Scintillation Detectors in PET

Electronics Repair Manuals

Schematic Diagrams and Service Manuals

Get Instant Access

As mentioned above, scintillation detectors are the most common and successful mode for detection of 511 keV photons in PET imaging due to their good stopping efficiency and energy resolution. These detectors consist of an appropriate choice of crystal (scintil-lator) coupled to a photo-detector for detection of the visible light. This process is outlined in further detail in the next two sections.

Scintillation Process and Crystals Used in PET

The electronic energy states of an isolated atom consist of discrete levels as given by the Schrodinger equation. In a crystal lattice, the outer levels are perturbed by mutual interactions between the atoms or ions, and so the levels become broadened into a series of allowed bands. The bands within this series are separated from each other by the forbidden bands. Electrons are not allowed to fill any of these forbidden bands. The last filled band is labelled the valence band, while the first unfilled band is called the conduction band. The energy gap, Eg, between these two bands is a few electron volts in magnitude (Fig. 2.19).

CONDUCTION BAND (empty)

ENERGY GAP (Eg)

CONDUCTION BAND (empty)

ENERGY GAP (Eg)

Scintillation Pet Band Gap

VALENCE BAND (full)

Figure 2.19. Schematic diagram of the energy levels in a scintillation crystal and the mechanism of light production after energy is absorbed. The photon energy is sufficient to move a valence band electron to the conduction band. In returning to the ground state, light photons are emitted.

VALENCE BAND (full)

Figure 2.19. Schematic diagram of the energy levels in a scintillation crystal and the mechanism of light production after energy is absorbed. The photon energy is sufficient to move a valence band electron to the conduction band. In returning to the ground state, light photons are emitted.

Electrons in the valence band can absorb energy by the interaction of the photoelectron or the Compton scatter electron with an atom, and get excited into the conduction band. Since this is not the ground state, the electron de-excites by releasing scintillation photons and returns to its ground state. Normally, the value of Eg is such that the scintillation is in the ultraviolet range. By adding impurities to a pure crystal, such as adding thallium to pure NaI (at a concentration of ~1%), the band structure can be modified to produce energy levels in the prior forbidden region. Adding an impurity or an activator raises the ground state of the electrons present at the impurity sites to slightly above the valence band, and also produces excited states that are slightly lower than the conduction band. Keeping the amount of activator low also minimizes the self-absorption of the scintillation photons. The scintillation process now results in the emission of visible light that can be detected by an appropriate photo-detector at room temperature. Such a scintillation process is often referred to as luminescence. The scintillation photons produced by luminescence are emitted isotropically from the point of interaction. For thallium-activated sodium iodide (NaI(Tl)), the wavelength of the maximum scintillation emission is 415 nm, and the photon emission rate has an exponential distribution with a decay time of 230 ns. Sometimes the excited electron may undergo a radiation-less transition to the ground state. No scintillation photons are emitted here and the process is called quenching.

There are four main properties of a scintillator which are crucial for its application in a PET detector. They are: the stopping power for 511 keV photons, signal decay time, light output, and the intrinsic energy resolution. The stopping power of a scintillator is characterized by the mean distance (attenuation length = 1/^) travelled by the photon before it deposits its energy within the crystal. For a PET scanner with high sensitivity, it is desirable to maximize the number of photons which interact and deposit energy in the de

Table 2.5. Physical properties of commonly used scintillators in PET. The energy resolution and attenuation coefficients (linear (p) and mass (p/p)) are measured at 511 keV

Property

NaI(Tl)

BGO

LSO

YSO

GSO

BaF2

Density (g/cm3)

3.67

7.13

7.4

4.53

6.71

4.89

Effective Z

50.6

74.2

65.5

34.2

58.6

52.2

Attenuation length

2.88

1.05

1.16

2.58

1.43

2.2

Decay constant (ns)

230

300

40

70

60

0.6

Light output (photons/keV)

38

6

29

46

10

2

Relative light output

100%

15%

75%

118%

25%

5%

Wavelength ^(nm)

410

480

420

420

440

220

Intrinsic AE/E (%)

5.8

3.1

9.1

7.5

4.6

4.3

AE/E (%)

6.6

10.2

10

12.5

8.5

11.4

Index of refraction

1.85

2.15

1.82

1.8

1.91

1.56

Hygroscopic?

Yes

No

No

No

No

No

Rugged?

No

Yes

Yes

Yes

No

Yes

V (cm-1)

0.3411

0.9496

0.8658

0.3875

0.6978

0.4545

V/p(cm2/gm)

0.0948

0.1332

0.117

0853

0.104

0.0929

tector. Thus, a scintillator with a short attenuation length will provide maximum efficiency in stopping the 511 keV photons. The attenuation length of a scintillator depends upon its density (p) and the effective atomic number (Zeff). The decay constant affects the timing characteristics of the scanner. A short decay time is desirable to process each pulse individually at high counting rates, as well as to reduce the number of random coincidence events occurring within the scanner geometry (see Ch. 6). A high light-output scintillator affects a PET detector design in two ways: it helps achieve good spatial resolution with a high encoding ratio (ratio of number of resolution elements, or crystals, to number of photo-detectors) and attain good energy resolution. Good energy resolution is needed to efficiently reject events which may Compton scatter in the patient before entering the detector. The energy resolution (AE/E) achieved by a PET detector is dependent not only upon the scintillator light output but also the intrinsic energy resolution of the scintilla-tor. The intrinsic energy resolution of a scintillator arises due to inhomegeneities in the crystal growth process as well as non-uniform light output for interactions within it. Table 2.5 shows the properties of scintil-lators that have application in PET. They are:

(i) sodium iodide doped with thallium (NaI(Tl)),

(ii) bismuth germanate Bi4Ge3O12 (BGO),

(iii) lutetium oxyorthosilicate doped with cerium Lu2SiO5:Ce (LSO),

(iv) yttrium oxyorthosilicate doped with cerium Y2SiO5:Ce (YSO),

(v) gadolinium oxyorthosilicate doped with cerium Gd2SiO5:Ce (GSO),and

(vi) barium fluoride (BaF2).

The energy resolution values given in this table are for single crystals. In a full PET system, variations between crystals and other factors such as light readout due to block geometry contribute to a significant worsening of the energy resolution. Typically, NaI(Tl) detectors in a PET scanner achieve a 10% energy resolution for 511 keV photons, while the BGO scanners have system energy resolution of more than 20%.

NaI(Tl) provides very high light output leading to good energy and spatial resolution with a high encoding ratio. The slow decay time leads to increased detector dead time and high random coincidences (see Energy Resolution and Scatter, below). It suffers from lower stopping power than BGO, GSO or LSO due to its lower density. BGO, on the other hand, has slightly worse timing properties than NaI(Tl) in addition to lower light output. However, the excellent stopping power of BGO gives it high sensitivity for photon detection in PET scanners. Currently, commercially produced whole-body scanners have developed along the lines of advantages and disadvantages of these two individual scintillators. The majority of scanners employ BGO and, when operating in 2D mode, use tungsten septa to limit the amount of scatter by physically restricting the axial field-of-view imaged by a detector area. This results in a reduction of the scanner sensitivity due to absorption of some photons in the septa. The low light output of BGO also requires the use of small photo-multiplier tubes to achieve good spatial resolution, thereby increasing system complexity and cost. The NaI(Tl)-based scanners [8] compromise on high count-rate performance by imaging in 3D mode in order to achieve acceptable scanner sensitivity.

LSO, a relatively new crystal, appears to have an ideal combination of the advantages of the high light output of NaI(Tl) and the high stopping power of BGO in one crystal [9]. In spite of its high light output (~75% of NaI(Tl)), the overall energy resolution of LSO is not as good as NaI(Tl). This is due to intrinsic properties of the crystal. Another disadvantage for general applications of this scintillator is that one of the naturally occurring isotopes present (176Lu, 2.6% abundance), is itself radioactive. It has a half-life of 3.8 x 1010 years and decays by P- emission and the subsequent release of y photons with energies from 88-400 keV. The intrinsic radioactivity concentration of LSO is approximately ~280 Bq/cc; approximately 12 counts per sec per gram would be emitted that would be detected within a 126-154 keV energy window. Thus its use in low-energy applications is restricted. This background has less impact in PET measurements due to the higher energy windows set for the annihilation radiation and the use of coincidence counting.

GSO is another scintillator with useful physical properties for PET detectors. One advantage of GSO over LSO, in spite of a lower stopping power and light output, is its better energy resolution and more uniform light output. Commercial systems are now being developed with GSO detectors.

Finally, the extremely short decay time of BaF2 (600 psec) makes it ideal for use in time-of-flight scanners (see Time-of-flight Measurement, below), which helps to partially compensate for the low sensitivity arising due to the reduced stopping power of this scintillator.

In addition to these scintillators, which have all been used in PET tomographs already, new inorganic scintillators continue to be developed. Many of the newer scintillators are based on cerium doping of lanthanide and transition metal elements. Examples include LuAP:Ce,Y2SiO5 (YSO), LuBO3:Ce, and others based on lead (Pb), tungsten (W) and gadolinium (Gd).

Photo-detectors and Detector Designs Used in PET

Generally, the photo-detectors used in scintillation detectors for PET can be divided into two categories, the photo-multiplier tubes (PMTs) and the semiconductor-based photodiodes. Photo-multiplier tubes (Fig. 2.20) represent the oldest and most reliable technique to measure and detect low levels of scintillation light. They consist of a vacuum enclosure with a thin photocathode layer at the entrance window. An incoming scintillation photon deposits its energy at the photocathode and triggers the release of a photo-electron. Depending upon its energy, the photo-electron can escape the surface potential of the photo-cathode and in the presence of an applied electric field accelerate to a nearby dynode which is at a positive potential with respect to the photo-cathode. Upon impact with the dynode, the electron, with its increased energy, will result in the emission of multiple secondary electrons. The process of acceleration and emission is then repeated through several dynode structures lying at in

To pre-amplifier

To pre-amplifier

High Energy Dynode

scintillator

Figure 2.20. Schematic diagram of a photomultiplier tube and a photograph of a hexagonal 6 cm-diameter tube (inset). Light entering the PMT displaces a photoelectron which is electrostatically focused to the first-stage dynode. Each dynode has a positive voltage bias relative to the previous one, and so electrons are accelerated from one dynode to the next. The increase in kinetic energy acquired by this process is sufficient to displace a number of electrons at the next dynode, and so on, causing large amplification by the end-stage dynode (usually tenth or twelfth).

scintillator

Figure 2.20. Schematic diagram of a photomultiplier tube and a photograph of a hexagonal 6 cm-diameter tube (inset). Light entering the PMT displaces a photoelectron which is electrostatically focused to the first-stage dynode. Each dynode has a positive voltage bias relative to the previous one, and so electrons are accelerated from one dynode to the next. The increase in kinetic energy acquired by this process is sufficient to displace a number of electrons at the next dynode, and so on, causing large amplification by the end-stage dynode (usually tenth or twelfth).

creasing potentials, leading to a gain of more than a million at the final dynode (anode). This high gain obtained from a photo-multiplier tube leads to a very good signal-to-noise ratio (SNR) for low light levels and is the primary reason for the success and applicability of photo-multiplier tubes for use in scintillation detectors. The only drawback of a photo-multiplier tube is the low efficiency in the emission and escape of a photo-electron from the cathode after the deposition of energy by a single scintillation photon. This property is called the Quantum Efficiency (QE) of the photo-multiplier tube and it is typically 25% for most of the photo-multiplier tubes. Different, complex arrangements of the dynode structure have been developed over the years in order to maximize the gain, reduce the travel time of the electrons from the cathode to the anode, as well as reduce the variation in the travel times of individual electrons. In particular, a fine grid dynode structure has been developed which restricts the spread of photoelectrons while in trajectory, thereby providing a position-sensitive energy measurement within a single photo-multiplier tube enclosure (Position Sensitive PMT or PS-PMT). More recently, a multi-channel capability has been developed which essentially reduces a single photo-multiplier tube enclosure into several very small channels. It uses a 2D array of glass capillary dynodes each of which is a few microns wide. Additionally, a multi-anode structure is used for electron collection, thereby providing a dramatically improved position-sensitive energy measurement with very little cross-talk between adjacent channels (Multi-Channel PMT, MC-PMT).

Photodiodes, on the other hand, are based upon semiconductors which, unlike the situation for detecting the photons, have high sensitivity for detecting the significantly lower energy scintillation photons. These detectors typically are in the form of PIN diodes (PIN refers to the three zones of the diode: P-type, Intrinsic, N-type). Manufacturing a PIN photodiode involves drifting an alkali metal such as lithium onto a p-type semiconductor such as doped silicon. Incident scintillation photons produce electron-hole pairs in the detector and an applied electric field then results in a flow of charge that can be measured through an external circuit. A significant disadvantage of the photodiodes is the low SNR achieved due to the presence of thermally activated charge flow and very low intrinsic signal amplification. In recent years, a new type of photodiode, called the Avalanche Photo Diode (APD), has been developed which provides an internal amplification of the signal, thereby improving the SNR. These gains are typically in the range of a few hundred and are still several orders of magnitude lower than the photo-multiplier tubes. More importantly, APD gains are sensitive to small temperature variations as well as changes in the applied bias voltage.

In general, there are three ways of arranging the scintillation crystals and coupling them to photo-detectors for signal readout in a PET detector. The first is the so-called one-to-one coupling, where a single crystal is glued to an individual photo-detector. A close-packed array of small discrete detectors can then be used as a large detector that is needed for PET imaging. The spatial resolution of such a detector is limited by the size of the discrete crystals making up the detector. In order to achieve spatial resolution better than 4 mm in one-to-one coupling, very small photo-detectors are needed. However, individual photo-multiplier tubes of this size are not currently manufactured. One solution is the use of photodiodes, or APDs instead of photo-multiplier tubes. The APDs are normally developed either as individual components or in an array, and so are ideal for use in such a detector design [10,11]. However, as mentioned earlier, the APD gain is sensitive to variations in temperature and bias voltage that can lead to practical problems of stability in their implementation for a complete PET scanner. Another option is the coupling of individual channels of a PS-PMT or a MC-PMT to the small crystals [12]. Due to the large package size of these photo-multiplier tubes, however, clever techniques are needed to achieve a close-packed arrangement of the crystals in the scanner design. Despite the very good spatial resolution and minimal dead time achieved by the one-to-one coupling design, the inherent complexity (number of electronic channels) and cost of such PET detectors limits their use at present to research tomographs; in particular, small animal systems.

The next two detector schemes are attempts at reducing these disadvantages by increasing the encoding for the detector. Both the designs involve the use of larger photo-multiplier tubes without intrinsic position-sensing capabilities. The Anger detector, originally developed by Hal Anger in the 1950s, uses a large (e.g., 1 cm thick x 30-50 cm in diameter) NaI(Tl) crystal glued to an array of photo-multiplier tubes via a light guide. This camera is normally used with a collimator to detect low-energy single photons in SPECT imaging. An application of the Anger technique to a PET detector, on the other hand, uses 2.5 cm-thick NaI(Tl) scintillators. An array of 6.5 cm-diameter photo-multiplier tubes can be used to achieve a spatial resolution of about 5 mm [8]. A weighted centroid positioning algorithm is used for estimation of the interaction position within the detector. This algorithm uses a weighted sum of the individual photo-multiplier tube signals

Siemens Pet Block Detector Bgo Crystal

Figure 2.21. A block detector from a Siemens-CTI ECAT 951 PET scanner is shown. The sectioned (8 x 8 elements) block of BGO is in the bottom left corner, with the four square PMTs attached in the center, and the final packaged module in the top right corner. The scanner would contain 128 such modules in total, or 8192 individual detector elements. (Figure courtesy of Dr Ron Nutt, CTI PET Systems, Knoxville, TN, USA).

Figure 2.21. A block detector from a Siemens-CTI ECAT 951 PET scanner is shown. The sectioned (8 x 8 elements) block of BGO is in the bottom left corner, with the four square PMTs attached in the center, and the final packaged module in the top right corner. The scanner would contain 128 such modules in total, or 8192 individual detector elements. (Figure courtesy of Dr Ron Nutt, CTI PET Systems, Knoxville, TN, USA).

and normalizes it with the total signal obtained from all the photo-multiplier tubes. The weights for the photo-multiplier tube signals depend exclusively upon the photo-multiplier tube position within the array. Since these detectors involve significant light sharing between photo-multiplier tubes, a high light-output scintillator such as NaI(Tl) is needed to obtain good spatial resolution. The use of large photo-multiplier tubes produces a very high encoding ratio, leading to a simple and cost-effective design. However, a disadvan tage of this detector, independent of the use of NaI(Tl) as a scintillator, is the spread of scintillation light within the crystal which leads to significant detector dead time at high count rates.

The block detector design uses Anger positioning in a restricted manner to achieve good spatial resolution and reduced dead time at the expense of a lower encoding ratio. The initial design used an 8 x 4 array of 6 x 14 x 30 mm3 BGO crystals glued to a slotted light guide [13]. The slots in the light guide are cut to varying depths with the deepest slots cut at the detector's edge (see Fig. 2.21, left and centre).

The read-out in this block design is performed by four 25 mm-square photo-multiplier tubes. The slotted light guide allows the scintillation light to be shared to varying degrees between the four photo-multiplier tubes depending upon the position of the crystal in which the interaction takes place. The centroid calculation is performed here as well to identify the crystal of interaction. An improved design of this detector allows the identification of smaller, 4 x 4 x 30 mm3, leading to an improved spatial resolution but with smaller 19 mm photo-multiplier tubes. Besides the advantages and disadvantages of BGO as a scintillator, the block detector design has the benefit of reduced detector dead time compared to the large-area Anger detector due to the restricted light spread. This, however, is achieved by increasing the number of detector channels (lower encoding ratio), thus leading to increased cost. A modification of the block design, called the quadrantsharing block design [14], can distinguish smaller (half the size in either direction) crystals by straddling the 19 mm photo-multiplier tube over four block quadrants (see Fig. 2.22, right). This design, in comparison to the standard block, results in a better spatial resolution with almost double the encoding ratio, but increased detector dead time due to the use of nine photo-multiplier tubes (not four) for signal readout from an event.

Block Detectors Pet
Figure 2.22. The standard block detector design from the side (left) and looking down through the crystals (middle). The quad-sharing block design as seen from the top through the crystals is shown on the right. Figures are not drawn to scale.

Figure 2.23. Schematic representation of detecting coincidence events in two detectors. Signal A results in a trigger pulse 1 which marks the start of the coincidence window of width At. Similarly, signal B results in a trigger pulse 2. A coincidence (AND) circuit then checks for coincidence between the pulse 2 and the coincidence window.

Time

SIGNAL A

SIGNAL B

COINCIDENCE WINDOW

Time

SIGNAL A

SIGNAL B

COINCIDENCE WINDOW

DETECTOR B

t2

kV2 V

Time

Time

Was this article helpful?

+8 -1

Responses

  • ines
    How valence band & conduction band produced in NaI?
    4 years ago
  • lina
    Do pet detectors have a septa attached to the detector?
    2 years ago

Post a comment