[Validation study in quantitative measurement of regional cerebral blood flow using N-isopropyl-p-[123I]iodoamphetamine (123I-IMP) and SPECT].
We validated following five methods to quantitate regional cerebral blood flow (rCBF) using 123I-IMP and SPECT; 1) microsphere method, which is based on a microsphere model using the SPECT image at five minutes after 123I-IMP injection with continuous withdrawal of arterial blood, 2) microsphere+ one-point sampling method, which is the same as the microsphere method except for using one-point sampling instead of continuous withdrawal, 3) conventional microsphere + one-point sampling method, which is the same as the microsphere + one-point sampling method except for using a later SPECT image corrected with the ratio of alteration of measured entire brain activity, 4) Table look-up method, which is based on a two-compartment model (influx; K1 and outflux; k2) using one arterial blood sample taken at 10 min and two SPECT images at 30 min and 180 min post-injection, and 5) functional IMP SPECT, which is based on the two-compartment model using one arterial blood sample taken at 5 min and two SPECT images at 30 min and 60 min post-injection. Those methods were applied to six patients with cerebral infarction and degenerative diseases, and rCBF results were compared with those estimated by non-linear least squares fitting (NLLSF) analysis based on the two-compartment model. The rCBF values obtained by the microsphere method was best correlated with those by NLLSF analysis (r = 0.940), followed by the microsphere+one-point sampling method (r = 0.885) and the functional IMP SPECT (r = 0.882). The table look-up method underestimated rCBF especially at the high flow level, however showed good correlation (r = 0.859). The conventional microsphere + one-point sampling method overestimated rCBF, however showed good correlation (r = 0.849). Distribution volume (Vd = K1/k2) was also estimated by both the table look-up method and the functional IMP SPECT. While values of Vd by the functional IMP SPECT were significantly correlated with those by NLLSF analysis (r = 0.785), the table look-up method overestimated Vd (43.4 +/- 6.6 ml/g) and showed not good correlation (r = 0.287).