Detectability of simulated brain activation using dual radioisotope SPECT based on size and intensity of the focal hyperactivity

Abstract

Simultaneous single-photon emission computed tomography (SPECT) neuroimaging with both technetium-99m (99mTc) hexamethylpropyleneamine oxime (HMPAO) and iodine-123 (123I) N-isopropyl-iodoamphetamine is a recently introduced method with potential for assessing activation phenomena in the brain. However, there is limited information on the accuracy of the technique for detecting focal cortical sites of neuroactivation. We determined, in vitro, what levels of activation could be detected as a function of the size of the activated region. A Lucite brain phantom was filled with both 123I and 99mTc so as to simulate both a nonactivated state (123I) along with focal sites of activation (99mTc). Simulated activations ranged from 0 to 18% in volumes of 7, 14, 20, and 27 cm3. Imaging was performed with a triple-detector gamma camera using a 10% symmetric window at 140 keV and 10% asymmetric window around 159 keV. No correction was made for gamma cross-talk. To determine whether a simulated activation was "detected," the 99mTc: 123I count ratios in the activated regions were compared by t test with ratios in nonactivated regions of similar volume. Detection sensitivities also were calculated as the fraction of the activated 99mTc: 123I ratios that were greater than the mean + 2 standard deviations of the corresponding nonactivated ratios. All sites of simulated activations of 10% or greater were detected. The detection sensitivity was 100% (95% confidence interval, 90-100%) for the two largest chambers with simulated activations of 13-18%. Activations in the 3-6% range, in the same-sized chambers, were detected with a limited sensitivity (67% with a confidence interval of 45-84%). In the 14-cm3 chamber, simulated activations in the 13-18% range were detected with 90% sensitivity (confidence interval, 74-98%). In general, the detection sensitivity was greater for larger chambers and higher levels of simulated activation. We conclude that the dual-radioisotope technique using triple-detector SPECT systems and low-energy all-purpose (LEAP) collimators should be highly reliable for identifying focal brain activations above 13% that cover at least 14 cm3 of brain cortex. Smaller, less intense sites of activation will be detected with reduced frequency. These conclusions are based on our assessment of only the physical parameters involved in this methodology and other factors (e.g., the possibility that the relation between cerebral radiotracer concentration and regional cerebral blood flow) may affect the results obtained with patients.