Meeting Banner
Abstract #0442

Fully Automated Unsupervised Multi-Parametric Classification of Adipose Tissue Depots in Skeletal Muscle

Alexander Valentinitsch1, Dimitrios C. Karampinos1, Hamza Alizai1, Karupppasamy Subburaj1, Thomas M. Link1, Sharmila Majumdar1

1Department of Radiology and Biomedical Imaging, University of California San Francisco, San Francisco, CA, United States

To introduce and validate a fully automated unsupervised multi-parametric segmentation method of the subcutaneous adipose tissue and muscle region to determine intermuscular adipose tissue (IMAT) and subcutaneous adipose tissue (SAT) volumes based on the images from a quantitative chemical shift-based water/fat separation approach.

Keywords

according accuracy accurate achieve additionally adipose agreement alterations amount amounts applied approaches asymmetry automated background bandwidth bland body bone bones calculation calf certain channel chemical classification closest clustering coil combination complementary conditions confidence contour contouring corr correction cortical cross decomposition degeneration density depots determine diagnostic dice distance distribution dotted ecology edge enabling error evaluated except expressed extract extracted extraction extremities extremity fatty features final fraction fully funded give grade grades ground highest histogram ideal identifying infiltration inherent inner intervals lack like link local makes manual manually mark mask masked masks matrix measure menopausal metabolic middle misclassification modal model modeling muscle muscles obesity operator parametric patterns peak perfect performance plots post presence processing property proposed quantitative radiology reduced reeder reliable reproducible require robust scale scoring section segmentation segmented semi separated separates separating separation served skeletal smallest squares stable still streaks studies subjects threshold tissue trend truth unsupervised validate validation volumes water women