Improved spillover correction model to quantify myocardial blood flow by ¹¹C-acetate PET: comparison with ¹⁵O-H₂O PET

Abstract

Objective

¹¹C-acetate has been applied for evaluation of myocardial oxidative metabolism and can simultaneously estimate myocardial blood flow (MBF). We developed a new method using two-parameter spillover correction to estimate regional MBF (rMBF) with ¹¹C-acetate PET in reference to MBF derived from ¹⁵O-H₂O PET. The usefulness of our new approach was evaluated compared to the conventional method using one-parameter spillover correction.

Methods

Sixty-three subjects were examined with ¹¹C-acetate and ¹⁵O-H₂O dynamic PET at rest. Inflow rate of ¹¹C-acetate (K1) was compared with MBF derived from ¹⁵O-H₂O PET. For the derivation, the relationship between K1 and MBF from ¹⁵O-H₂O was linked by the Renkin-Crone model in 20 subjects as a pilot group. One-parameter and two-parameter corrections were applied to suppress the spillover between left ventricular (LV) wall and LV cavity. Validation was set using the other 43 subjects’ data. Finally, rMBFs were calculated using relational expression derived from the pilot-group data.

Results

The relationship between K1 and MBF derived from ¹⁵O-H₂O PET was approximated as K1 = [1–0.764 × exp(−1.001/MBF)] MBF from the pilot data using the two-parameter method. In the validation set, the correlation coefficient between rMBF from ¹¹C-acetate and ¹⁵O-H₂O demonstrated a significantly higher relationship with the two-parameter spillover correction method than the one-parameter spillover correction method (r = 0.730, 0.592, respectively, p < 0.05).

Conclusion

In ¹¹C-acetate PET study, the new two-parameter spillover correction method dedicated more accurate and robust myocardial blood flow than the conventional one-parameter method.

Appendix: 1-parameter and 2-parameter spillover method

Relational expression among R(t), C _t(t), and C_a(t) was represented as Eq. 3 for the conventional method and as Eq. 4 for the two-parameter method.

$$ R\left( t \right) \, = \, \alpha \, \times \, C_{\text{t} } \left( t \right) \, + \, \left( {1 - \alpha } \right) \, \times \, C_{\text{a} } \left( t \right) $$

(3)

where α denoted myocardial tissue ratio in the ROI R, and (1–α) is spillover from blood into the ROI R.

$$ R\left( t \right) \, = \, \alpha \, \times \, C_{\text{t} } \left( t \right) \, + \, v\alpha \, \times \, C_{\text{a} } \left( t \right) $$

(4)

where vα is spillover from blood into the ROI R for the two-parameter method.

Activity concentration in the LV blood cavity was modeled as a partial-volume mixture of arterial blood and myocardial tissue as Eq. 5.

$$ \text{LV} \left( t \right) \, = \, \beta \, \times \, C_{\text{a} } \left( t \right) \, + \, \left( {1 - \beta } \right) \, \times \, m \, \times \, C_{\text{t} } \left( t \right) $$

(5)

where β denoted a recovery coefficient in the LV ROI, (1–β) is spillover from myocardium into blood pool, and m is the density of myocardial tissue (1.04 g/ml). Radioactivity in the LV blood pool was calculated using Eq. 5 with β = 85 % [4].

The change in tissue activity concentration was modeled using the one-tissue compartment model as Eq. 6.

$$ dC_{\text{t} } \left( t \right) \, = \, K1 \, \times \, C_{\text{a} } \left( t \right) \, - \, k2 \, \times \, C_{\text{t} } \left( t \right) $$

(6)

where K1 (mL/g/min) is the uptake rate from blood into the tissue and k2 (/min) is the washout rate from myocardial tissue into the blood C _a(t) (Bq/mL). The parameters K1, k2, α and vα were estimated by the non-linear least-squares method using Eqs. 3, 5, and 6 for the one-parameter method and Eqs. 4, 5, and 6 for the two-parameter method. Estimated data and the curve R(t) dedicated the spillover-corrected pure blood curve C _a(t).