Development of fully automatic technique to parametric images of myocardial blood flow with 15O-water and positron emission tomography using basis function method

Regional myocardial blood flow (MBF) can be measured with I5O-water and PET using the one-tissue compartment model (IT model) with perfusable tissue fraction (PTF), which provides MBF value that is free from the partial volume effect (PVE). Study with I5O-water has several advantages such as the ability of repeated scan. However, the image quality of 15O-water is limited, which impedes the computation of MBF and PTF values at the voxel level. We implemented the basis function method (BFM) for generating parametric images of MBF, PTF and arterial blood volume (Va) with 15O-water and PET. The BFM is to linearize the solution of IT model, which results in a computationally much faster method than the conventional non-linear least squares fitting method (NLM) in estimating the parameters. In order to fully automate the process of the BFM, PET images were realigned to a reference image using Affine transformation. Arterial input function with PVE correction was derived from the realigned PET images using templates of regions of left ventricle and myocardium. In order to validate the BFM, series of PET studies were performed on infarction model pigs (n=4). PET scans with 15O-water were performed with varying doses of adenosine (5-7 scans for each pig). The results by the BFM were well correlated to ones by the NLM. The method automatically generated functional maps computationally fast enough for clinical application.