eanm-logo eanm-logo
European Nuclear Medicine Guide
eanm-logo eanm-logo
European Nuclear Medicine Guide
Chapter 10

Nuclear Medicine Instrumentation

Basic instrumentation, SPECT/CT and PET/CT scanners

Positron emission tomography (PET) and single photon emission computed tomography (SPECT), the two most common nuclear medicine medical imaging technologies, are based on the detection of gamma radiation. In the case of PET, radiotracers are β+ emitters, which upon an annihilation with an electron produce pairs of gamma rays. In the case of SPECT radiotracers, gamma photons are directly emitted upon a nuclear de-excitation of one of the short-lived excited nuclei.

SPECT and PET detectors are both based on the same principle technology: a scintillator coupled through a light guide to a photo-detector. The principal aim of the scintillator is to stop gamma rays and convert the absorbed energy into visible scintillation light, which are subsequently collected by the photo-detector and converted into an electrical pulse.

There are several types of photo-detectors, and photomultiplier tubes (PMT) and silicon photomultipliers (SiPM) are the most popular. PMTs consist of a vacuum tube with a photocathode at the entrance, a series of dynodes at successively high voltages, and an anode at the end of the PMT. The photocathode detects the optical photons emitted by the scintillator, and, by photoelectric effect, electrons are emitted in the direction of the dynodes, through the focusing electrodes, and accelerated by the existing electric field. When electrons interact with the dynodes, these are refocused to the next dynode, and new electrons are generated from the dynode. The amplification of electrons is exponential and results in an exceptional gain, when electrons reach the anode. PMTs have excellent timing and gain properties. However, these devices are bulky, delicate, operate at high voltages, and the electrons-focusing stage along the dynodes does not function in the presence of electromagnetic fields.

SiPMs are considered the semiconductor analogue of PMTs. A photon travelling through silicon may interact with a bound electron and transfer its energy. This energy transfer causes the electron to move to the valence shell and generate an electron-hole pair. SiPMs are based on a densely packed array of avalanche photodiodes (p-n junction) operating in Geiger mode, i.e., in reverse-bias voltage over the breakdown voltage. All photodiodes are connected through a polysilicon quenching resistor and are read in parallel. When a photon from the scintillator interacts with a photodiode, it causes a breakdown discharge that results in a photoelectron pulse of high gain. The output of the SiPM is the sum of all the pulses from every photodiode that detected a photon. SiPMs have gain comparable to PMTs operating at low voltages, are very compact and fast, and can function in the presence of electromagnetic fields[68].

From the physical principle’s perspective, the basic differences between PET and SPECT cameras, besides the different radiotracers to which they are sensitive, originate in the dual gamma-emission per nuclear decay for PET, while radionuclides used in SPECT emit only one gamma-photon per nuclear decay. This difference determines the different geometries used in PET and SPECT systems. PET systems consist of a ring of detectors, while SPECT systems may consist of one to three rotating detectors, but normally there are only two detectors fixed at 90° or 180° on a rotating gantry.

The detection principle in PET relies on electronic coincidence detection (also known as electronic collimation) with high detection efficiency, as compared to SPECT, which requires a high density (usually made of lead or tungsten) physical collimator to determine the direction of the incoming photons. Collimators are designed to absorb gamma-photons coming from non-perpendicular directions, which implicitly limits the detection efficiency. Collimators can have several structures, like parallel-hole, multi-pinhole, multi-slit or fan-beam structure, among others. Parallel-hole collimators consist of an array of parallel holes which can be large or small depending on whether high sensitivity or spatial resolution is aimed. Pinhole collimators consist of an array of cone-shaped holes and have a better trade-off between sensitivity and resolution than parallel-hole collimators, but these usually cover a small field of view. While pinhole collimators magnify the object that is being imaged, their sensitivity is dependent on the distance to the object.

Another difference between PET and SPECT cameras resides in the different scintillators used for each modality. The scintillators are especially designed to provide the best properties in a range of energies depending on radionuclide. Gamma-photons produced from PET radiotracers have an energy of 511 keV, while the energy from gamma-photons emitted from SPECT radiotracers varies depending on the radionuclide and range from 110 keV to >350 keV. The energy of the gamma photons, and how they interact with the SPECT collimators, has a significant impact on the spatial resolution. Some common radionuclide for PET and SPECT are shown in Tables 1 and 2, respectively.

Table 1. PET radionuclides

Radionuclide

Positron energy (Mean/Max)

Positron range (Mean/Max)

Half life

18F

0.28/0.65 MeV

0.7/2.5 mm

110.0 min

68Ga

0.84/1.90 MeV

2.9/9.0 mm

67.7 min

11C

0.39/0.96 MeV

1.2/4.2 mm

20.4 min

15O

0.73/1.73 MeV

3.0/8.4 mm

2.0 min

13N

0.49/1.20 MeV

1.8/5.5 mm

10.0 min

 

Table 2. SPECT radionuclides

Radionuclide

Energy

Half life

99mTc

141 keV (89%)

6.02 hours

177Lu

208 keV (10.4%) and 113 keV (6.8%)

6.70 days

123I

159 keV (83%)

13.22 hours

111In

171 keV (91%) and 245 keV (94%)

2.80 days

155Tb

87 keV (32%) and 105 keV (25%)

5.32 days

Scintillators in SPECT are generally continuous slabs decoded by an underlying array of photo-detectors. Upon an interaction, light propagates in the scintillator, and the intrinsic spatial resolution degrades with the scintillator thickness. To overcome the dependence between spatial resolution and scintillator thickness, there are already some systems with pixelated scintillators[69]. In the case of PET scintillators, they generally consist of an array of crystals separated by reflector material to confine the light in each crystal when an interaction takes place. The crystal size is directly related to the reconstructed image spatial resolution, but small crystals may require a significant number of photo-detectors to accurately decode the gamma interaction position, and this increases the total cost of the PET system.

Scintillators have several physical properties that affect the efficiency, time, and energy resolutions of PET/SPECT detectors and cameras. The density and atomic number determine the efficiency of the detector, which impacts the detector sensitivity. The light yield determines the number of optical photons generated from each gamma-interaction within some range which affects the time and energy resolution. The decay time determines how fast the luminescence signal decays after an excitation. Finally, the emission wave-length is the wave-length of the produced optical photons, which needs to be compatible with the wave-length to which the photo-detector is sensitive to. A fraction of the produced optical photons is converted into electrons in the photo-detector depending on the quantum efficiency of the photo-detector. The scintillator properties are directly related to intrinsic detector properties, namely the sensitivity, energy resolution, and time resolution. Other aspects to consider from scintillators are the hygroscopicity, how difficult is to cut them and how finely they can be cut, and radiation hardness among others. Some typical scintillators used in PET and SPECT are shown in table 3.

As mentioned before, PET systems consist of a ring of detectors to detect the pairs of gamma photons from a positron-electron annihilation. In the case of SPECT cameras, usually one to three heads located in a rotating gantry are used to detect the gamma photons. PET and SPECT systems are usually attached to a computer tomography (CT) system to acquire anatomical information without moving the patient and accurately overlay the functional information from the PET or SPECT on a perfectly aligned anatomical reference[70].

Table 3. PET and SPECT scintillators

Scintillator

Technology

Density
(g/cm3)

Light yield
(Ph/MeV)

Emission
wavelength
(nm)

Decay time
(ns)

LSO(Ce)

PET

7.40

27,000

410

40

BGO

PET

7.13

9,000

480

300

La3Br(Ce)

PET

5.29

61,000

358

35

NaI(Tl)

SPECT/PET

3.57

38,000

415

230

CsI(Tl)

SPECT

4.51

54,000

550

1,000

Special SPECT configurations

Typical SPECT systems consist of two gamma cameras mounted on a rotating gantry. However, there are several SPECT systems with singular configurations especially designed for specific organs. Most attention has been directed towards designing dedicated systems for cardiac SPECT to study myocardial perfusion.

There are several dedicated cameras based on cadmium-zinc-telluride (CZT) technology[71]. CZT detectors directly convert gamma radiation to electrical signals, thereby circumventing the need of a scintillator and improving the energy and spatial localization accuracy of gamma interactions. There are conventional whole-body SPECT systems based on CZT such as the Discovery NM/CT 670 (GE Healthcare, Waukesha, WI, USA), but there are also several dedicated cardiac SPECT systems based on CZT technology with a variety of collimation systems: stationary multi-pinhole, enhancing spatial resolution (Discovery 5(3)70c, GE Healthcare, Waukesha, WI, USA)[72], and rotational parallel-hole (D-SPECT, Spectrum Dynamics, Haifa, Israel) which enhance sensitivity[73].

In the GE Discovery system, each hole is focused on one pixel from the CZT detector (2.46x2.46 mm), thereby eliminating the light sharing problem. The system has 19 cameras focused on the heart, fully covers the heart and its surrounding tissue, and produces a high resolution image in a volume of 19 cm diameter, which needs to be centred in the heart (guided by an automatic algorithm) to achieve optimum performance.

The Spectrum Dynamics D-SPECT system consists of 9 rotating cameras placed in a curved configuration. The collimator is thinner (21.7 mm) and has larger holes (2.26 mm) than conventional collimators used in whole-body SPECT, and this results in a significantly larger solid angle and, hence, sensitivity. As in the case of the stationary multi-pinhole system, each hole is focused in one single CZT pixel, which is 2.50x2.50 mm.

There are also approaches to develop dedicated brain SPECT systems focused on Parkinson’s and similar neurodegenerative diseases, which, rather than a two-head structure, have a ring geometry similar to that from PET systems, like the Inspira HD system (Neurologica, Boston, USA). The main aim of these dedicated brain systems is to achieve a spatial resolution in the range of ~3 mm compared to 7-12 mm obtained with whole-body systems[74]. The INSERT research project also requires special mention, whose aim is to implement a high resolution MR-compatible brain SPECT insert based on mini-slit-slat collimators coupled to a monolithic CsI scintillator on a SiPM array[75].

Digital PET vs conventional PET

Conventional PET is based on detectors consisting of a scintillator coupled to a photo-detector based on analogue electronics such as PMTs and SiPMs. Optical photons produced in a scintillator following an interaction by a gamma-photon are emitted isotropically through the scintillator. A fraction of these optical photons eventually reaches the entrance of the photo-detector through an optical guide (grease, glue, or similar). Each photon interacting in a micro-cell generates an avalanche of carriers resulting in a pulse of current containing 105-106 electrons and lasting a few nanoseconds. For accurate determination of energy and time information of the detected photons, the output signal is usually amplified through a low noise pre-amplifier, and the pulse shape is enhanced and cleaned before the analogue signal is digitized.

A current trend in the development PET detectors is to perform the digitization process at an early point in the electronic chain, like CZT detectors that perform the digitization upon gamma photon detection. However, CZT detectors are currently used only in SPECT[76] and small animal PET systems[77].

Another trend is performing the digitization directly in the photo-detector upon detection of optical photons. The digital photon counter (DPC), developed by Philips (Philips Healthcare, Best, Netherlands), was first presented in 2009[78]. With digital SiPMs, the pulse generated in each SiPM diode is immediately digitized upon generation eliminating any signal variations due to electronic noise and temperature fluctuations. Digital SiPMs have several benefits over analogue SiPMs, which are briefly discussed below.

In the case of analogue SiPMs, the time stamp of an interaction is obtained from a discriminating leading edge as the sum of all the photodiodes. In the case of digital SiPMs, the signal from each photodiode is digitized, effectively functioning as a photon counter, and potentially providing the time stamp of each detected photon from the scintillator, which has been shown to improve the system coincidence time resolution (CTR)[79]. The only commercial digital SiPM to date is the DPC manufactured by Philips, but there are other research projects developing digital SiPMs[80].

Several efforts have investigated the time performance of digital SiPMs coupled to scintillators of different sizes reaching a CTR of ~130 ps[81]. The commercial Philips PET/CT scanner Vereos, based on DPC detectors, has a CTR of ~320 ps[82]. as compared to other PET scanners based on PMTs with a CTR of ~500 ps and SiPMs with a CTR of ~250-350 ps.

In order to identify the position following a gamma interaction with pixelated scintillators based on light sharing, centre of gravity is commonly used. In addition to the great performance in timing resolution, using the digitized SiPM outputs, advanced algorithms can be used to improve the interaction position accuracy[83].

There are other designs with the digitization process close to the SiPM, such as the LightBurst PET detector (GE Healthcare, Waukesha, WI, USA) that integrates the ASIC and the analogue SiPM in the same chip. Together with a one-to-one crystal coupling, this results in outstanding sensitivity and volumetric spatial resolution.

Benefits from current advancements

There have been significant advances in the development of novel PET detectors in the last few years. Depth of interaction by using multi-layer or monolithic scintillators to increase the spatial resolution uniformity across the field of view has been the subject of several studies including whole-body scanners[84], but especially in dedicated brain[85] and small animal PET scanners[86] due to the close proximity of the PET detectors and the field of view.

The change from PMTs to SiPM detectors has opened the possibility of increasing the amount of light collected in the photo-detector and the solid angle coverage to produce a significant boost in detected radiation and hence sensitivity and count rate. This increase in sensitivity enables a number of improvements in PET imaging, namely reduced injected radiotracer (facilitating the scan of children and pregnant women), short scan times, and improved image quality.

Overall, most technical advances in clinical systems have been focused on developing fast detectors to obtain high CTR, mainly driven by time-of-flight PET applications. Increased CTR has been shown to produce significantly high image quality and increasing the image contrast, improving small lesions detectability, and increasing the spatial resolution.

CTR relies on the scintillator, photo-detector and electronic chain, and these three parts have been the focus of different research studies. All PET manufacturers have put special effort on improving the CTR of their scanners and setting the minimum at 250 ps for the moment in a clinical PET/CT scanner.

There is an active community working on new scintillating materials, new doping ions, and new photo-detectors, and some research groups are working on improving the timing and efficiency properties of scintillators and photo-detectors with the aim of 10 ps on the horizon[87].