Aim: Statistical parametric mapping (SPM) is used worldwide to compare brain perfusion single photon emission computed tomography (SPECT) data. The default template within the SPM package used for SPECT image normalization includes images of a group of healthy subjects studied with [(99m)Tc]HMPAO. Since [(99m)Tc]HMPAO and [(99m)Tc]ECD have shown to distribute differently in SPECT studies, we formulated the hypothesis that comparing set of [(99m)Tc]ECD data normalized by means of a [(99m)Tc]HMPAO template may lead to incorrect results.
Methods: A customized [(99m)Tc]ECD template was built with SPECT and magnetic resonance imaging (MRI) images of 22 neurologically healthy women. Then, two sets of subjects, i.e. a group of patients with very early Alzheimer's disease (eAD) and a matched control group, studied by means of [(99m)Tc]ECD SPECT, were chosen for comparisons. The same statistical approach (t-test between eAD patients and controls and correlation analysis between brain SPECT and a cognitive score) was applied twice, i.e. after normalization with either the default [(99m)Tc]HMPAO template or the customized [(99m)Tc]ECD template.
Results: In the comparison between eAD and controls, a cluster of difference in the posterior cingulate gyrus of both hemispheres was only highlighted when using the customized [(99m)Tc]ECD template, but was missed when using the default [(99m)Tc]HMPAO template. In the correlation between brain perfusion and a cognitive score, the significant cluster was more significant and far more extended, also including the right superior temporal gyrus, using the customized [(99m)Tc]ECD template than using the default [(99m)Tc]HMPAO template.
Conclusion: These data suggest the need of customized, radiopharmaceutical-matched SPECT templates to be used within the SPM package. The present customized [(99m)Tc]ECD template is now freely available on the web.