Positron Imaging Centre

The First Birmingham Position Camera

The first Birmingham Positron Camera was designed and constructed at the Rutherford Appleton Laboratory by Bateman et al and completed in 1984. It was one of two RAL-MkII PET systems built (of which the other, known as MUP-PET, was used for medical PET at the Royal Marsden Hospital, Sutton, UK). The design was developed from the earlier RAL-MkI MWPC, a small scale prototype.


The camera consists of two multi-wire chambers, each of active area 600x300 mm2, operating in coincidence. The diagram below shows part of a single set of planes from a detector; each detector contains a stack of such assemblies. In each anode plane there are 300 parallel wires, 20 mm in diameter and 2 mm apart. On either side of the anode plane are two cathode planes, consisting of lead plated onto printed circuit board for strength. The anode wires are held at a potential of 3.5 kV relative to the cathode planes, and the intervening volume is filled with gas (isobutane plus 0.1% freon) at atmospheric pressure. The lead of the cathode planes acts as the principal converter for 511 keV photons, and is only 50 mm thick in order that the electron released when a photon interacts by photoelectric absorption or Compton scattering should have a reasonable probability of escaping into the gas, where it creates ionisation. The resulting free electrons are accelerated towards the nearest anode wire. In the vicinity of this wire the electric field is so high that the electrons gain enough energy between collisions with gas molecules that they cause further ionisation, triggering an avalanche which is detected as a pulse of induced voltage on the anode wire and the two cathodes. The lead cathodes are divided into strips, 2.2 mm wide and separated by 0.8 mm, and the strips on the two cathodes run in orthogonal directions, so that the position of the ionisation within the plane can be determined from identifying which strips receive an induced voltage.


Only 1% of incident 511 keV photons interact in traversing 50 mm lead, and of these interactions only about 35% result in release of an electron into the gas. In order to achieve a useful efficiency, each detector contains a stack of 20 assemblies as shown, comprising a total of 21 cathode planes interspersed with 20 anode planes. The overall detection efficiency of this stack is around 7% at 511 keV.

The voltage pulse from the anode wires provides the fast trigger to the readout system. A coincidence is recognised if anode pulses are received from both detectors within a 12.5ns resolving time. The wires within each plane are electrically connected, but the signals from the separate planes are distinguished in order that the plane within which the detection occurred can be identified and used for parallax correction (the total depth of the stack of planes is 132 mm).

For simplicity the positional readout from the cathode planes is achieved using delay lines. The parallel lead strips of each cathode plane are fed onto a delay line and the time delay between the fast trigger pulse from the anode wires and the arrival of the cathode pulse at the end of the delay line determines how far along the delay line the event occurred. Only two delay lines are used in each detector, a single 600 mm (300 ns) delay line serving all the cathode planes with strips parallel to the short side of the detector, and one half that length serving those whose strips run in the orthogonal direction. In order that events where more than one pulse occupied a delay line simultaneously can be discarded, signals are taken from both ends of each delay line and a check is made that the sum of the times to reach the two ends is equal to the delay line length. At detection rates above about 2x105s-1 a significant number of detected events are discarded because of this pile up. Due to the detector efficiency, the coincidence rate is at best 7% of the singles rate in either detector, and in practice the coincidence rate which can be recorded saturates at around 5000 events/s. There would be relatively little advantage in improving the readout system to accommodate higher rates, since in practice the rate of random coincidences usually limits the useful coincidence rate to around 3000 events/s.

The spatial resolution is governed in approximately equal proportions by the distance the electron travels transversely along the gap parallel to the converter planes and by timing spreads in the delay line readout. In addition, scattering of photons in the detector prior to detection adds long tails to the spatial response. Under optimum conditions the back projected image of a central point source has a full width at half maximum of around 8 mm. In most actual imaging studies the limited statistics obtained mean that additional smoothing is needed, so that the practical resolution is poorer than this figure suggests.

Data from the camera are recorded event by event on computer for subsequent processing. For each event, the detection coordinates are recorded together with additional information such as the elapsed real and live times.

Although this camera is still fully operational, its role is gradually being taken over by the Forte.