Thursday, July 03, 2008

Killer SPECT/CT?

OK, wrong Hawkeye, but the title is correct.

To keep at the bleeding edge of nuclear technology, my hospital is in the market for a hybrid SPECT/CT. This latest and greatest idea in nuclear medicine recapitulates the brilliance and innovation that brought us PET/CT, and applies it to PET's cousin SPECT.

I'm going to assume that you know what these modalities are all about. PET uses positron-emitting radioisotopes to produce an image, while SPECT, Single Photon Emission Computed Tomography, uses more conventional radiopharmaceuticals, generally labelled with Technetium 99m, or Indium 111, or Gallium 67. PET is acquired with a ring detector, and SPECT uses two rotating gamma camera heads (sometimes one, rarely three) to produce an image. You really don't want me to get into the mathematics of the process, not that I really could; suffice it to say that by rotating all around the radiation source (the patient), the radioactivity detected by the heads is reassembled into a series of slices. This technology has been around for a long time.

In the early days of PET, there were attempts to modify gamma cameras to act as cheap PET scanners. We had one, the ADAC Vertex, which used thicker NaI (Sodium Iodide) crystals to stop the higher energy annihilation gamma rays (511 KeV vs 140 KeV for the main gamma emission from Technetium). It was a good way to get our feet wet, but the images weren't great.

One of the SPECT gamma-cameras that was marketed as having PET capabilities started life as the Elscint Varicam:

This was a camera well ahead of its time, with body cowling designed by Jaguar (or was it Porsche?). Its computer was a little esoteric, and I'm not sure many were ultimately sold. But with thick crystals, it could be made to do PET imaging.

In 1998, GE bought Elscint, and assimilated the Varicam into its collective product line, where it became the Hawkeye. (As an interesting historical aside, I was actually talking MRI with some of the developers at Elscint HQ in Haifa, Israel, the very week everyone else at Elscint was in Milwaukee completing the deal.)

I think the innovation to put a CT in the same machine as the SPECT gantry came from UCSF, where an old GE 9800 and a GE 600 XR/T SPECT were sort of loosely combined:

I suppose I should stop at this point to answer the question of why one would want to do this. As with PET and its marriage to CT, there are two reasons. First, the human body is inhomogeneous, i.e., it is full of things that block x-rays to varying degrees. Bones block more than watery structures (like organs), which block more than air in the lungs. To properly reconstruct a tomographic, SPECT image, the computer should know what the anatomy looks like. Most of the time in SPECT we ignore this, although in cardiac work, where chest wall or breast or diaphragmatic attenutation (blocking) could create false defects in the image of the heart, attenuation correction is critical. This correction can be accomplished by sweeping a radioactive source across the area in question, sometimes during the scan itself, or one could use CT. CT actually works much better, because it is designed to look at anatomy, not physiology. Second, linking CT and SPECT gives me a better chance to localize those pesky little hot-spots, just like we see with PET/CT.

Anyway, under GE tutilage, the Varicam became the Infinia. A bit later, a low-resolution CT was added to create the Hawkeye:

The "bump" to the right of the ring holds the x-ray tube, and the black strip to the left, inside the ring, is a detector. Thus, a rather limited CT is created. GE continues to sell this device today, and it now has 4-slice capability (Hawkeye Infinia 4).

Philips and Siemens have gone more traditional, with offerings hearkening back to the original UCSF design with a real CT scanner. The Symbia from Siemens uses their Emotion CT scanner (in various slice increments, either one, two, or six, and soon 64) mated with a dual-head gamma camera. Very pretty cowling; you have to love that Bondi-Blue made famous by the original iMacs.

Philips Precedence SPECT/CT mates a Philips CT (which is coincidentally a descendant of an Elscint design as well) to an ADAC (I mean Philips) Skylight gamma-camera:

I've heard that the Skylight, a room-sized framework suspending two gamma camera heads, was commissioned to do bone scans on racehorses, and having two in my lab (Skylights, not racehorses), I can see where this might have been the case.

So, once I get my hands on one of these, what will I do with it? The possibilities are really endless. Anything we do with SPECT, we can do better with SPECT/CT. A lesion on a bone scan can be precisely located with the CT component. The blobs of activity that are almost unreadable from a ProstaScint scan suddenly become diagnostic of nodes invaded by prostate cancer when I can see where they actually lie. Cardiac scans gain accuracy with the improvement in attenuation correction. And so on. But there is one theme common to these new and exciting applications: They depend heavily on the CT component. The nuclear literature is chock-full of reports of improvements in diagnosis when using SPECT/CT, especially with diagnostic quality CT's done with IV and oral contrast.

Which leads us into the editorial part of this opus.

The Medical University 100 miles down the road is getting four Symbias. Yes, I said FOUR. Our community hospital here in the boonies has to be competitive, but with luck, I will get one SPECT/CT scanner, so I had better choose the right one.

Equipment vendors are in the business of selling equipment, oddly enough, and they try hard to convince you that their machine is the best, and the other guy's is trash. When I was looking at PET/CT scanners, GE sent me at least a dozen articles that "proved" their older BGO crystals were better than Siemen's LSO crystals. I didn't fall for it, and I'm happy with my PET/CT choice of the Siemens Biograph. (However, I'm very UNhappy with the fact that Siemens released their higher resolution TruePoint machine within a year of our system's purchase, which can't be retrofit our older model.)

GE knows very well that I'm not a Hawkeye fan, but they continue to try to convince me of the error of my ways.

I'll make the assumption that the gamma camera (i.e., the SPECT) components are similar, and that the attenuation correction on each will yield good images. The only reference I could find indicates that the Hawkeye's SPECT resolution is slightly better than that of the Symbia series.

But.... When my Chief Tech asked about the quality of the Hawkeye's CT, she was told, "Why would you need more than attenuation correction?" Excuse me? Obviously, there is a little lack of understanding of one's product here. Attenuation correction is only part of the story, and as the interpreting radiologist, it is the lesser part. When I see hand-waving like this, I know there is something to hide. And it is indeed the CT component.
The Hawkeye's CT (we'll be generous and call it that) can acquire 4 slices simultaneously, although I'm not sure how fast it rotates. You sure as heck wouldn't want the open gantry rotating beyond a snail's pace, or someone is going to get hurt. Slice thickness is (per RT-Image) 5mm. OK, but the darn thing puts out only 1-2.5 mA with a 350 Watt generator! Compare this to the Symbia T6, with a 6-slice CT, that has a 50 KW (50,000 Watts), that can develop 20-345 mA. The Emotion CT platform can scan down to 0.63 mm. That's a real CT, with diagnostic images for correlation. In a departmental pinch, a Symbia can even be used as a stand-alone CT.

Why do I need more than attenuation correction? Because CT is what SPECT/CT is all about, folks. Yes, attenuation correction is important, and it greatly improves the image. But I need the diagnostic CT, and it needs to register perfectly with the SPECT image. In this way, I can match the anatomy to the physiology, and that is what I need to do. Sadly, GE doesn't want to acknowledge this, mainly because its product doesn't have a diagnostic CT component. No doubt they will couple their gamma camera with a proper CT someday, but until then, this little deficiency doesn't seem to bother them. In fact, their strategy takes a rather different turn:

The following link is to an article regarding the possible link between multiple CTs and cancer risks, especially in children. Aiding us in our cause for our low-dose SPECT/CT Infinia Hawkeye 4 system. Maximum patient dose with the Infinia Hawkeye 4 is 2.5ma. Why more dose for Attenuation Correction and Anatomical Mapping in Nuclear Medicine environment?
Maximum patient dose is 2.5 ma? I'm not quite sure what that is supposed to mean. The initials "mA" stand for milliamperes, a measure of current, or loosely, flow of electricity. The Hawkeye's tube current is indeed rated at 2.5 mA. Yes, this lower-powered tube does yield a lower radiation dose.

The article in the link discusses the supposed danger of too much radiation from too many CT scans, and was inspired by a now-infamous New England Journal article. There has been tremendous debate in our literature, both before and after this somewhat inflammatory piece, concerning the carcinogenic potential of diagnostic radiological procedures in general, and CT in particular. I won't rehash that now. But most agree that it is desireable to keep the amount of radiation to the lowest reasonable level.

GE deserves accolades for its attempts to reduce radiation exposure from its devices, especially in the realm of cardiac CT. But in the SPECT/CT venue, the argument is hollow at best. Because the Hawkeye's CT images are not diagnostic, the patient will have to undergo another CT, and the dose will be added to that from the Hawkeye. In other words, the total dose will exceed what the patient would have received had he or she been scanned on a SPECT/CT unit that had a diagnostic CT in the first place.

Now, being board-certified in both Nuclear Medicine and Radiology, I rather take offense to that last line in the quote from GE about the "Nuclear Medicine Environment". The implication here is that Nuclear Medicine doesn't need diagnostic images. Well, sorry to disappoint you guys, but likely the majority of nuclear medicine exams are read by diagnostic radiologists, and because of PET/CT, most pure Nuc Med physicians are now pretty well versed in reading CT's, so don't even think of going there.

One simply cannot justify the purchase of a SPECT/CT camera on the basis of attenuation correction alone, and that seems to be the majority of what the Hawkeye offers. So, let's not play the radiation dose card, when ultimately a patient scanned on the Hawkeye might get a higher dose than someone scanned on a Symbia or a Precedence. If the extra radiation is so dangerous, and it needs to be limited as much as possible so the scans don't turn out to be harmful (or maybe fatal?), then we need to do only one CT scan and not two, yes?

I'm anxiously awaiting the day that GE delivers a SPECT/CT camera with a real CT in the box. No doubt this will occur. But you will forgive me if I have to pass on the Hawkeye. It just isn't going to be my "killer app".


Several comments to this post have led me to look further into the Philips BrightView XCT SPECT/CT device, which will be available early 2009. From the Philips website, we learn that this device differs significantly from the other contenders:

CoPlanar FP uniquely integrates Philips BrightView SPECT in a co-planar design with advanced Philips flat-detector X-ray CT technology.
This is the only application of flat-panel detectors in this realm that I could find. Sadly, I didn't look closely at this machine at SNM, but RT-Image gives us the CT specs for the Hawkeye, the BrightView, and the Symbia T16:

I don't know how the flat-panel detector will change things, but physics is physics, and the CT appears underpowered relative to the Symbia, although it is significantly better than the Hawkeye. I'll post further when I get to see some images.


mal said...


Yo didn't mention the Philips new Brightview XCT (volumetric SPECT/CT). I was wondering what you thought of this approach.


Anonymous said...

I was at the Philips unveiling of the new Brightview XCT at SNM in NOLA, and it really deserves a look. They have figured out what the market is looking for in SPECT/CT and are able to show some very nice patient images using flat panel volume CT (over 300 thin slices!). From a NM department perspective, reliable AC and decent anatomical information is mainly what's needed to improve diagnosis.

Anonymous said...


The Brightview XCT is a flat panel CT-detector and a cone beam CT and is far away from diagnostic CT quality. The back and forth rotation to get the CT images and the stitching together of 2 half CT scan (small detector!!!) are a patient hazard and increases the chances of artifacts greatly,...!
just my 2 cents


Anonymous said...

The image quality for bone imaging is awesome and as good as any high end CT (with less dose!!). The soft tissue imaging has improved greatly since its first release and is very acceptable. I have used this system and it is extremely patient friendly and the least hazardous system in the market today. I have experienced the Hawkeye and Symbia T2 too. The Hawkeye is very poor quality and cumbersome when compared with the Siemens and Philips systems, and the T2 is a beast ... a noisy one at that! I love the flexibility and intuitiveness of the XCT and the image quality is advanced. It's well worth a look at any tradeshows and a sincere contender for any department. I guess you can tell, I'm now a fan :)