SEPARATION OF TWO RADIONUCLIDES IN SIMULTANEOUS DUAL- ISOTOPE IMAGING WITH INDEPENDENT COMPONENT ANALYSIS

Simultaneous dual-isotope imaging has contributions from two different energy gamma rays, such as Tc-99m (140KeV) and I-123 (159KeV). Dual-isotope Tc-99m/I-123 imaging has capability in nuclear medicine applications for brain perfusion or presynaptic and postsynaptic dopamine system and for myocardial perfusion system. It reduces acquisition time and makes registration much easier. But the disadvantages of dual-isotope technique are the scatter and cross-talk between the two emitted gamma-ray photons. We proposed a new method to correct for cross-talk from primary and scattered photons using independent component analysis (ICA), which was compared to conventional energy-window method (CONW). In this paper we investigated the performance of ICA to separate Tc-99m and I-123 energy spectra from dual-isotope simulated data to obtain images of each isotope. In Tc-99m images, the estimated image was extracted from mixed dual-isotope imaging using ICA with the percentage bias (P) about -11% and the mean square error (MSE) of 80.48. The counts of Tc-99m were overestimated by CONW method with P < 66% and MSE of 1860.28. ICA could reduce the contamination significantly from I-123 to Tc-99m window. In I-123 images, the cross-talk was disappeared in both methods, but the percentage bias for ICA (< -18%) was less than that of CONW (< -57%). The MSE was 104.63 for the ICA case and 960.15 for the CONW case. The results demonstrated that ICA method improved the quantitative accuracy and can be a promising and effective tool for dual-isotope imaging.