Abstract
Secondary breast angiosarcoma is a rare and highly aggressive malignant cancer, occurring most commonly on the skin surface and could be a result of radiation exposure during primary breast conserving therapy (BCT). Detecting the extent of the lesion is imperative to ensure complete tumor resection and reducing the probability of loco- regional tumor recurrence. Dynamic contrast enhanced MRI (DCE-MRI) is a highly sensitive imaging modality to detect the lesion. However, current quantitative methods of detecting the lesion are sensitive to the imaging parameters, timing of bolus injection, contrast-to- noise ratio, temporal sampling rate etc. We present a novel quantitative method to diagnose and stage angiosarcomas on DCE-MRI. The result of the algorithm has been compared to the radiologist's segmentation and the output of a commercial software, CADstream for 14 cases. Compared to the radiologist's segmentation, the algorithm detected the sarcoma with 96.1% sensitivity and 89.2% accuracy with a mean overlap, measured in terms of the Dice Similarity Coefficient (DSC), of 0.63. In 15.4% (2/13) of the cases, CADstream failed to detect the lesion. For the remaining cases, the algorithm detected the tumor with 91.7% sensitivity and accuracy and a DSC of 0.59, considering the CADstream output as the gold standard. The LDS based algorithm shows potential for automatically segmenting the lesion on DCE-MRI and to further help in planning surgical resection.
© 2013 Optical Society of America
PDF ArticleMore Like This
Michelle Heijblom, Daniele Piras, Frank M. van den Engh, Joost M. Klaase, Mariël Brinkhuis, Wiendelt Steenbergen, and Srirang Manohar
880004 European Conference on Biomedical Optics (ECBO) 2013
Haoyu Wang and Shanglian Bao
83110Y Asia Communications and Photonics Conference and Exhibition (ACP) 2011
David R. Busch, Wensheng Guo, Regine Choe, Turgut Durduran, Michael D. Feldman, Carolyn Mies, Brian J. Czerniecki, Julia Tchou, Angela DeMichele, Mark A. Rosen, Michael D. Schnall, and Arjun G. Yodh
BSu5A.7 Biomedical Optics (BIOMED) 2012