Meeting Banner
Abstract #3873

Classical and Knowledge-Based Pharmacokinetic Model Selection Techniques in Analysis of Dynamic Contrast Enhanced MRI Studies: Performance and Bias Comparison

Hassan Bagher-Ebadian 1,2 , Mohammadreza Mohammadian-Behbahani 3,4 , Azimeh Noorizadeh Vahed Dehkordi 3,5 , James R Ewing 2,6 , Alireza Kamali-Asl 3 , Siamak P Nejad-Davarani 7 , Hamed Moradi 8 , Stephen Brown 2,9 , Brent Griffith 10 , Ali S Arbab 11 , Tom Mikkelsen 12 , Lisa Scarpace 12 , and Hamid Soltanian-Zadeh 1,13

1 Radiology and Research Administration, Henry Ford Hospital, Detroit, Michigan, United States, 2 Physics, Oakland University, Rochester, Michigan, United States, 3 Nuclear Engineering, Shahid Beheshti University, Tehran, Iran, 4 Nuclear Engineering, Amir-Kabir University of Technology, Tehran, Iran, 5 Nuclear Engineering, Najaf Abad Branch, Islamic Azad University, Isfahan, Iran, 6 Neurology, Henry Ford Hospital, Detroit, Michigan, United States, 7 Neurology, Henry Ford Hospital, Michigan, Iran, 8 Nuclear Engineering, Shiraz University, Shiraz, Fars, Iran, 9 Radiation Oncology, Henry Ford Hospital, Detroit, Michigan, United States, 10 Radiology, Henry Ford Hospital, Detroit, Michigan, United States, 11 GRU Cancer Center, Georgia Regents University, Atlanta, Georgia, United States, 12 Neurosurgery, Henry Ford Hospital, Detroit, Michigan, United States, 13 CIPCE, School of Electrical and Computer Engineering, University of Tehran, Tehran, Iran

Given Dynamic-Contrast-Enhanced MRI data, accurate estimation of Pharmacokinetic (PK) parameters strongly relies on appropriate selection of the best PK model to fit the data. This study investigates the impact of different classical and adaptive MS technique such as F-Statistic (F-test), Akaike-Information-Criterion (AIC), Bayesian-Information-Criterion (BIC), Log-Likelihood-Ratio (LLR) and Artificial Neural Network (ANN) on estimation of vascular permeability parameters. Results imply that ANN generates significantly less-biased estimates of PK parameters compared to other techniques while both the LLR and BIC methods outperform the other classical MS techniques, the ANN, LLR and BIC are the best candidates for PK analysis of DCE-MRI data.

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

Join Here