Naked Engineering - Designing Combined PET/MRI Scanners

23 January 2011

Interview with

Richard Ansorge, University of Cambridge

Meera -  For this week's Naked Engineering, Dave and I have come along to the Addenbrookes Biomedical campus to investigate antimatter, but more specifically, just how antimatter can be used for medical imaging.  So here to tell us a bit more about this is Dr. Richard Ansorge from the Cavendish Laboratory at the University of Cambridge.  Now Richard, if you could just set the scene then as to what are the different types of medical imaging?

Richard -   The oldest type of medical imaging is just straightforward x-rays.  Although these days, that is usually done as a CT scan which actually gives you a 3-dimensional X-ray image by combining many slices.  In the 1980s, I guess it was, Magnetic Resonance Imaging [MRI] started to be developed based on the magnetic properties of the hydrogen atoms in all the water molecules, and other organic molecules inside every living person.

Dave -   So because you're looking at the hydrogen atoms rather than x-ray, which is just looking at the really heavy atoms in your bones, you can see a lot more detail of the soft, squishy bits.

Richard -   Yes, that's exactly right.  One of the major applications of MRI imaging is, PET Image of the human brain showing energy consumptionfor example, to look at people's brains.

Meera - But as well as CT and MRI scans another technique is PET scanning which is your particular area...

Richard -   Yes.  PET stands for Positron Emission Tomogrpahy, the positron of course is the antimatter equivalent to the electron, and is like an electron except for the fact it has a positive charge.  It's possible to manufacture radioactive isotopes that will emit positrons, for example Fluorine-18, which has a 2-hour half life, is used a great deal in PET for example in a molecule called FDG which is a glucose analogue.  When injected into a patient, it will be metabolised and the radioactive Fluorine tends to accumulate in positions in the body where the tracer has been metabolised.  The end result of that is that one literally has hot spots of radioactivity, in places where there are hot spots of metabolic activity.

Dave -   So for example, your brain or especially I guess a tumour will be very, very active and respiring a lot so taking in a lot of glucose, so that will be very radioactive.

Richard -   Yes and so, for the last 20 years or so, PET has been used mainly for oncology.

Meera -   Metabolic function is the end result, but how does a PET scanner actually work to find where those metabolic areas are?

Richard -   What happens is that the radioactive substance decays to produce a positron which wanders around in your tissue, travelling typically a millimetre or less and then meets a normal electron in another atom, and annihilates with it.  The electron and the positron disappear and 2 gamma rays, each having an energy equivalent of the rest mass of an electron and positron are produced, travelling in fairly precisely opposite directions.

Dave -   So this is the E=mc2 thing where all of the mass of the electrons has been converted into energy in those photons.

Meera -   And these are what are essentially picked up by detectors in the scanner.

Richard -   Yes.  An important feature is that 2 gamma rays are produced and are detected simultaneously.  That means that since you have injected a radioactive substance, there are a lot of background counts, but the coincidence of the arrival of two gammas tells you that you've actually got a good event associated with the decay, somewhere on a line, joining the two detectors that went off.

Dave -   And then I guess you just look at many, many events and you know all of these appeared on lots of different lines, and somehow you've got to work out where the original radioactivity was?

Richard -   You can run quite complicated mathematical algorithms on these lines of response to reconstruct the most probable distribution of the tracer.  It's a three-dimensional image for each small volume region in the body, showing the number of decays that were detected from that region.  Typically, these images are displayed in false colour.  So you get different shades of colour representing the intensity of the emission.

Meera -   But what about positioning it say, within our actual bodies?

Richard -   Tissue which is not absorbing the tracer is invisible in a PET scan.  So, it can be problematic, locating where a suspected cancer is actually located.  Is it in a lung or is it somewhere nearby?

Meera -   And so I guess to get around that problem, a recent development has been the combination of PET scanners with CT scanners to then get the structure and the metabolism.

Richard -   Yes, that's right.

Dave -   This has all the same disadvantage of a normal CAT scan. It's really finding the bones, but not so good at the soft tissues.

PET/MRIMeera -   This is where your research really comes in, Richard.  So we're currently in an imaging suite, standing next to a large PET scanner, but this is a combined PET/MRI scanner, which is what you're developing.

Richard -   MRI gives different types of contrast as compared to CT.  So for example in the brain, you can get a much more accurate view of where the PET activity is taking place.  There are a number of technical challenges involved in combining PET and MRI.  The photomultipliers used in PET scanners don't work in magnetic fields which are essential for MRI scanners.

Meera -   How, with your design, have you tried to overcome these problems?  Looking at the scanner now, it's a large, white cylinder, about 3 metres long and there are "wings" attached to the centre of it.

Richard -   If you look carefully at our system, you'll see that whereas a conventional MRI scanner is a single cylinder, we have sliced our scanner in the middle and opened up a gap.  The gap means that we can put a conventional PET detector in the centre of the magnet and the light that's produced can be brought out of the magnetic field using fibre-optic guides, travelling transverse to the axis of the magnet, and end in a region where the magnetic field is sufficiently low, that the photomultipliers will operate satisfactorily.

Dave -   So the wings which Meera mentioned earlier are basically the ends of these light guides and all the electronics.

Richard -  Yes, that's right.  So the end result of that is that our system has a sensitivity of about 5% which actually compares fairly favourably with unmodified systems which would be 6 or 7%.  A further advantage is that, unlike combined PET/CT systems, where the CT is run once as a quick snapshot at the beginning, the MRI can be run continuously, for the 10 minutes or so that the PET scan goes on, and one can track, for example, breathing, and allow for that to make much higher quality images than is possible with a PET/CT system.

Add a comment