An alternative approach to estimation of the brain perfusion index for measurement of cerebral blood flow using technetium-99m compounds

Cerebral blood flow (CBF) has been quantified non-invasively using the brain perfusion index (BPI) determined from radionuclide angiographic data generated by technetium-99m hexamethylpropylene amine oxime(^99mTc-HMPAO) or technetium-99m ethyl cysteinate dimer(^99mTc-ECD). The BPI is generally calculated using graphical analysis (GA). In the present study, BPI was measured using spectral analysis (SA), and its usefulness evaluated in comparison with GA. The BPI was calculated from the sum of spectral data obtained by SA. We applied this method to radionuclide angiographic data collected from the bilateral brain hemispheres of 20 patients with various brain diseases using ^99mTc-HMPAO and from those of 20 patients using ^99mTc-ECD. We also measured BPI using GA. The BPI values obtained by SA (BPI^S) (x) and by GA (BPI^G) (y) correlated closely (y=0.708x+0.038, r=0.945 for ^99mTc-HMPAO and y=0.559x+0.093, r=0.931 for ^99mTc-ECD). However, the BPI^G values were underestimated by 22.9%±6.6% (mean±SD) for ^99mTc-HMPAO and by 27.9%±7.5% for ^99mTc-ECD as compared with the BPI^S values. The extent of underestimation tended to increase with increasing BPI^S values. These findings were considered to be a result of the BPI^G values being affected by the first-pass extraction fraction of the tracer. We also compared the BPI^S and BPI^G values with those of CBF measured using N-isopropyl-p-[¹²³I]iodoamphetamine (CBF^IMP) in 16 patients (six for ^99mTc-HMPAO and ten for ^99mTc-ECD). Although both BPI^S and BPI^G values correlated significantly with the CBF^IMP values, the correlation coefficient in BPI^S was always better than that in BPI^G (r=0.869 for ^99mTc-HMPAO and r=0.929 for ^99mTc-ECD in BPI^S, r=0.629 for ^99mTc-HMPAO and r=0.856 for ^99mTc-ECD in BPI^G). These results suggest that SA can provide a more reliable BPI for quantifying CBF using ^99mTc-HMPAO or ^99mTc-ECD than the conventional method using GA. Our method will be useful especially when using a tracer with a low first-pass extraction fraction and/or when performing activation studies using pharmacological intervention. Received 3 November 1998 and in revised form 27 May 1999