PDF Print

Structure of Gamma Cameras

The heart of the camera is a scintillation detector that is complemented with a collimator ensuring position sensitivity. The figure below shows a schematic depiction of the parts of the camera. The gamma photon to be detected passes through a collimator first, then it generates scintillation (light) photons in the scintillaton crystal covered with a reflective coating. The reflective coating is necessary so that the scintillation light travels towards the detector and is not lost. The generated photons move through light conducting glass, then hit the PMT matrix. There is a permalloy magnetic shielding around the PMTs so that the magnetic field which might be present does not interfere with the operation of the photoelectron multipliers. The outermost coating is actually a lead shielding housing that protects the crystal from the scattered gamma photons.

Figure 1.

The mount holding and moving the head of the gamma camera is called gantry. Designing the gantry is not trivial, since it has to precisely move a head weighing 200-300 kg. Furthermore, the electronics creating, amplifying and sampling the signal of the PMTs are located in the head of the gamma camera as well. Data are collected by one or more computers, which also perform image corrections and evaluation.

Let us discuss the individual elements in detail now.

Scintillation crystals
In gamma cameras crystals with a large surface area are needed, in some cases as large as 40 cm x 50 cm; typically sodium iodide single crystals doped with thallium are used, as they have good parameters and they are cheap. Their light yield is 38 photon/keV and they have a moderate density (3.7 g/cm^3). They also have a moderate gamma attenuation factor, and they are generally used for technetium-99m (140 keV) or iodine-131 (365 keV) isotopes, so the typical thickness of the crystal is 6-12 mm.

Figure 2.

They have a high refractive index (1.85), which makes light coupling from the crystal more difficult. In gamma cameras timing is not necessary, so crystals with a short response time can be used (in NaI light yield decreases with a time constant of 230 ns).

Their disadvantages are that they break easily and they are very hygroscopic, i.e. they absorb humidity present in the air. Consequently, they have to be placed in a special housing. A further important feature is the emission spectrum of the crystal; it is practical to adjust it to the optical properties of the light conductor, the reflective coating and the photocathode of the detector.

Figure 3.

PMT detector matrix
The photoelectron multipliers are typically situated in a quadrilateral or a hexagonal grid on the crystal, there are 30-60 of them altogether. One of their characteristics is the so-called transit time which is the time required for the scintillation photon or the electrons generated from this photon to pass through the PMT (the photons generate electrons that are multiplied to the adequate number). This time can be different for the individual PMTs but the difference is not significant, unless timing matters. In such a case the transit time of the individual PMTs can be corrected by changing cable lengths. This is what they do in PET devices, where coincidence circuits make timing very important.

The PMT can be either DC or AC coupled, the difference lies in the fact that the housing of the PMT is connected with negative high voltage when DC coupled, however, in case of AC coupling no special contact protection is needed. The output signal is not the same either, since in case of AC coupling we get a bipolar signal, while in case of DC coupling we do not, as it can be seen in the figure below.

Figure 4.
Figure 5.

As previously discussed, the noise of the output signal of the PMT is determined by the statistics of the number of photons and not by the gain of the PMT.

Sampling of signals
The sampled signal is preamplified and shaped by the front-end electronics. This is necessary because the light scattered by the light conducting layer has a Gaussian distribution, thus constant noise appears on each PMT. This deteriorates the precision of positioning far from the point of arrival of the gamma photon.

There are analogue and digital gamma cameras, depending on the level of digitalization of the incoming signal. In case of analogue cameras a matrix of resistors produces corner signals (2 x and 2 y signals), which it then digitalizes; in contrast, in case of digital cameras each amplified and shaped PMT signal is digitalized by separate channels.

ADCs operate at 40-50 MHz and typically have a resolution of 12 bits. Out of this there are 48-64 independent ADC channels on the sampling electronics (used today), the signals of which arrive in series at the FPGA (Field Programmable Gate Array) that performs the processing. The frequency of the series signals is ~0.6 GHz, so a special high frequency PCB design is required. The paralleling of the ADC channels and the integration with a sliding time window take place in the FPGA, a register is continuously filled with the incoming signals during this process and its contents are read out at specified intervals. In addition, the rejection of pile-up events, the correction of the base line (if there has not been a scintillation signal for a long time), the formation and transmission of the ethernet packets, the communication with the computer collecting the data and the position calculation may all take place here.

The use of a collimator ensures that instead of just obtaining information about the presence of the source, we can even image the spatial distribution of the isotope as well. There are more types of collimators; parallel hole collimators (typically those that have a honeycomb structure) are the most widespread. This arrangement can be created by either casting or folding lead foil. Besides, converging fan beam and cone beam structures also exist; the difference between them is the direction(s) in which the collimator holes converge. Pinhole collimators, which consist of one or several apertures not close to each other, magnify the image, just like converging collimators do. However, we get a narrow field of view at the expense of magnification.

Figure 6.
Figure 7.

The imaging properties of the collimator, its sensitivity and resolution (the full width at half maximum of the point response function) depend highly on the distance of the collimator and the source, and of course on the type of the collimator. The sensitivity of collimators is very low, they typically transmit only ~0.05% of the photons.

Parallel hole collimators can be characterized by three parameters: hole size, bore length and septal thickness. The desired ratio of the resolution to sensitivity can be achieved by choosing the parameters properly. This adjustment is an optimizing problem because sensitivity decreases when increasing resolution. Therefore, different collimators need to be produced for the different applications in accordance with the required imaging properties. There are LEHR (low energy – high resolution), UHR (ultra-high resolution) and LEGP (low energy general purpose) collimators available on the market.

Figure 8.
Figure 9.

The point response function of the collimator-camera system approximately follows a Gaussian distribution, the standard deviation depends on the distance of the collimator to the source according to the following formula: \sigma (r)=\sqrt{(p_1+p_2 r)^2+\sigma_{intr}^2}, where \sigma_{intr} is the intrinsic resolution of the camera.

The aperture of a pinhole collimator during small animal studies is 1 mm, while in case of human applications it is 4-6 mm. The imaging is practically equivalent to camera obscura, and due to magnification its resolution can fall below the intrinsic resolution.
D_{eff}=\sqrt{D^2+\frac{2Dtan(\alpha /2)}{\mu} and \sigma_{pinhole}=\sqrt{(\frac{\sigma_{intr}}M)^2+(1+ \frac 1 {M})D_{eff}^2}
, where M is the magnification, D_{eff} is the aperture hole diameter increased by edge penetration and μ is the linear attenuation factor of the aperture at a given gamma energy. It is obvious from the formulae that edge penetration is dominant in case of a small hole size (~1 mm) and in order to decrease it a material with a high attenuation factor should be chosen as aperture. Examples include wolfram and gold.

Sensitivity (S) can be increased by using more apertures instead of only one (multi-pinhole imaging):
S=cos^3(\gamma) \frac{D_{eff,sens}^2}{16 r^2}

Site Language: English

Log in as…