Use of singular value decomposition to characterize age and gender differences in SPECT cerebral perfusion

The main objective of this study was to characterize changes in brain perfusion associated with normal aging and gender.

Perfusion SPECT images using 99mTc-hexamethyl propyleneamine oxime (HMPAO) were obtained from 152 healthy subjects (67 men, 85 women) aged 50-92 yr. An automated method was developed to objectively assess image data from a large number of brain regions. Image data were reduced with singular value decomposition (SVD), which produced 20 eigenvectors capturing 97.05% of the total information content of 4320 regions from each subject. Subjects were scored individually on each vector.

Multivariate analyses demonstrated that there were no significant differences in whole-brain HMPAO uptake with age, but age-related regional declines were seen in lateral ventricular regions. Women had higher HMPAO uptake than men in estimates of global perfusion and regional perfusion in the midcingulate/corpus callosum, inferior temporal and inferior parietal areas.

These discriminations demonstrate that singular value deomposition of SPECT data may be used to assess differences in perfusion patterns between groups of subjects. They replicate several previous findings, both with respect to age-related changes in perfusion and with respect to gender differences. In addition, they identify a previously unreported gender difference in biparietal regions.