Matthew R. Orton1, Keiko Miyazaki1, Dow-Mu Koh2, David J. Collins1, David Atkinson3, David J. Hawkes3, Martin O. Leach1
1Clinical Magentic Resonance Research Group, Institute of Cancer Research, Sutton, Surrey, UK; 2Academic Department of Radiology, Royal Marsden Hospital, Sutton, Surrey, UK; 3Centre for Medical Image Computing, Univeristy College London, London, UK
Kinetic modelling of DCE-MRI liver data requires portal and arterial input functions. Patient specific portal input functions are necessary for accurate kinetic estimation, but direct measurement is challenging, and often highly variable. We present a methodology whereby the portal input function is derived from the arterial input function by applying a delay and a dispersion factor to the first-pass bolus. These parameters are estimated from the tissue data itself, thus avoiding many of the pitfalls of direct measurement. Results from 10 cases of patients with liver metastases are presented demonstrating a correlation between the estimated portal delay and dispersion.