RT Journal Article SR Electronic T1 Principal component analysis (PCA)-based radionuclide separation in simultaneous dual-isotope 201Tl/123I cardiac imaging JF Journal of Nuclear Medicine JO J Nucl Med FD Society of Nuclear Medicine SP 419P OP 419P VO 48 IS supplement 2 A1 William Erwin A1 Richard Wendt A1 S. Cheenu Kappadath A1 Isis Gayed YR 2007 UL http://jnm.snmjournals.org/content/48/supplement_2/419P.3.abstract AB 1751 Objectives: Dual-isotope 201Tl chloride/123I MIBG scanning allows simultaneous assessment of myocardial perfusion and innervation. However, accuracy is limited by cross-talk between the two radionuclides. This study investigated a PCA-based method of creating 201Tl and 123I photopeak images with suppression of cross-talk contamination. Methods: A six-energy-window (SEW) acquisition was defined, to capture photopeak (123I 159 keV γ, 201Hg X-ray) events, and contamination events (Compton scatter, downscatter, septal penetration, Pb X-rays) present in the photopeak images (in keV: 50-60, 60-80, 80-95, 96-142, 143-175 and 200-300). Scans of a torso phantom containing myocardial inserts filled with either 201Tl, 123I or both, were acquired for 10 minutes into 256×256 matrices, through low-energy (LEHR) and medium-energy (MELP) collimation. For each collimator, PCA of the SEW images from a scan with two spatially separated inserts, each filled with one of the radionuclides, was used to determine the linear combinations of PCs that produced images of either the 201Tl insert or 123I insert alone. PCA was then applied to SEW scans of the same insert filled with either 201Tl alone, 123I alone, or both in roughly equivalent amounts, to assess whether those linear combinations produced separated 201Tl and 123I images. Results: PCA yielded four significant components for both collimators. For LEHR, 201Tl was generated from the mean image + 1.0×PC1 + 1.0×PC2 - 0.3×PC3 - 1.2×PC4, and 123I from the mean + 0.9×PC1 - 2.0×PC2 - 3.0×PC3 + 2.0×PC4. For MELP, the images were mean + 1.0×PC1 + 0.32×PC2 + 1.0×PC4 and mean + 0.13×PC1 - 1.0×PC2, respectively. As expected, for the 201Tl-only scans, reconstructions of 123I contained comparatively few counts, and vice versa. The 201Tl and 123I reconstructions were of comparable intensity for the scans containing both. Compared to the original photopeak images, contamination events appeared substantially reduced. Conclusions: PCA of multiple-energy-window, dual-isotope 201Tl/123I scans appears capable of producing highly-separated and contamination-suppressed 201Tl and 123I images, potentially improving accuracy of the technique.