Meeting Banner
Abstract #3202

Gradient Nonlinearity Correction to Improve ADC Accuracy and Standardization in Breast Cancer Clinical Trials

David C. Newitt1, Ek T. Tan2, Thomas L. Chenevert3, Lisa J. Wilmes1, Suchandrima Banerjee4, Luca Marinelli2, Nola M. Hylton5

1Radiology and Biomedical Imaging, University of California, San Francisco, San Francisco, CA, United States; 2Diagnostics and Biomedical Technologies, GE Global Research, Niskayuna, NY, United States; 3Radiology MRI, University of Michigan, Ann Arbor, MI, United States; 4Applied Science Lab, GE Healthcare, Menlo Park, CA, United States; 5Radiology and Biomedical Imaging, University of California San Francisco, San Francisco, CA, United States

Gradient nonlinearity (GN) is a significant source of error for quantitative diffusion MRI, and the extent of GN varies with the scanner used. This confounds results from multi-center and longitudinal studies required in clinical trials. GN effects are of particular concern in breast imaging, where the anatomy has large offsets from magnet isocenter. Retrospective GN correction (GNC) was evaluated in a multi-center setting with phantoms and in normal and breast cancer subjects. GNC significantly reduced the spatial-dependence of ADC values and improved quantitative accuracy of ADC, which could in turn improve the sensitivity of cancer-detection and cancer-monitoring in diffusion MRI.

Abstract PDF Presentation Video

Keywords

absolute accuracy allows anatomically anterior apart apparent appeared applied arbor arrows artifact artifacts axial axis bias biomedical breast cancer clinical closer coefficient coil common comparing concern configuration configurations confounding consistent corrected correction cover decreased determine developed diagnostics diffusion distance done drawn dual error errors female function generally generate global good grad gradient grants great greatly harmonics human improve improved improvement increasing indicative inferior inherently inter intra known larger linearity magnet maintaining maps matched materials measure measured measures mode much nonlinearity offsets original oscillations park partially percent phantom phantoms placed positions posterior press program quality radiology reduce reduced reduces reduction repeatability reproducibility reproducible residual respectively retrospectively scanner scanners science shorter shot significantly site sites slice slightly smaller source spacing spatial spherical standardization subject subjects superior supported system systems table technologies trace trial trials tumor typical uncorrected utilizing variation variations volunteer water whole yellow zoom