Analysis of myocardial 82Rb PET studies
Background
Positron emission tomography enables noninvasive quantification of myocardial perfusion. PET tracers that are used for estimation of blood flow include diffusible tracer [15O]H2O and partially extracted tracers 82Rb and [13N]ammonia. During one capillary pass, monovalent potassium analogue 82Rb ([82Rb]Rb+) is only partially extracted by the myocardial cells via the Na+/K+ adenosine triphosphatase pump, and extraction is inversely and nonlinearly proportional to perfusion. Furthermore, extraction and retention, at a given perfusion level may be affected by drugs or severe acidosis, hypoxia and ischemia.In 82Rb studies the scanner may be near its saturation limit, which limits the amount of tracer that can be injected. Together with the short physical half-life, this leads to a low signal-to-noise ratio. That has to be taken into account, when deciding between 2D and 3D scanning (Votaw & White, 2001). Model selection can be misguided and analysis may lead to biased results, if scanner is saturated or dead-time correction is not working properly. The relatively high positron energy of 82Rb leads to relatively poor image quality and reduced spatial resolution.
Analysis methods used in literature
Retention model
Retention model is simple to calculate, but requires an extraction correction (Herrero et al., 1990; Yoshida et al., 1996) and the result is dependent on the time of imaging the tissue concentration. Maximum flow that can be measured with this method is about 2-2.5 ml/(g * min) (Herrero et al., 1990).Note that retention model accounts for the efflux from tissue to blood (k2) only by keeping the calculation time as short as possible and by applying nonlinear extraction correction. Therefore the extraction correction formula is different for different PET scan times and injection protocols, and is not the same as may be determined for K1 from compartment models
Compartment models
Physiological model for 82Rb in myocardium contains three tissue compartments, capillary space, interstitial space, and intracellular space (Coxson et al., 1997). This model, and the reduced model with two tissue compartments (Herrero et al., 1992), cannot be applied to noisy PET data without a priori values for some of the model parameters, even if the signal-to-noise ratio of PET images is improved using wavelet-based noise reduction protocol (Lin et al., 2001). When recovery coefficient and distribution volume of the first tissue compartment were fixed, good repeatability can be achieved (Knešaurek et al., 2009).
The further reduced one-tissue compartment model (Coxson et al., 1995; Golanowski et al., 2000) provides results with good reproducibility in rest and hyperemic conditions ()Manabe et al., 2009), and is therefore better suited for clinical setting. The result of one- or two-tissue compartment models (K1) is dependent on both extraction and blood flow, and requires (non-linear) extraction correction, if quantitative flow estimates are needed (Lortie et al., 2007).
El Fakhri's approach with factor analysis
El Fakhri et al. (2005) applied generalized form of least-squares factor analysis of dynamic sequences (GFADS) to generate left and rightventricular (LV and RV) TACs. A 1-tissue compartment model was used to estimate the flow and extraction dependent parameters k1 and k2. The contribution from both LV and RV blood to the myocardial activity was taken into account in the model as fitted parameters fiv and riv. The model was applied to TACs calculated as average of groups of voxels. Voxels were grouped with orthogonal grouping with predetermined number of groups (100).Correction for spillover and partial volume effects
Commonly, the spillover and partial volume effects are taken into account in compartment models for 82Rb by assuming a geometrical model.Simulations
Herrero et al (1992) and Meyer et al. (2007) have published input functions for 82Rb simulation studies. Coxson et al. (1997) provide compartment model parameters for simulations.Suggested analysis method
Our current suggestion is to apply one-tissue compartment model withgeometrical model correction for spillover and partial volume effects. This method is implemented in CarimasTurku.To calculate quantitative perfusion values, a nonlinear extraction correction is needed. Parameters for the correction functions may need to be determined for each institute, unless the study protocols are similar. To start with, CarimasTurku implements the extraction correction function from (Yoshida et al., 1996).
