CT-based cerebral blood flow measures stand up to comparison with PET

CONTEXT: Dynamic perfusion CT is easy to perform and requires no specialized equipment, yet little concrete information exists about the accuracy of its blood flow measures, according to researchers at Osaka City University in Japan. Findings from their head-to-head comparison of cerebral blood flow (CBF) values obtained with perfusion CT and PET were presented in March at the European Congress of Radiology.

RESULTS: The researchers examined 11 patients with various cerebrovascular diseases. Perfusion CT CBF maps were obtained using the central volume principle (relating CBF to blood volume and mean transit time) and a deconvolution method. Regions of interest in the frontal, temporal, parieto-occipital territories, caudate nucleus, thalamus, and white matter in both hemispheres were identified manually for both modalities.

The average CT-based CBF value was 19% higher than the PET-based CBF value in gray matter and 32% lower in white matter. Linear regression analysis revealed a moderate, although statistically significant, correlation between the mean CBF values acquired with the two modalities, however. Correlation coefficients ranged from 0.657 to 0.833 for individual analyses, dipping to 0.533 for the overall comparison.

IMAGE: Freehand-drawn region of interest on CBF maps derived from CT (A) and PET (B) scanning of the same subject. Comparison of paired scans such as these reveals moderate overall correlation of CBF values. (Provided by H. Yokote)

IMPLICATIONS: The differences were attributed to freehand drawing of regions of interest on the CBF maps and to PET's lower spatial resolution compared with CT. The researchers tended to overestimate CBF in gray matter and underestimate it in white matter with PET, said lead author Hiroyuki Yokote, a radiology fellow at Osaka City University. In evaluating arterial input function, the choice of artery can also influence CT-based CBF data in patients with complex blood flow changes.