Hybrid image and blood sampling input function for quantification of small animal dynamic PET data

Kooresh I. Shoghi, Michael J. Welch

Research output: Contribution to journalArticlepeer-review

33 Scopus citations


We describe and validate a hybrid image and blood sampling (HIBS) method to derive the input function for quantification of microPET mice data. The HIBS algorithm derives the peak of the input function from the image, which is corrected for recovery, while the tail is derived from 5 to 6 optimally placed blood sampling points. A Bezier interpolation algorithm is used to link the rightmost image peak data point to the leftmost blood sampling point. To assess the performance of HIBS, 4 mice underwent 60-min microPET imaging sessions following a 0.40-0.50-mCi bolus administration of 18FDG. In total, 21 blood samples (blood-sampled plasma time-activity curve, bsPTAC) were obtained throughout the imaging session to compare against the proposed HIBS method. MicroPET images were reconstructed using filtered back projection with a zoom of 2.75 on the heart. Volumetric regions of interest (ROIs) were composed by drawing circular ROIs 3 pixels in diameter on 3-4 transverse planes of the left ventricle. Performance was characterized by kinetic simulations in terms of bias in parameter estimates when bsPTAC and HIBS are used as input functions. The peak of the bsPTAC curve was distorted in comparison to the HIBS-derived curve due to temporal limitations and delay in blood sampling, which affected the rates of bidirectional exchange between plasma and tissue. The results highlight limitations in using bsPTAC. The HIBS method, however, yields consistent results, and thus, is a substitute for bsPTAC.

Original languageEnglish
Pages (from-to)989-994
Number of pages6
JournalNuclear Medicine and Biology
Issue number8
StatePublished - Nov 2007


  • Blood input function
  • Image quantification
  • Small animal PET imaging


Dive into the research topics of 'Hybrid image and blood sampling input function for quantification of small animal dynamic PET data'. Together they form a unique fingerprint.

Cite this