José A Rosado-Toro1, Tomoe Barr2, Jean-Philippe Galons3, Marilyn T Marron4, Alison Stopeck5, Cynthia Thomson4, Patricia Thompson6, Danielle Carroll3, Eszter Wolf3, María I Altbach7, Jeffrey J Rodríguez1. 1. Department of Electrical and Computer Engineering, University of Arizona, Tucson, Arizona 85721. 2. Department of Biomedical Engineering, University of Arizona, Tucson, Arizona 85721. 3. Department of Medical Imaging, University of Arizona, Tucson, AZ 85724. 4. Arizona Cancer Center, Tucson, Arizona 85721. 5. Arizona Cancer Center, Tucson, Arizona 85721; Department of Medicine, University of Arizona, Tucson, Arizona 85724. 6. Arizona Cancer Center, Tucson, Arizona 85721; Department of Cellular and Molecular Medicine, University of Arizona, Tucson, Arizona 85721. 7. Department of Medical Imaging, University of Arizona, Tucson, AZ 85724. Electronic address: maltbach@radiology.arizona.edu.
Abstract
RATIONALE AND OBJECTIVES: To develop and test an algorithm that outlines the breast boundaries using information from fat and water magnetic resonance images. MATERIALS AND METHODS: Three algorithms were implemented and tested using registered fat and water magnetic resonance images. Two of the segmentation algorithms are simple extensions of the techniques used for contrast-enhanced images: one algorithm uses clustering and local gradient (CLG) analysis and the other algorithm uses a Hessian-based sheetness filter (HSF). The third segmentation algorithm uses k-means++ and dynamic programming (KDP) for finding the breast pixels. All three algorithms separate the left and right breasts using either a fixed region or a morphological method. The performance is quantified using a mutual overlap (Dice) metric and a pectoral muscle boundary error. The algorithms are evaluated against three manual tracers using 266 breast images from 14 female subjects. RESULTS: The KDP algorithm has a mean overlap percentage improvement that is statistically significant relative to the HSF and CLG algorithms. When using a fixed region to remove the tissue between breasts with tracer 1 as a reference, the KDP algorithm has a mean overlap of 0.922 compared to 0.864 (P < .01) for HSF and 0.843 (P < .01) for CLG. The performance of KDP is very similar to tracers 2 (0.926 overlap) and 3 (0.929 overlap). The performance analysis in terms of pectoral muscle boundary error showed that the fraction of the muscle boundary within three pixels of reference tracer 1 is 0.87 using KDP compared to 0.578 for HSF and 0.617 for CLG. Our results show that the performance of the KDP algorithm is independent of breast density. CONCLUSIONS: We developed a new automated segmentation algorithm (KDP) to isolate breast tissue from magnetic resonance fat and water images. KDP outperforms the other techniques that focus on local analysis (CLG and HSF) and yields a performance similar to human tracers.
RATIONALE AND OBJECTIVES: To develop and test an algorithm that outlines the breast boundaries using information from fat and water magnetic resonance images. MATERIALS AND METHODS: Three algorithms were implemented and tested using registered fat and water magnetic resonance images. Two of the segmentation algorithms are simple extensions of the techniques used for contrast-enhanced images: one algorithm uses clustering and local gradient (CLG) analysis and the other algorithm uses a Hessian-based sheetness filter (HSF). The third segmentation algorithm uses k-means++ and dynamic programming (KDP) for finding the breast pixels. All three algorithms separate the left and right breasts using either a fixed region or a morphological method. The performance is quantified using a mutual overlap (Dice) metric and a pectoral muscle boundary error. The algorithms are evaluated against three manual tracers using 266 breast images from 14 female subjects. RESULTS: The KDP algorithm has a mean overlap percentage improvement that is statistically significant relative to the HSF and CLG algorithms. When using a fixed region to remove the tissue between breasts with tracer 1 as a reference, the KDP algorithm has a mean overlap of 0.922 compared to 0.864 (P < .01) for HSF and 0.843 (P < .01) for CLG. The performance of KDP is very similar to tracers 2 (0.926 overlap) and 3 (0.929 overlap). The performance analysis in terms of pectoral muscle boundary error showed that the fraction of the muscle boundary within three pixels of reference tracer 1 is 0.87 using KDP compared to 0.578 for HSF and 0.617 for CLG. Our results show that the performance of the KDP algorithm is independent of breast density. CONCLUSIONS: We developed a new automated segmentation algorithm (KDP) to isolate breast tissue from magnetic resonance fat and water images. KDP outperforms the other techniques that focus on local analysis (CLG and HSF) and yields a performance similar to human tracers.
Authors: Thorsten A Bley; Oliver Wieben; Christopher J François; Jean H Brittain; Scott B Reeder Journal: J Magn Reson Imaging Date: 2010-01 Impact factor: 4.813
Authors: Snehal M Pinto Pereira; Valerie A McCormack; John H Hipwell; Carol Record; Louise S Wilkinson; Sue M Moss; David J Hawkes; Isabel dos-Santos-Silva Journal: Cancer Epidemiol Biomarkers Prev Date: 2011-06-21 Impact factor: 4.254
Authors: Zhiqiang Li; Christian Graff; Arthur F Gmitro; Scott W Squire; Ali Bilgin; Eric K Outwater; Maria I Altbach Journal: Magn Reson Med Date: 2009-06 Impact factor: 4.668
Authors: Michael Khazen; Ruth M L Warren; Caroline R M Boggis; Emilie C Bryant; Sadie Reed; Iqbal Warsi; Linda J Pointon; Gek E Kwan-Lim; Deborah Thompson; Ros Eeles; Doug Easton; D Gareth Evans; Martin O Leach Journal: Cancer Epidemiol Biomarkers Prev Date: 2008-09 Impact factor: 4.254
Authors: Maria E C Sandberg; Jingmei Li; Per Hall; Mikael Hartman; Isabel dos-Santos-Silva; Keith Humphreys; Kamila Czene Journal: Breast Cancer Res Date: 2013 Impact factor: 6.466
Authors: Leon Lenchik; Laura Heacock; Ashley A Weaver; Robert D Boutin; Tessa S Cook; Jason Itri; Christopher G Filippi; Rao P Gullapalli; James Lee; Marianna Zagurovskaya; Tara Retson; Kendra Godwin; Joey Nicholson; Ponnada A Narayana Journal: Acad Radiol Date: 2019-08-10 Impact factor: 3.173
Authors: Cynthia A Thomson; H H Sherry Chow; Betsy C Wertheim; Denise J Roe; Alison Stopeck; Gertraud Maskarinec; Maria Altbach; Pavani Chalasani; Chuan Huang; Meghan B Strom; Jean-Philippe Galons; Patricia A Thompson Journal: Breast Cancer Res Treat Date: 2017-05-30 Impact factor: 4.872
Authors: Jie Ding; Alison T Stopeck; Yi Gao; Marilyn T Marron; Betsy C Wertheim; Maria I Altbach; Jean-Philippe Galons; Denise J Roe; Fang Wang; Gertraud Maskarinec; Cynthia A Thomson; Patricia A Thompson; Chuan Huang Journal: J Magn Reson Imaging Date: 2018-04-06 Impact factor: 4.813
Authors: Patricia A Thompson; Chuan Huang; Jie Yang; Betsy C Wertheim; Denise Roe; Xiaoyue Zhang; Jie Ding; Pavani Chalasani; Christina Preece; Jessica Martinez; H-H Sherry Chow; Alison T Stopeck Journal: Clin Cancer Res Date: 2021-06-10 Impact factor: 12.531