Fortunately, I have friends in high places. I posed the question to MedGadget, wherein we find another review of the Optima. Somehow, their email was answered, and here is the response:
You can tell him/her we chose the BGO crystal as it delivers a proven track record of speed and performance and is optimized for oncology imaging especially for Fluorine-based tracers.I don't like to parse words, but this basically says that "we've used it before and we're using it again." And the energies of the other PET tracers all peak at 511 KeV, standard for positron/electron annihilation. So what's this about being best for F-18? (As we will see below, F-18 might be best for a machine with BGO crystals, but not vice versa.) If BGO is the end-all to end-all, why not mention it in the extensive Optima product brochure? If BGO is the best choice, pat yourselves on the back for making it.
Personally, I'm not buying any of this. Literally or figuratively. As I mentioned in the initial post, this is a deal-breaker, not that I have the funding to buy a new PET/CT in the immediate future.
I think we can assume that the rationale behind the decision not to go with LYSO as used in the Discovery 690 was cost. BGO is cheaper, and adequate (barely) for the job. (Ummm, by the way, did GE ever actually make any 690's? Did they sell any? Reports from the field appreciated...)
As I've said again and again, LSO beats BGO hands down for positron imaging.
AuntMinnie has a great reference paper on PET fundamentals. Therein, we find this passage (emphasis mine):
Bismuth germinate oxide (BGO) crystals are generally used in conventional PET imaging systems. BGO crystals have a high stopping power (high efficiency), high spatial resolution, and are 50% more efficient than thallium-doped sodium iodide -- NaI (Tl) -- crystals. Most crystals are 3-6 mm thick and they are not hydrophilic. The detection efficiency for 25 mm BGO crystal is approximately 80% . The spatial resolution approaches 5 mm, which nears the theoretical limit of resolution. The disadvantages of BGO crystals are that they have a much lower light output (15% of NaI (Tl) crystals), long photofluorescent decay times (decay constant of 300 ns which limits count rates/coincidence timing resolution ), and poorer energy resolution than sodium-iodide crystals. Energy resolution of BGO is normally worse than 20-25% in FWHM at 511 keV. A typical energy window for a BGO scanner is 300-350 keV to 650 keV . This poor energy resolution makes it difficult to remove scattered events by energy discrimination. Therefore, lead-tungsten septa are interposed between detector rings to reduce interplane scatter . The coincidence time window is normally set for 10-20 ns. The inferior time resolution causes larger accidental detections and greater dead times. BGO detectors are best suited for imaging isotopes with long half-lives such as F-18 and C-11.
Lutetium Oxyorthosilicate (LSO) crystals offer the best combination of properties for PET imaging . LSO has a higher effective Z (number of protons per atom) and density compared to BGO which results in an equal or higher detection efficiency [3,31]. It has very good energy resolution (about 12% - with a typical energy window set to 425-650 keV), a short decay constant for good coincidence timing (a decay constant of 40 ns and coincidence time window of 4.5 ns), and higher light output (five fold more light compared to BGO crystals ) . The coincidence time window is set to 4.5 ns. These characteristics enable image formation is less time when LSO crystals are used . The crystal is rugged and nonhygroscopic.
BGO versus LSO detector: The images below were acquired in 2D mode with a scan time of 5 min/bed emission and 3 min/bed transmission. The upper row of images were acquired on a BGO camera system and the lower row of images were acquired on a ECAT Accel LSO PET camera system. The data was reconstructed using normalized attenuation weighted OS-EM reconstruction. Image courtesy of Northern California PET Imaging Center, Sactamento, CA and CTI, The Power Behind PET.