Attenuation correction in cardiac SPECT without a transmission measurement

The accuracy of SPECT cardiac perfusion imaging is impaired by artifacts induced by nonuniform gamma-ray attenuation. This study proposes a method to estimate attenuation in the cheat of patients without the additional hardware and expense of transmission imaging. Methods: After the standard ²⁰¹Tl or ^99mTc-sestamibi delayed images were obtained, ^99mTc macroaggregated albumin (MAA) was injected and dual-energy SPECT acquisition was performed with windows cantered at 140 keV and 94 keV. Lung contours were obtained by thresholding the on-peak (140 keV) reconstructions. Outer body contours were defined from images produced by reconstruction of the lower energy scatter window obtained simultaneously at the time of the lung (MAA) imaging. Following assignment of standard attenuation values to the lung and nonlung (soft tissue) regions attenuation correction was achieved by means of a modified iterative Chang algorithm. The results were quantitatively evaluated by imaging of a cardiac phantom filled with uniform activity placed in a chest phantom. Sensitivity to the choice of lung and soft tissue attenuation values, the choice of the threshold used for lung segmentation, and errors in registration of the attenuation map were assessed. Results: Application of this technique in a chest phantom and in patients imaged with both ²⁰¹Tl and ^99mTc-sestamibi resulted in improvement in artifactually decreased inferior wall activity without adversely affecting the other walls. The results were relatively insensitive to choice of values for lung and soft- tissue attenuation, lung thresholding, and small (≤ 1.3 cm) registration errors. Conclusion: This simple method corrects for nonuniform attenuation in males; studies are underway to adapt the method to determine breast contour in females and to determine the value of the method in clinical practice.