The purpose of this study was to develop a quantitative method for myocardial blood flow (MBF) measurement that can be used to derive accurate myocardial perfusion measurements from dynamic multidetector computed tomography (MDCT) images by using a compartment model for calculating the first-order transfer constant (K1) with correction for the capillary transit extraction fraction (E). Six canine models of left anterior descending (LAD) artery stenosis were prepared and underwent first-pass contrast-enhanced MDCT perfusion imaging during adenosine infusion (0.14-0.21 mg/kg/min). K1, which is the first-order transfer constant from left ventricular (LV) blood to myocardium, was measured using the Patlak plot method applied to time-density curve data of the LV blood pool and myocardium. The results were compared against microsphere MBF measurements, and the extraction fraction of contrast agent was calculated. K1 is related to the regional MBF as K 1=EF, E=(1-exp(-PS/F)), where PS is the permeability-surface area product and F is myocardial flow. Based on the above relationship, a look-up table from K1 to MBF can be generated and Patlak plot-derived K 1 values can be converted to the calculated MBF. The calculated MBF and microsphere MBF showed a strong linear association. The extraction fraction in dogs as a function of flow (F) was E=(1-exp(-(0.253F+0.7871)/F)). Regional MBF can be measured accurately using the Patlak plot method based on a compartment model and look-up table with extraction fraction correction from K1 to MBF.