Attenuation correction for cardiac dual-head γ camera coincidence imaging using segmented myocardial perfusion SPECT

The diagnostic accuracy of cardiac FDG imaging obtained with the dual-head coincidence gamma camera (DHC) is impaired by artifacts induced by nonuniform attenuation. This study proposed a new method (registration and segmentation method for attenuation correction [AC-RS]) to correct these attenuations in the chest region without the need for additional hardware or expensive transmission scanning equipment. Methods: Before DHC imaging, Tc-99m-tetrofosmin SPECT was performed using dual-energy acquisition from both the photopeak and Compton scatter windows. The scatter window images of the Tc-99m-tetrofosmin were then registered 3-dimensionally with the cardiac DHC images and segmented into anatomic regions to obtain body and lung contours by applying the optimal threshold method on localized histograms. Theoretic attenuation coefficient Values were assigned to the corresponding anatomic regions, and the DHC emission images were reconstructed using these attenuation correction factors. The results were quantitatively evaluated by imaging a cardiac phantom filled with a uniform solution and placed in a chest phantom. Eight nondiabetic subjects were also examined using this technique, and the results were compared with those of measured attenuation-corrected PET images. Results: Use of this technique in phantom and clinical studies decreased the degree of artifacts seen in the inferior wall activity and corrected the emission images. When the results were compared with those of PET scans, the regional relative counts of the uncorrected DHC scan did not correlate with the results of the PET scan. However, the regional relative counts of the AG-RS-corrected DHC scan exhibited a linear correlation with the results of the PET scan (r = 0.73; P < 0.001), Conclusion: Reasonably accurate attenuation-corrected cardiac DHC images can be obtained using AC-RS without the need for transmission scanning.