Feasibility of an Ultra-Low-Dose PET Scan Protocol with CT-Based and LSO-TX–Based Attenuation Correction Using a Long–Axial-Field-of-View PET/CT Scanner
Hasan Sari, Mohammadreza Teimoorisichani, Marco Viscione, Clemens Mingels, Robert Seifert, Kuangyu Shi, Michael Morris, Eliot L. Siegel, Babak Saboury, Thomas Pyka, Axel Rominger
Abstract
Long–axial-field-of-view (LAFOV) PET scanners enable substantial reduction in injected radiotracer activity while maintaining clinically feasible scan times. Whole-body CT scans performed for PET attenuation correction can significantly add to total radiation exposure. We investigated the feasibility of an ultra-low-dose PET protocol and the application of a CT-less PET attenuation correction method (lutetium oxyorthosilicate background transmission [LSO-TX]) that uses <sup>176</sup>Lu background radiation from detector scintillators with low-count PET data. <b>Methods:</b> Each of the 4 study subjects was scanned for 90 min using an ultra-low-dose <sup>18</sup>F-FDG protocol (injected activity, 6.7–9.0 MBq) with an LAFOV PET scanner. PET images were reconstructed with different frame durations using low-dose CT-based and LSO-TX–based attenuation maps (μ-maps). The image quality of PET images was assessed by the signal-to-noise ratio (SNR) in the liver and the contrast-to-noise ratio in the brain. Absolute errors in SUVs between PET images reconstructed with LSO-TX–based and CT-based μ-maps were assessed at each scan duration. <b>Results:</b> Visual assessment showed that 20–30 min of PET data obtained using <sup>18</sup>F-FDG activities below 10 MBq (i.e., 0.1 MBq/kg) can yield high-quality images. PET images reconstructed with CT-based and LSO-TX–based μ-maps had comparable SNRs and contrast-to-noise ratios at all scan durations. The mean ± SD SNRs of PET images reconstructed with the CT-based and the LSO-TX–based μ-maps were 9.2 ± 1.9 dB and 9.8 ± 2.0 dB at 90-min scan duration, 6.8 ± 1.7 dB and 6.9 ± 1.8 dB at 30-min scan duration, and 5.5 ± 1.2 dB and 5.6 ± 1.2 dB at 20-min scan duration, respectively. The relative absolute SUV errors between PET images reconstructed with LSO-TX–based and CT-based μ-maps ranged from 3.1% to 6.4% across different volumes of interest with a 20-min scan duration. <b>Conclusion:</b> PET scans with an LAFOV scanner maintained good visual image quality with <sup>18</sup>F-FDG activities below 10 MBq for scan durations of 20–30 min. The LSO-TX–based attenuation correction method yielded images comparable to those obtained with the CT-based attenuation correction method in such protocols.