Pharmacokinetic perfusion curves estimation for liver tumor diagnosis from DCE-MRI

Dynamic-Contrast Enhanced MRI (DCE-MRI) is a method to analyze the perfusion dynamics in the tissues. The contrast agent concentration along the time, after the bolus injection, depends on the type of tissue observed, namely on its vascularization density and metabolic activity. The number of acquired volumes in this type of exam is usually very small, typically < 10, and the volumes are misaligned due to respiratory and cardiac activities. In this paper an algorithm to automatically characterize the malignancy of the tumor is presented based on the perfusion curves on each voxel of the tumor, obtained from DCE-MRI. A non-rigid registration procedure based on Mutual Information (MI) criterion is used to align the small volumes representing the region of interest (ROI) containing the tumor along the time. A pharmacokinetic (PK) third order linear model is estimated from the observations and its parameters are used to classify the malignancy of tumor.