Clinical validation of SPECT attenuation correction using x-ray computed tomography–derived attenuation maps: Multicenter clinical trial with angiographic correlation

Abstract

Nonuniform attenuation artifacts cause suboptimal specificity of stress single photon emission computed tomography (SPECT) myocardial perfusion images. In phantoms, normal subjects, and patients suspected of having coronary artery disease (CAD), we evaluated a new hybrid attenuation correction (AC) system that combines x-ray computed tomography (CT) with conventional stress SPECT imaging. The effect of CT-based AC was evaluated in phantoms by assessing homogeneity of normal cardiac inserts. AC improved homogeneity of normal cardiac phantoms from 11% +/- 2% to 5% +/- 1% (P < .001). Attenuation-corrected normal patient files were created from 37 normal subjects with a low likelihood (<3%) of CAD. The diagnostic performance of AC for detection of CAD was evaluated in 118 patients who had stress technetium 99m sestamibi or tetrofosmin stress SPECT imaging and coronary angiography. SPECT images with and without AC were interpreted by 4 blinded readers with different interpretative attitudes. Overall, AC improved the diagnostic performance of all readers, particularly the normalcy rate. The degree of improvement depended on interpretative attitude. Readers prone to high sensitivity or with less experience had the greatest gain in the normalcy rate, whereas a reader prone to higher specificity had improvements in sensitivity and specificity but not the normalcy rate. Importantly, improvement of one diagnostic variable was not associated with worsening of other variables. CT-based AC of SPECT images consistently improved overall diagnostic performance of readers with different interpretive attitudes and experience. CT-based AC is well suited for routine use in clinical practice.