Fig. 3

Error AUC of sPBIF relative to AUC of AIF. The sPBIF was scaled using four blood samples with 5 min interval and scaling method 1. The relative errors are shown as a function of the sampling time of the 1st blood sample, and AUCs are calculated up to the sampling time of the 4th blood sample, e.g. for the data points at 30 min, we scaled the PBIF using blood samples at 30 min, 35 min, 40 min and 45 min, and the AUCs are calculated up to 45 min. Clearly, the best PBIF is based on h₄ and the use of data up to 120 min, as seen by the low AUC errors within 1%, i.e. best MR_FDG estimates