PT - JOURNAL ARTICLE AU - Beverley Holman AU - Brian Hutton AU - Kris Thielemans TI - Method to determine the voxel-wise blood volume in the lung from dynamic PET data DP - 2017 May 01 TA - Journal of Nuclear Medicine PG - 1310--1310 VI - 58 IP - supplement 1 4099 - http://jnm.snmjournals.org/content/58/supplement_1/1310.short 4100 - http://jnm.snmjournals.org/content/58/supplement_1/1310.full SO - J Nucl Med2017 May 01; 58 AB - 1310Objectives: Accurate determination of pulmonary blood volume (VB) is important for deriving stable and reproducible static and kinetic parameter estimates in the lung [1]. VB estimates also yield useful diagnostic and therapeutic information such as providing a marker for disease progression. Standard methods of calculating VB from dynamic PET data result in global (whole lung) or regional estimates. However, in lung diseases such as idiopathic pulmonary fibrosis (IPF), where the quantity of blood reaching regions of diseased lung is poorly understood, determination of VB on a local (voxel-wise) level would be advantageous. Here we describe and test a methodology to accurately determining voxel-wise VB using PET/CT kinetic analysis.Methods: 5 patients with diagnosed IPF underwent dynamic FDG PET/CT. An image derived input function (IDIF) was determined in 6 different regions in each patient to allow determination of variability in VB estimates from imperfect IDIFs. Then, for each patient, 100 voxels from within the lung were chosen at random. The time activity curves from each voxel were fit to each of 3 different models (all accounting for time delay): directly to the IDIF, an irreversible 1-tissue compartment model and a reversible 1-tissue compartment model. This analysis was repeated for different durations of the scan time to allow measurement of parameter stability and appropriate scan length. Blood volume maps were determined and the voxel values compared with and without time delay to confirm the requirement of calculating blood volume at the voxel level in the lung. Comparison of the blood volume per unit tissue was determined in normal appearing tissue and fibrotic regions in the IPF lung as determined on CT.Results: The variation in blood volume determined from the different IDIFs was <10%, well within the normal errors of lung imaging. Determination of stable VB at the voxel level requires use of a 1-tissue reversible compartment model which accounts for time delay with a dynamic acquisition which lasts a minimum of 5mins. Time delay between IDIF and different voxels varied considerably within the IPF lung (maximum difference 17s), confirming our previous findings [2]. The VB in the fibrotic and normal appearing tissue of the IPF lung was found to be 0.22±0.11 and 0.42±0.10 respectively.Conclusion: Calculating stable and reproducible VB in the lung requires use of a 1-tissue reversible compartment model which accounts for time delay and must be performed on a voxel basis. The location used for the thorax IDIF has little effect on the variability of VB. The fibrotic lung regions appear to have a lower VB than the normal appearing tissue and this may have an effect on the choice of treatment pathway. Research Support: We acknowledge funding support from GlaxoSmithKline to UCL (STU100028576 and COL29165). This project is also supported by researchers at the National Institute for Health Research, University College London Hospitals Biomedical Research Centre.