11. Resultados
11.3 Correlación de los resultados obtenidos por la muestra en la prueba de las
11.3.2 Correlación de los resultados obtenidos por la muestra en la prueba de
Scintillators
A number of gamma detectors have been and continue to be investigated for PET, including semiconductors (e.g., CdTe and CZT), Liquid Xenon, and ceramics (e.g., GluGag); however, all modern detectors for whole-body PET are based on inorganic scintillators, and so the discussion will be focused on this technology. Scintillators
function in nuclear medicine imaging to convert the energy of gamma rays into light that is collected by a photodetector, and may be broadly classified as either organic or
inorganic. Organic scintillators make up a large variety of scintillators, including liquid and plastic scintillators; however, because of their improved detection efficiency, all scintillators used for clinical scanners are inorganic, and the discussion henceforth will be limited to this type. For these scintillators, excited electrons result in three types of
scintillation: 1. An electron excited from the valence band to the conduction band, leaving electron-hole pairs, may return to the valence band, resulting in fluorescence. 2. An excited electron forbidden from returning to the ground state absorbs thermal energy, resulting in phosphorescence, which generally has a longer wavelength and a longer characteristic decay time. 3. Quenching, in which a transfer of thermal energy from certain excited states to the ground state occurs without radiation and again with a long time constant, resulting in a decrease in conversion efficiency and long afterglow. Activator dopants are often added to these scintillators to create energy levels in the forbidden gap, in order to increase the efficiency of scintillation and allow the
wavelength of the emitted photons to be in the visible range as the electrons return to the valence band [50].
The qualities that make up an ideal scintillator are: high sensitivity (high density), a high light output (to improve energy and timing resolution), a large fraction of incident photons converted to prompt fluorescence, scintillation light that is transparent to the scintillator (to avoid reabsorption) and compatible with the photodetector absorption spectrum, an output that is proportional to the energy of the incident photons over a wide range, short decay time (both to limit the coincidence window and to improve TOF
resolution), a refractive index close to that of glass (the photodetector encasing), as well as good properties for commercial manufacture (i.e., cost, availability, ease of
manufacturing). No scintillator exists that maximizes performance in each category, requiring a trade-off in performance; some of the most common scintillators with relevant characteristics are shown in Table 1.2 [21, 47, 83-84].
Table 1.2: Properties of some of the scintillators used for PET.
Cerium-doped Lutetium Oxyorthosilicate (LSO) was introduced in the 1990’s, and Cerium-doped Yttrium Lutetium Oxyorthosilicate (LYSO) later; since then, the quality of their production has improved enough that they are the most common
scintillators used in current clinical scanners, because of the good balance between high
Scin%llator Rela%ve Light Output Density (g/cm3) Decay Constant (ns) A?enua%on Length for 511 keV (mm) Max Emission Wavelength (nm)
NaI (Tl)
(Thalium-doped Sodium Iodide)100
(~38000
ph/MeV)
3.67
230
23
410
BGO
(Bismuth Germinate)15
7.13
300
10.4
480
LYSO
(Cerium-doped LuteKum YLrium Oxyorthosilicate)75
7.3
40
11.4
420
LaBr
3 (Lanthanum Bromide)160
5.29
15
22.3
380
GSO
(Gadolinium Oxyorthosilicate19
6.71
30-60
14.1
440
light output, fast decay time, high density, and an emission spectrum that matches the absorption spectrum of common photodetectors well; additionally, its emission does not consist of components with a slower decay time. The performance of these two crystals is generally quite similar, with some studies indicating differences with regard to
afterglow, and a slightly lower density for LYSO. These scintillators allowed for the implementation of TOF in modern PET scanners in the 2000’s, because the fast
scintillation time was paired with a high sensitivity, unlike any TOF scintillator until that point.
LYSO/LSO do exhibit a number of features that must be considered when using them. Both scintillators have an afterglow (phosphorescence) when exposed to radiation, resulting from charge trapping within the scintillator; this afterglow gradually decays with time and requires that the scintillators be shielded from outside light. At the low count rates typically encountered in the lab, this afterglow does not pose a problem, though baseline shifts have been reported at higher count rates. Additionally, because Lu3+ is naturally radioactive, these scintillators exhibit a natural background of ~300 cts/s/cc, though this does not exhibit a strong effect on the coincidence rate. The light output for these also exhibit a non-proportionality with respect to incident energy, driving their energy resolution to be slightly worse than that of GSO, despite a higher light output [4, 85-91].
Photodetectors
The photodetector converts the energy of the scintillation light into an electrical signal through the photoelectric effect. The SNR provided by a photodetector is a
function of dark (or thermal) noise in the detector as well as the quantum efficiency, defined as the number of electrons produced per incident photon, which determines the noise resulting from the statistical nature of photodetection. The quantum efficiency is determined by a number of factors, including light reflection at the protective glass covering a photodetector; the photocathode material, which determines the absorptive efficiency and reflectivity of the photocathode; and the thickness of the photocathode, which affects both the absorptive efficiency and the number of electrons that escape the cathode. The resolution allowed by the photodetector is a function of sampling width, determined by the sampling pitch of the detector, and the amount of signal averaging, determined by the detector aperture width. The sensitivity of the photodetector is wavelength-dependent and largely determined by the spectral response of the
photocathode, which is a function of its composition. Last, another important metric is the pulse rise time resulting from scintillation pulses, determined by the quantum
efficiency of the photodetector as well as the electron multiplication scheme. Through the years, photodetectors have evolved greatly in their method of operation, physical design, and performance; details on some of the important current photodetectors are given below [4, 57].
Photomultiplier Tube
The photomultiplier tube (PMT) has been the workhorse of nuclear medicine since its inception. Gains are typically ~4-6 electrons per dynode, resulting in an amplification for 10 stages of ~106-107. The maximum quantum efficiency for those typically used is ~25-35%. Some of the advantages of PMTs include their reliability and low noise, with thermal emission of electrons orders of magnitude lower compared to the
current induced by a photoelectric event. Photocathode nonuniformities resulting from variations of photocathode thickness (especially in large area PMTs), result in
nonuniform sensitivity, as well as nonuniform collection of photoelectrons at the first dynode depending across the photocathode area [4, 50].
Multianode PMT
Multianode PMTs (MAPMTs), or position sensitive PMTs, offer two benefits compared to PMTs: compact size, with a length of ~12mm compared to the length of a PMT of ~150mm; and a grid of anodes that allows for positioning information (Figure 1.9). For the purposes of this thesis, this discussion will focus on the Hamamatsu H8500. A fundamental requirement of position sensitive PMTs is that the process of electron multiplication retains the spatial separation of the original electron cloud resulting from the photoelectric interaction. MAPMTs may be constructed using a number of
techniques to do this, including the use of a fine mesh layer to channel the electrons from one dynode layer to the next; the H8500 uses 12 stages of metal channel dynodes that are arranged to channel electrons between layers. To retain the positional information at the anode, each anode is read out separately for the H8500. Recent models are compact, with an active area of 49.7mm x 49.7mm (total area of 51.7mm x 51.7 mm), arranged in an 8x8 grid of anodes (each 6mm x 6mm), and a ~2mm thick window. The cross talk between the anodes is ~1%, and the performance metrics for the H8500 (e.g., quantum efficiency, gain, timing, dark current) are comparable to standard PMTs. The gain of the anodes may vary by a factor of 3 within the array, because of both nonuniformities in the thickness of the photocathode and variations in the efficiency in collecting photoelectrons
as a function of emission position; this may be corrected to some extent using gain correction factors. Another disadvantage of some MAPMTs is the presence of
nonlinearities near the edges of the detector, resulting in an unusable area that reduces the sensitivity of the detector [92-95]. The MAPMT was chosen for this work because its active area matched the size of the crystals studied in this work well and because of its favorable noise properties.
Figure 1.9: Photograph and diagram of the H8500 MAPMT, used throughout this work. Top Left: Photograph of Hamamatsu H8500 MAPMT. Top right: Diagram of the 64- anode layout. Bottom: Diagram of a continuous crystal coupled to the H8500 MAPMT, viewed from the side.
49mm
64-channel
grid of 6.1-mm
anodes
49mm m 6.125mm mSilicon Photomulitplier
Silicon photomultipliers (SiPMs) detect radiation by using the junction between n-type and p-type silicon, created by doping, to create an electrical current when electron- hole pairs are formed after a photon deposits energy in the photodetector; an electrical field (generated by an applied bias voltage) increases this current. SiPMs used in scanners are designed as a 2D array of pixels ranging from 1-4mm, each divided into thousands of micropixels composed of avalanche photodiodes, which operate in Geiger mode. While each micropixel registers the same output after absorbing enough energy, the collection of many micropixels allows for the output to be linear with total absorbed energy. Since their introduction into the field a decade ago, research into their use has blossomed and produced detectors that offer a number of advantages: compactness, cheapness, insensitivity to magnetic fields (important for incorporation in a PET/MR scanner), high gain (up to 106), good timing performance (recent detectors have a timing resolution <250ps when coupled to small scintillators), and a large quantum efficiency. This has lead to the introduction of the Philips Digital Photon Counter, which is a fully digital detector, with electronics (e.g., the analog-to-digital converter) built into the readout chip [16]. Aside from less readout electronics, this detector offers the added advantages of an improved dark count rate and timing performance. The major disadvantages of SiPMs have been the dead area between the pixels, the temperature dependence of the performance, relatively large bias voltage required and the dark count rate, though in recent years the latter has decreased enough that they are used in clinical scanners with cooling [4, 38, 47, 37, 92, 96-99].