MC responses across scales.A region growing method is used to delineate the final MCcontour curve, with pixel aggregation constrained by the MCcontour point estimates. The segmentation accuracy of theproposed method was quantitatively evaluated by means of areaoverlap by comparing automatically derived borders withmanually traced ones provided by an expert radiologist. The roposed method achieved an area overlap of 0.68±0.13 on adataset of 67 individual microcalcifications, originating frompleomorphic clusters.
AB - Accurate Microcalcification (MC) segmentation is acrucial first step in morphology based Computer Aided Diagnosissystems for microcalcifications in mammography. In this articlewe present an automated segmentation method of individualMCs adaptive to both size and shape variations. Size is estimatedby active rays (polar-transformed active contours) on continuouswavelet representation while shape adaptivity is achieved by asubsequent region growing step. Following MC seed pointannotation, contour point estimates are obtained byimplementing active rays on an analytic scale-spacerepresentation in a coarse-to-fine strategy. Initial coarsest scale iautomatically defined by analyzing MC responses across scales.A region growing method is used to delineate the final MCcontour curve, with pixel aggregation constrained by the MCcontour point estimates. The segmentation accuracy of theproposed method was quantitatively evaluated by means of areaoverlap by comparing automatically derived borders withmanually traced ones provided by an expert radiologist. The roposed method achieved an area overlap of 0.68±0.13 on adataset of 67 individual microcalcifications, originating frompleomorphic clusters.
KW - Active rays
KW - Continuous wavelet transform
KW - Microcalcification segmentation
KW - Region growing
KW - Size-adapted scale-space analysis
UR - http://www.scopus.com/inward/record.url?scp=67549117495&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=67549117495&partnerID=8YFLogxK
U2 - 10.1109/BIBE.2008.4696833
DO - 10.1109/BIBE.2008.4696833
M3 - Conference contribution
AN - SCOPUS:67549117495
SN - 9781424428458
T3 - 8th IEEE International Conference on BioInformatics and BioEngineering, BIBE 2008
BT - 8th IEEE International Conference on BioInformatics and BioEngineering, BIBE 2008
T2 - 8th IEEE International Conference on BioInformatics and BioEngineering, BIBE 2008
Y2 - 8 October 2008 through 10 October 2008
ER -