Original Research Improved Bolus Arrival Time and Arterial Input Function Estimation for Tracer Kinetic Analysis in DCE-MRI Anup Singh, PhD, 1 Ram K. Singh Rathore, PhD, 1 * Mohammad Haris, PhD, 2 Sanjay K. Verma, MSc, 1 Nuzhat Husain, MD, 3 and Rakesh K. Gupta, MD 2 Purpose: To develop a methodology for improved estima- tion of bolus arrival time (BAT) and arterial input function (AIF) which are prerequisites for tracer kinetic analysis of dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) data and to verify the applicability of the same in the case of intracranial lesions (brain tumor and tuber- culoma). Materials and Methods: A continuous piecewise linear (PL) model (with BAT as one of the free parameters) is proposed for concentration time curve C(t) in T 1 -weighted DCE-MRI. The resulting improved procedure suggested for automatic extraction of AIF is compared with earlier methods. The accuracy of BAT and other estimated parameters is tested over simulated as well as experimental data. Results: The proposed PL model provides a good approxi- mation of C(t) trends of interest and fit parameters show their significance in a better understanding and classifica- tion of different tissues. BAT was correctly estimated. The automatic and robust estimation of AIF obtained using the proposed methodology also corrects for partial volume ef- fects. The accuracy of tracer kinetic analysis is improved and the proposed methodology also reduces the time com- plexity of the computations. Conclusion: The PL model parameters along with AIF mea- sured by the proposed procedure can be used for an im- proved tracer kinetic analysis of DCE-MRI data. Key Words: DCE-MRI; BAT, AIF; piecewise linear model; tracer kinetic analysis J. Magn. Reson. Imaging 2009;29:166 –176. © 2008 Wiley-Liss, Inc. DYNAMIC CONTRAST-ENHANCED (DCE) magnetic resonance imaging (MRI) provides noninvasive methods for studying the vasculature of different tissues/lesions based on their response to the passage of intravenously injected contrast agent. The study of tissue vasculature is essential for understanding a wide range of disease processes. DCE-MRI data results in a signal intensity time curve, S(t), at individual voxels that can be con- verted into a concentration time curve, C(t) (1–3). The shape of C(t) is an important criterion for differentia- tion/characterization of different tissues. In the case of an intact blood brain barrier (BBB) the contrast re- mains intravascular. BBB breakdown results in leak- age of contrast into the extracellular extravascular space, which leads to the enhancement of contrast (1). The shape of the curve in these tissues is different from that in normal tissues. A set of well-recognized mathe- matical models is available for T 1 -weighted (W) DCE- MRI data that can provide useful information on tissue vasculature (1– 4). Accurate and precise estimation of the indices of these models is essential for understand- ing tissue vasculature that may help in the diagnosis of different pathologies. With the presently available res- olution and the number of feasible scans the under- standing and the accuracy of the resulting C(t) is still not fully resolved, as issues like compartmentalization of a voxel, effect of water exchange between different compartments of the same voxel, partial volume effect (PVE), inflow of fresh spins (time of flight), etc, continue to be gray areas. Continuous efforts are being made to achieve accurate and precise quantitation of these in- dices and extracting more information (indices) from C(t) by finding newer analysis methods (5–10). The knowledge of accurate arterial input function (AIF) is a prerequisite for fitting tracer kinetic models to C(t). The AIF can be selected manually on major vessels (like carotid artery or middle carotid artery). However, the 1 Department of Mathematics and Statistics, Indian Institute of Tech- nology, Kanpur, India. 2 Department of Radiodiagnosis, Sanjay Gandhi Post Graduate Institute of Medical Sciences, Lucknow, India. 3 Department of Pathology, King George’s Medical University, Lucknow, India. Contract grant sponsor: Department of Science and Technology, New Delhi, India; Contract grant number: SP/SO/HS-50/2002; Contract grant sponsor: Council of Scientific and Industrial Research, New Delhi, India (to A.S.). *Address reprint requests to: R.K.S.R.,. Department of Mathematics and Statistics, Indian Institute of Technology, Kanpur, UP, India. E-mail: rksr@iitk.ac.in Received May 23, 2008; Accepted September 10, 2008. DOI 10.1002/jmri.21624 Published online in Wiley InterScience (www.interscience.wiley.com). JOURNAL OF MAGNETIC RESONANCE IMAGING 29:166 –176 (2009) © 2008 Wiley-Liss, Inc. 166