Abstract #4788

Prostate Cancer Diffusion Signal Analysis: Combination of Multiple Fit Parameters Improves Tissue Discrimination

Stephan E. Maier^1,2, Thiele Kobus³, Andriy Fedorov¹, Fredrik Langkilde², Ruth Dunne¹, Robert V. Mulkern⁴, and Clare M. Tempany¹

¹Radiology, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, United States, ²Radiology, Sahlgrenska Academy, Gothenburg University, Gothenburg, Sweden, ³Department of Radiology and Nuclear Medcine, Radbound University Medical Center, Netherlands, ⁴Radiology, Children's Hospital, Harvard Medical School, Boston, MA, United States

Diffusion signals over an extended b-factor range 0-3500 s/mm² were measured with an endorectal coil at 3 Tesla in 56 prostate cancer patients. For each pixel, signal decay fits were computed assuming biexponential, kurtosis, stretched exponential and gamma distribution diffusion signal models. The potential of individual parameters and linear parameter combinations to differentiate normal from cancerous tissue was evaluated with ROC analysis. For the kurtosis and stretched exponential models, single parameters yield the highest AUCs, whereas for the biexponential and gamma distribution models, only combinations of parameters produce the comparably high AUCs.

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

Join Here