2010-02-13
Half-life Of Technetium-99m
My Dad had a Sestamibi myocardial perfusion imaging test on Tuesday as part of a cardiac assessment for surgery he requires in a few days. He is an almost 20 year survivor of a fairly bad MI which has left his heart unfortunately quite compromised with a poor EF, naturally his doctors were concerned about his ability to tolerate the required general anaesthesia. The appointment for the nuclear medicine test was made at short notice, so I didn't have a lot of time to prepare for the event - had I more time I would have built dosimetry equipment to document the event more closely - none the less I couldn't help but make use of the situation to do a little geeky experimentation.
Sestamibi is a Technetium-99m labelled tracer, composed of Methoxyisobutylisonitrile coordinated with Technetium. The MIBI caging the Tc is lipophilic and gives the radiopharmaceutical high affinity for heart muscle (amongst other tissues). The Tc-99m nuclide itself is a pure gamma emitter, with a half-life of about 6 hours and emits 140 keV gammas. It decays to Tc-99 which has a much longer half life (211 ka). Tc-99m seems almost ideal of nuclear medicine procedures, it has a short half-life so it has high activity and relatively small amounts can be given so total dose is kept reasonable (i.e bright, but brief). The gamma energy is around the same as diagnostic x-ray photons giving them similar properties (gamma and x-rays are basically identical, it is really only a naming convention based on what generated the particular photon, not their wavelength). As a pure gamma emitter it is not as damaging to tissue as isotopes which emit charged particles, and its decay product, despite being a beta emitter, has such a long half-life that it offers little danger and would likely be eliminated before much decays at all. Most importantly Tc-99m can be produced and purified easily, on site from longer-lived isotopes which helps make it practical despite its short half-life. The Molybdenum-99 based generators (aka the Moly Cow) have made the technology practical, and likely save thousands of lives every year.
Anyway, on the day of the test, a very nice technician injected a few hundred microlitres of the tracer (containing what he said was about 370 MBq of Tc-99m) into a vein in my Dad's arm. The test is completely painless, but involves several hours of sitting around waiting for the tracer to redistribute, then imaging with a Gamma Camera before and after cardiac stressing by exercise or drug (Persantine). The injection was at about 8:20 am. By the time we got home it was past midday. Naturally I grabbed my Geiger counters to see how much activity persisted. He was extremely hot, detectable by ear using my toy unit 5 metres or more away, and pegged my calibrated Technical Associates unit at over 100,000 CPM over the chest and upper abdomen.
Measuring Half-life
My dad was a real trooper, and was easily persuaded to supply a urine sample for further experimentation. Working with a liquid sample was far more practical than fixing a detector to him for the days required to observe the complete decay of the tracer (biological clearance would have made that a very poor experiment anyway). I hacked together a counting rig using my toy "clicking detector" in the lab and left the urine sample to decay right down into the background.
Just looking at the curve shows the half-life is about 6 hours, but the right way to extract the decay constant is to take the logarithm of the activity so as to reduce the exponential decay into a linear relation from which the slope directly gives the constant. GNUplot has the ability to directly fit a regression line to the data, greatly simplify the process. Actually GNUplot can directly fit the exponential data (or any arbitrary function) if given some hints, but linearising the data gives a better idea of the quality of the data.
The dataset is quite well behaved (and beautifully displays Poisson distribution), but reduction gives a value for half-life that is close, but high by a significant amount (about 5%, my raw fit gives 6.3 - 6.4 hours depending on how much of the data I use for the fit). The cause is the uncorrected dead-time of the counting system (and to a lesser extent the background level). The dead-time has more impact at higher count rates, making those counts read low and hence make the decay rate appear smaller than it indeed is. This is a systematic error that can be modelled and corrected for.
Dead-time Correction
Using the CRO and the old Americium-241 source I observed a minimum delay between counts of approximately 420 us. Long term integration of the background radiation gives a figure of 24.5 CPM. Putting both of these figures into a corrective model gives a figure 6.041 hours, around 0.5% from the established half-life of Tc-99m.
Further error is likely poorly controlled counting geometry, inadequate GM power supply regulation and additional unmodelled counting systematic error (perhaps from the MCU firmware or pin-change detection synchronisation hardware). I have not yet performed extensive statistical analysis to determine what kind of confidence my data provides, but my figure is fairly close considering how primitive my lash-up was and how poorly prepared I was for the entire experiment.
Other Experiments
Upon first coming home from the test and even early the next day my Dad was sufficiently radioactive to trigger the Nuclear War Detector toy ion chamber when held over his chest. Even with the end-window closed the thin tin plate walls offered little significant effect on the sensitivity. You can hear the Geiger counter (sitting a few metres away) going nuts in the background of this video as well.
Upon waking up on the day after the test I realised I might be able to use the tracer radiation still in his body to do some coarse imaging. I built a makeshift gamma camera by rolling a collimator tube out of 2 mm lead sheet for my RS-232 interfaced Geiger counter (now in a box) and 50 mm pancake tube, mounting the resulting device on a tripod.
I taped out a tomographic reference guide on the living room floor, intending to make about 80 projection measurements to allow reconstruction of a coarse tomograph of his chest. While the detector worked quite well I had missed my chance, he was no longer a bright enough source to achieve a reasonable SNR in practical integration times.
Using the collimated detector I noticed the radiation seemed to be fairly localised to the centre of the chest with a noticeable dip over the sternum, maybe the result of a bit of adsorption in the bone? The liver was very hot as well, and over time the radiation seemed to move more into the lower abdomen, suggesting perhaps biliary excretion from the liver into the intestines? While his extremities and head were much hotter than the room background, they were very much less luminous than the torso. I wish I had built the collimated detector sooner, not only so I could have attempted imaging while he was still sufficiently bright, but also so as to do more controlled measurements of relative activity over each major organ. If I ever get the opportunity again I'd love to have a mechanical scanning system and collimated detectors available.
6 comments.
Attachments
| title | type | size |
|---|---|---|
| The Urine Dataset | text/plain | 104.412 kbytes |