Abstract #2400

Post processing decisions, independent of institution and fit, influence the perceived diagnostic utility of prostate cancer diffusion maps

Sean D McGarry¹, John D Bukowy², Kenneth Iczkowski³, Allison K Lowman², Anjishnu Banerjee⁴, Kenneth Jacobsohn⁵, Petar Duvnjak², Michael Griffin², Dariya Malyarenko⁶, Tom Chenevert⁶, Yuan Li^7,8, DaeKeun You⁷, Yue Cao^6,7, Andrey Fedorov⁹, Laura C Bell¹⁰, Chad Quarles¹⁰, Melissa A Prah¹, Kathleen M Schmainda¹, Stefanie Hectors¹¹, Bachir Taouli¹¹, Eve LoCastro¹², Yousef Mazaheri^12,13, Amita Shukla-Dave^12,13, Thomas Yankeelov¹⁴, David Hormuth II¹⁴, Ananth J Madhuranthakam¹⁵, Keith Hulsey¹⁵, Mark Muzi¹⁶, Michael Jacobs¹⁷, Meiyappan Solaiyappan¹⁷, William See⁵, Mark Hohenwalter², and Peter S LaViolette¹⁸

¹Biophysics, Medical College of Wisconsin, Milwaukee, WI, United States, ²Radiology, Medical College of Wisconsin, Milwaukee, WI, United States, ³Pathology, Medical College of Wisconsin, Milwaukee, WI, United States, ⁴Biostatistics, Medical College of Wisconsin, Milwaukee, WI, United States, ⁵Urological Surgery, Medical College of Wisconsin, Milwaukee, WI, United States, ⁶Radiology, University of Michigan, Ann Arbor, MI, United States, ⁷Radiation Oncology, University of Michigan, Ann Arbor, MI, United States, ⁸Biomedical Engineering, University of Michigan, Ann Arbor, MI, United States, ⁹Radiology, Brigham and Womens Hospital, Boston, MA, United States, ¹⁰Neuroimaging Research, Barrow Neurological Institute, Phoenix, AZ, United States, ¹¹Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY, United States, ¹²Medical Physics, Memorial Sloan Kettering Cancer Center, New York, NY, United States, ¹³Radiology, Memorial Sloan Kettering Cancer Center, New York, NY, United States, ¹⁴Institute for Computer Engineering and Sciences, University of Texas, Austin, TX, United States, ¹⁵Radiology, University of Texas Southwestern Medical Center, Dallas, TX, United States, ¹⁶Radiology and Neurology, University of Washington, Seattle, WA, United States, ¹⁷Radiology, Johns Hopkins University, Baltimore, MD, United States, ¹⁸Radiology, Medical College of Wisconsin, wauwatosa, WI, United States

Diffusion data from 33 prospectively recruited patients was distributed to 14 sites, who returned monoexponential, biexponential, and diffusion kurtosis fits of the data. Receiver operator characteristic curve AUC was evaluated while varying the positive condition in the ROC analysis, minimum lesion size included, and method of extracting values from the ROI. Cluster size and stratification technique have a significant impact on the AUC and should be explicitly stated in future rad-path studies.

This abstract and the presentation materials are available to members only; a login is required.

Join Here