Medical Imaging 101 pt 7: SPECT

Medical Imaging 101 pt 7: SPECT

I know it’s been a long time since I added to my #CHMedicalImagingSeries . I’ve been whittling away at this for a few weeks and finally decided to post it instead of trying to refine it ad naseum. Today I’ll be talking about Single Photon Emission Computed Tomography (SPECT). If you missed the first part, I gave an overview of the major imaging modalities and why I’m doing this (see Medical Imaging 101 pt 1). Like PET imaging, SPECT requires an exogenous agent, i.e., there is no background signal. That can be a good thing. If your agent is targeted, you would have the best contrast, i.e., you only see the targeted tissue/organ. SPECT is most often used in cardiology. There are some SPECT scans for brain and bone imaging.

Physics for raw data acquisition

SPECT scanners look a lot like a PET and CT scanners. Recall CT scanners have a rotating gantry with an x-ray source and a detector (see Medical Imaging 101 pt 2: CT). Most SPECT scanners have a rotating gantry with 2-4 detectors (see the GE video below). The detectors are crystals, specifically scintillation crystals. Scintillation is the phenomenon where light is emitted  when the scintillator (liquid or crystal) absorbs radiation.  Two examples are sodium iodide doped with thallium (NaI(Tl)) and cadmium zinc telluride (CZT). Thallium doping allows the NaI crystals to scintillate at room temperature. Similarly, cesium iodide crystals can be doped with Tl. Gamma rays are lower energy than beta rays (PET imaging) and NaI(Tl) has higher efficiency at lower energy (95% for 140 keV) and only 10% for 511 keV (beta rays). As you can see, different crystals have different efficiencies and sensitivities to different energy windows. The next step is a photo-multiplier tube (PMT). The PMT is used to get the light from the crystal to an electronic detector. A more compact and modern alternative is an avalanche photodiode (APD).

So how do the detectors and radionuclide work together? The radionuclide for SPECT has to undergo gamma decay, i.e. high energy photons. Like x-rays, it’s a form of electromagnetic radiation. Unlike PET, SPECT detection does not rely on coincidence detection (recall beta decay has two positrons collide). Gamma rays are lower energy and high penetrating power. Positrons are even higher energy (~170 keV vs. 511 keV). SPECT uses collimators because it can’t take advantage of the 180 degree annihilation that PET has. Parallel collimators have holes parallel to the detector. They have less attenuation and therefore high sensitivity. Pinhole collimators are a set of hole(s) at the center of the collimator. They have more attenuation but they haven an important advantage, they provide magnification (see figure below).

Contrast

Because there is no background gamma rays from within your body, the contrast is completely from the injected/ingested radiotracer.  The majority of SPECT images use Tc-99m (technetium-99 m, the m is for metastable) labeled radiotracers. That’s one of the advantages. Say you have an antibody you wish to detect in vivo. You can chelate (chemically attach) Tc-99m, i.e. attaching Tc-99m is relatively easy so you can turn many agents into radiotracers.

How is the 3D image made: FBP

I discussed Filtered Back Projection (FBP) in the CT post. SPECT is inherently much lower resolution compared to CT. So there is a need for more tricks with reconstruction to improve the spatial resolution. For example, an iterative reconstruction method that uses expectation maximization (EM) is often less sensitive to noise compared to FBP.  Ordered subset expectation maximization (OSEM) is a popular method as it is faster than FBP and often has less artifacts.  The standard iterative method is maximum likelihood expectation maximization (MLEM). OSEM is a faster subset of MLEM, no pun intended. MLEM uses a model of the system to simulated projections of the current estimate of the image. The acquired projections are compared to the simulated projections and the ratio between the current estimate and the measured estimate are used to modify the current estimate to produce an updated estimate (which should be better) and becomes the n + 1 iteration. MLEM is slow but accurate and can take up to 200 iterations. OSEM is an accelerated version of MLEM in that, the projections are divided into subsets and MLEM is run on each subset.

Strengths

SPECT is much more sensitive than MRI or CT, i.e., it can detect much lower concentrations of the contrast agent. In general it is less sensitive than PET as it uses collimators and PET does not, i.e., the signal is attenuated by the object and the collimator. In theory it can detect in the nanomolar range. SPECT radiotracers often use a chelator to attach technetium so the chemistry is relatively straightforward to label antibodies or proteins of interest.

Weakness

The use of radiation carries with it a whole host of issues. Obviously there are health and regulatory issues.  Because SPECT uses collimators, you attenuate the signal and lower sensitivity. You can counter that with either longer scan times or higher dose of radioisotope. 

Something unique: multiple isotopes

Unlike PET, which is more or less 511 keV, SPECT covers a range a spectrum of energy. Some isotopes have multiple energy peaks. If you acquire the whole energy spectrum, which most scanners can do, you can differentiate one isotope from another by reconstructing the data with different energy windows, i.e., filter the data. So you could label a drug with one isotope and a radiotracer that is specific for bone, for example, and image them simultaneously to see if the drug goes to the bone.

Example:

The video below is of a patient that was found to have an abdominal bleed. Look at the lower left. You’ll see the bolus of Tc-99m go from the pelvic region towards the head. That is not normal (see the right side of the patient). The second example is very interesting. Dr. Peng and colleagues at the Mayo Clinic have developed an oncolytic virus based on a measles virus. The virus has a sodium-iodide symporter (channel) which uptakes radioactive iodine (I-125 or I-123), which you can use SPECT imaging to visualize. The virus selectively destroys myeloma plasma cells. Replication of the virus peaks at 8 days post injection (two different patients are shown below, one of the patients is in complete remission). The PET image in part C is interesting because it seems that mostly the smaller tumors take up the measles virus. I’ll write separately about this.

Here’s an informative video from GE Healthcare showing their clinical scanner (note the detectors can move to change their angle and not just rotate around the patient).

http://www.youtube.com/watch?v=QNgstUOnOZc

Medical Imaging 101 pt 1

http://goo.gl/LTWUf

Medical Imaging 101 pt 2: CT

http://goo.gl/IHaFw

Medical Imaging 101 pt 3: MRI

http://goo.gl/UVbiU

Medical Imaging 101 pt 4: PET

http://goo.gl/YNAVhX

References and image sources

How stuff works.

http://goo.gl/xGqmuq

Washington University in St. Louis, Radiology case studies

The site is down, I’ll add a link to the abdominal bleed if the site comes back up.

http://gamma.wustl.edu/home.html

Parallel vs. Pinhole collimation

http://goo.gl/pzbB9a

Analytic and Iterative Reconstruction Algorithms in SPECT

J. Nuc. Med. V43 2002

http://jnm.snmjournals.org/content/43/10/1343.long

Remission of Disseminated Cancer After Systemic Oncolytic Virotherapy

Mayo Clinic Proceedings, Volume 89, Issue 7, Pages 926–933, July 2014

http://goo.gl/Y61fX4

#ScienceSunday  

8 thoughts on “Medical Imaging 101 pt 7: SPECT”

  1. Before I knew any better, I once left off the “m” when writing Tc-99m. People understood what I meant in context, but I guess technically one letter makes a big difference since Tc-99 has a half-life of over 200,000 years and Tc-99m has a half-life of 6 hours. Details shmeetails… 🙂

    Chad Haney For the abdominal/pelvic bleed example, what would be the reason for choosing SPECT rather than something else like CTA? Is it because the former is better (more sensitive) at detecting slower rates or smaller amounts of extravasation for intermittent bleeds? or is it the difference in radiation exposure to the pelvis?

  2. Johnathan Chung, I don’t know the answer. The Wash U site is down but they didn’t have a lot of information there anyway. There are a few possibilities. Fluoroscopy is fast but 2D. Not knowing what’s wrong with the patient, I think fluoroscopy is not a good first choice. Most CT scanners (except the newer slip-ring design) are not fast enough to catch some perfusion events. I bet they did plain CT or MRI first and didn’t find anything. SPECT gave them a fast 3D scan to see what was happening dynamically.

    So yes, SPECT has higher sensitivity for smaller bleeds and can image either fast or slow bleeds.

    CT and MRI contrast agents tend to be small molecular weight. They could diffuse to fast and lose contrast. A SPECT agent, especially one linked  to albumin would work well.

  3. Perhaps if your post was the only one that happened, but it’s not. It’s just one of those days when you’ve read too much/too tired for information to permeate anything.

Leave a Reply