PET uses natural compounds such as water or glucose labelled with positron emitters such as 15O (H215O) or 18F (18F-fluorodeoxyglucose or FDG) to measure blood flow or glucose metabolic activity in the human brain, heart, etc. These small quantities of labelled compound are injected intravenously into our subjects. The two 511 keV gamma rays emitted back-to-back when the ejected positron stops in tissue (within 1-2 mm of the labelled atom) are detected by our PET Camera. We detect the two 511 keV gamma rays in coincidence (i.e. at the same time) to record a positron decay. We then know the positron decay, for example of the labelled glucose atom, lies along the line joining the 2 detection sites (see Figure).
Imagine for example you had been injected with 18F-FDG (labelled glucose) when you started reading this guide. The glucose, like unlabelled or (cold) natural glucose, is taken up most in the more metabolically active parts of your body such as your heart and brain. Specifically, the act of reading (and hopefully comprehension) mean that Wernickes area of your left temporal lobe is particularly metabolically active and therefore uses more glucose. Thus, a PET scan would show increased (labelled) glucose uptake in that area of your brain: it would appear as a bright area of the brain because of the accumulation of 18F-FDG detected by the camera.
PET is unique in being able to measure physiological activity in vivo in real time. It makes absolute measurements in terms of flow (ml/min/100g) or metabolic rates (mg/min/100g of tissue). It can do this because we are able to obtain absolute activity concentrations in the brain or body by performing accurate corrections for the absorption of gamma rays in tissue.
From the corrected regional intensity of positron emissions (from the coincident gamma rays) we may obtain the activity concentration in each volume element (or voxel of 2 × 2 × 2 mm) in the brain or body. This is a measure of how metabolically active that part of the brain or body is. By relating this map of physiological activity to a corresponding map of the patient's anatomy we can see where there is abnormal physiology or how we normally respond physiologically to various stimuli. MRI scans of PET study subjects are usually taken to provide the required anatomic information. The applications of the PET technique are many. Examples performed at our PET Center include: (i) finding metabolically active tumors in the body via 18F-FDG uptake, (ii) detecting ischaemia in the human myocardium (via FDG or 13NH3 uptakes), (iii) measuring the metabolic rate (18F-FDG) and blood flow (H215O) changes in the brains of healthy volunteers and patients at rest and when performing cognitive tasks and responding to emotional stimuli.
In a number of our studies, for example of the healthy and diseased workings of the human brain, we are able to use our activity concentration measurements to calculate absolute metabolic rates or flow using a kinetic model. A schematic overview of such studies is shown in Figure 2.
Many of the functional shifts and deficits in the human brain now being studied by PET groups are very subtle. Often they become clearer with activation or demands on the cognitive or emotional functioning of subjects. These activations are most often studied by measuring regional blood flow in the brain, which is expected to increase with increased synaptic activity.
While baseline differences in brain function between controls (e.g. as a function of age) and between controls and patients (e.g. in schizophrenia or Parkinsons disease) continue to be studied there is more and more interest in activation studies.
Since there is a natural variability in brain physiology both between controls and in patients (as a function of disease severity) many PET groups now take baseline (no activation) and activation scans in each subject so the subject may act as his/her own controls. Thus a CBF study may involve 3-5 scans (1-2 baselines, 2-3 activations) in the same subject. This is possible because the 15O used to label H2O has a 2 minute halflife and so decays away between scans. The maximum dose used for such studies is 75 mCi in total, so that no more than 15 mCi is injected per measurement. This is considerably less than the 20-50 mCi injected in single bolus scans in PET cameras with septa (septa are tapered lead-tungten rings used to exclude out-of-plane coincidences).
The desire to perform multiple 15O-water cerebral blood flow (CBF) scans has meant that septa are retracted in PET scanners with septa to improve sensitivity. The PET scanners at HUP have the unique advantage that their standard operating configuration is without septa, their scatter fraction (background) is substantially lower than other PET cameras without septa (15% vs 30-40%) and the peak countrate capability of 120,000 cps trues is reached at 1 mCi in the FOV (5-6 mCi injected). The new HEAD PENN PET scanner with 25.6 cm axial FOV and no septa has 1.5 times the sensitivity of the CPET scanner making it even more suited to investigating subtle changes in brain physiological activity.
We generally perform 6 patient protocols per day, 4 days a week. A number of these protocols consist of more than one study. or up to 120 scans (for multiple ramped infusion studies) per subject. Common study types are listed below.
| Brain Studies | Radioisotope | type | for | # of studies | duration | population | H215O | ramped | CBF | 5 | 90m | epilepsy, healthy | equilib infusion | 5 | 100m | schizophernia | C15O | respiration | CBV | 3 | 40m | stroke risk | 18F-FDG | bolus (late scan) | CMR(glu) | 40m | ageing, depression | 18F-spiperone | bolus | D2 receptor | 90m | Parkinsons | Body Studies | Radioisotope | type | for | # of studies | duration | population | 18F-FDG | 1m infusion | myocardial | 1 | 90m | cardiac | attenuation scan | viability | 18F-FDG | bolus (late scan) | cancer | 3-15 | 90-120m | cancer | 13NH3 | 30s infusion | myocardial | 2 | 20 m | cardiac |
The plan of the PET Center is shown in Figure 3. The functions of the various rooms should become clear from the description below.
The positron emitters 18F, 15O, 13N and 11C are produced in a JSW Minicyclotron 400 meters from the PET Center. The reactions used and chemical forms of the isotopes are:
| Energy | particle | target | reaction | chemical form | 15 MeV | 2H or d | 15N2 | 15N(d,n)15O | H215O, 15O2, C15O | 8 MeV | 1H or p | H218 | O 18O(p,n)18F | 18F-FDG, 18F-spiperone | 16 MeV | p | 16O | 16O(p,a)13N | 13NH3 | 8 MeV | p | 14N2 | 14N(p,a)11C | 11C-acetate |
The longer lived isotopes: 11C (20.3 mins), 13N (9.96 mins) and 18F (109.8 mins) may be delivered via a pneumatic transport system (in sealed capsules) from the cyclotron to the Hot Laboratory in the PET Center. 15O is delivered to the PET suites themselves via gas lines (directly from the cyclotron): H215O via blue lines; 15O2 or C15O via red lines. In the case of H215O delivered as water vapor the gas is collected in a specially designed cart where it is diffused into a sterile saline water reservoir prior to infusion via a venous line into the subject. 15O2 or C15O is inhaled via a mask.
In all cases the positron emitter labelled compounds are subject to pyrogen and sterility tests (Cyclotron and Microbiology Laboratories) and chemical purity (chromatography). Activity levels are monitored prior to injection.
For all quantitative brain studies it is necessary to sample arterial blood during uptake and scanning in order to obtain the input function (activity concentration in the blood delivered to the brain). This is achieved by inserting an arterial line (catheter) into the radial artery under lidocaine local anaesthesia.
For most studies the radiolabel is administered via a venous injection or infusion. The various protocols are given below:
CMR(glu) (cerebral metabolic rate for glucose), CBF (cerebral blood flow), CBV (cerebral blood volume), OER (oxygen extraction ratio) studies:
| Study | arterial line | isotope | injection | cancer | yes | 18F-FDG | bolus 30-90 m prior to scans | cardiac | yes | 18F-FDG | 1 m infusion | cardiac | yes | 13NH3 | 1 m infusion | brain trauma | yes | 18F-FDG | bolus 30 m prior to scan | brain: CMR(glu) | no | 18F-FDG | bolus 30m prior to scans | brain: CBF | no | H215O | equilib infusion 8m prior and thoughout 80 m scan | brain: CBF | no | H215O | ramped infusion 4 m | brain: CBV | no | C15O | inhalation | brain: OER | no | 15O2 | inhalation |
Note: for all studies post-injection singles transmission scans (1 min/axial position) are peformed Since the emission and transmission scans are interleaved (ie they follow each other for each bed position) and the transmission scans are so short, there is little likelihood of patient motion between emission and transmission scanning which may cause misregistration artefacts in the attenuation corrected images.
Once the relevant lines have been inserted and any (e.g.
EEG or psychological) tests completed, the subject is positioned centrally
in the PET Camera port so that, for example, their brain is centered in
the 256 mm diameter FOV for a brain study or their heart is well centered
axially in the (256 mm axial field) camera for a cardiac study. The subject
is restrained from movement during the scan as much as possible. Blood
aliquots of 0.25 ml are sampled during uptake and measured in the Tennelec
NaI(Tl) well counter in the Bioassay Laboratory as whole blood or, after
centrifuge, as plasma. This provides the arterial activity concentration
in nCi/ml via the relation:
| Isotope | L / min | calibration factor | 18F | 0.006313 | 0.002367 | 15O | 0.3409 | 0.002599 | 13N | 0.06956 | 0.002352 | 11C | 0.00340 | 0.002330 | 68Ga | 0.01018 | 0.003249 |
For 18F-FDG studies of the brain used to calculate CMR(glu) (cerebral metabolic rate for glucose) the patient is generally asked to fast prior to the study to reduce the dilution of labelled glucose (18F-FDG) by the higher levels of natural or "cold" glucose and insulin present in the blood stream after a meal. Cold glucose levels are measured from the sampled blood using a Beckman glucose analyser situated in the Bioassay Laboratory. For cardiac FDG scans the subject is loaded with 50-100 ml of glucose an hour before scanning to increase FDG transport into myocardial muscle. For whole body cancer studies subjects are asked to fast for 4 hours prior to the study to reduce FDG uptake in the heart which may mask nearby tumor uptake. Blood glucose is monitored during these scans.
In addition to the NaI(Tl) well counter, centrifuges (for plasma separation) and computer interface, the PET Center also has a dynamic blood sampler. This is used in order to provide finer time sampling of blood activity during slow bolus H215O infusion studies. The sampler is based on 2 BGO crystals (coupled a single PMT) on opposing sides of an arterial sampling line. 1022 keV summed gamma energy is required for coincident 511 keV detection.
The PET Center has two Positron Emission Tomographs: an ADAC-UGM CPET scanner with a 25.6 cm axial FOV and a HEAD PENN PET scanner also with a 25.6 cm axial FOV. Both of these PET Cameras are manufactured by UGM Medical Systems, Philadelphia. Schematic diagrams of these type of scanners are shown in Figure 4. The CPET scanner characteristics are listed below. The scanner comprises six curved sodium iodide (NaI(Tl)) crystals of size 50 × 30 × 2.5 cm. Each crystal is viewed by 48 photomultiplier tubes (PMTs) of 5 cm × 5 cm extent. Coincidences are accepted between opposing detectors (on pairs) and the 2 detectors adjacent to the opposing detectors (off pairs). There are no septa in this camera and sampling is in 2 mm × 2 mm × 2 mm voxels.
CPET Scanner characteristics
The simplified camera electronics are illustrated in Figure 5. The light from a 511 keV gamma-ray interaction in the NaI(Tl) crystal is summed for an energy signal (E), and the light distribution over the PMTs yields a position (X,Z) signal. Only coincidence gamma-rays from a positron decay are accepted. The basic pre-processing (performed via the rack electronics and computers within the PET suites) are:
(1) The preamplifier signals are digitized and integrated over 160 ns in the digitizer. This integration time is much shorter than that used in BGO PET Cameras (typically 1000 ns) since the NaI(Tl) crystal light output is high. Full integration of the light would lead to better spatial resolution but also higher deadtime. The controller passes on the digitized signals for a coincidence event (determined in the coincidence logic unit) to the position calculator.
(2) The calculator calculates the energy and position of the event. The calculation for each detector in coincidence is performed in parallel in two calculators.
(3) The events are corrected for detector spatial distortions using a lookup table, and are gated on energy (460-580 keV), field of view (e.g. 256 mm diameter) and axial angular acceptance (±15°). This step rejects scattered and randoms events.
(4) Finally, the data are rebinned into a maximum of 128 sinograms with 2mm axial separation.
Typical sinograms are a matrix of 256 "rays" (1mm apart for 256 mm FOV) × 192 angles (0.9375° apart). Each sinogram is essentially a stack of projections. This is most readily understood for a human brain distribution in the camera: See Figure 6a. The corresponding sinogram for a body activity distribution is illustrated in figure 6b. The set of sinograms are then available for image reconstruction on the Sun SPARC Station and array processor.
The large crystal detectors of the PENN PET cameras are divided into smaller zones for coincidence detection, energy calculation and position calculation of coincident 511 keV gamma rays. For the CPET camera, there are 3 coincidence triggers (16 PMTs each), 5 zones for light signal integration (24 PMTs each) and 16 "FOV"s (fields of view) of 9 PMTs each used in the position centroid calculation. The effects of these zones are: First, the processing of coincidences is speeded up and there is also less event rejection (loss) due to pileup, compared to the camera without zones This reults in a higher countrate capability. Given in Figure 7 are the countrate curves for a 20 cm dia x 30 cm long cylinder (body like) and for a 20 cm dia x 20 cm cylinder (brain-like) object in the FOV. The 1.1 liter object The resulting sinogram (scan) statistics for typical studies are given in the table on page 9.
Secondly, in zone configuration the camera calculates the positions of coincidence gamma rays using smaller areas of the detctor crystal so that simultaneous events (gamma rays striking the detector) will not affect (distort) the true position calculation. This results in very stable axial positioning even at the highest countrates: this is illustrated in Figure 8 (for a PENN PET Quest camera) in which there is no significant axial mispositioning even at 3.4 mCi in the FOV. Thus high image quality is acheived even at high countrates.
The HEAD PENN PET scanner is based around a single annular NaI(Tl) crystal 42 cm in diameter and 1.9 cm thick. 180 PMTs are coupled to this crystal via light guides. The crystal is divided into zones with 15 trigger channels. This camera is now used routinely for clinical and resting quantitative brain studies. The position resolution has been measured at 3.7 mm FWHM and the sensitivity is around 650 kcps/microCi/cm3. Due to the large axial acceptance angle of this Volume Imaging Camera (±28°) 3D rebinning and reconstruction is performed.
From a practical point of view we need to generate images of physiological activity of good image quality with accurate and precise quantitation. For accurate quantitation we need to subtract scatter and randoms (accidental coincidences) background and to perform quantitative attenuation correction (to correct for the gamma rays absorbed in the body) and correct for camera deadtime at high countrates. For precision we need high scan count densities to measure activity concentrations in the brain to better than ± 3%. These requirements are summarised below along with typical scan parameters for our key study types which show the statistical quality of our data.
Clinical and Patient Research Studies: High Countrate Imaging with the PENN-PET Cameras require:
| Study | mCi | true coinc | duration | total counts | counts/slice | type | in FOV | [kcps] | [mins] | [Mcts] | [Mcts] | FDG | 0.2 | 50 | 30.0 | 110 | 5.5 | OXY (equil) | 0.4 | 70 | 10.0 | 40 | 4.0 | OXY (slow bolus) | 0.5 | 80 | 3.5 | 15.0 | 1.5 |
The PENN-PET Cameras are calibrated to generate activity
concentrations in tissue (after background subtraction and attenuation
correction) from the relation:
where cpm/voxel is the measured count density (counts per minute per voxel), the calibration factor is 3.2 ± 0.1 nCi/ml/cpm/voxel (s.d: 3%) and is independent of object size and is constant over time. The ratefactor corrects for camera deadtime and is defined as: ratefactor = 1/(1-deadtime). For a volume imaging scanner the deadtime is determined by the activity (mCi) in the field of view (independent of the size of the object in the FOV), and the activity in the FOV is measured by the average singles countrate at the detector crystals. These points are illustrated in Figures 9. In practice we measure the singles countrate during a scan and use these numbers to calculate the ratefactors. The resulting activity concentrations are within ± 5% of true activity concentrations for activities up to 2.5 mCi in the FOV.
The PENN PET Center staff perform full image reconstruction, with corrections for attenuation, scatter, randoms, deadtime and decay, and generate the appropriate blood curve data within 1-3 days of a research scan. Within a week the data is available to investigators. Whole brain CBF or CMR(glu)'s will have been calculated by the PET Center Group as a cross check of quantitation. In addition all data is backed up and relevant films generated. In the case of clinical studies reconstruction and filming should be completed on the same day as the study.
Since 1991 quantitative studies have been performed in over 1500 subjects. The majority of these have been 18FDG (fluorodeoxyglucose) and H215O brain studies. The range of research brain studies is outlined here:
| controls: | patient groups: | mood induction | schizophrenia | aphasics | learned helplessness | Parkinsons disease | brain tumor recurrence | pain centers | Alzheimers | stroke risk | ageing population | Seizures | cocaine addicts | language processing | multiple sclerosis | childhood neuroblastomas |
In addition quantitative body studies are routinely performed, namely: (1) whole body cancer studies (FDG): over 2500 subjects with breast, lung, liver, colorectal, renal, ovarian, prostate and stomach cancer have been scanned, (2) FDG and ammonia cardiac studies: over 200 subjects have been studied via FDG myocardial uptake and/or NH3 (myocardial blood flow) uptake, (3) FDG brain studies: for seizure as well as brain trauma patients.
Currently around 25 subjects are studied per week although the PET Center is capable of performing 50 subject studies per week. Computing and analysis facilities in the PET Center are:6 Sun computer terminals, 10 hard disks (total >40GB), 5 optical disk drives and 2 exabyte tape drives (for long term data archival). In addition, available from the PET Center are PETview software (ROI, oblique reslicing, whole body and T-S-C display, etc) and programs to generate CMRglu and CBF (equilibrium infusion studies, also blood flow images), as well as modelling packages for metabolic rates, blood flow and receptor density calculations.
The dedicated brain scanner (HEAD PENN PET) is used for
clinical and research studies, including brain trauma, brain cancer and
children/infant studies. Thus both whole body (cardiac or cancer studies)
and brain studies may be performed simultaneously. The PET Group provides
all necessary support to investigators new to PET: from processing of images
and blood curves, to programs for reslicing and registering PET and MRI
images, to quantitation of regional CBF or metabolic rates. In addition,
clinical PET studies are routinely performed in the PET Center at HUP in
the areas of:
Oncology
Regional analysis using predefined or MRI measured anatomic templates is performed by the individual investigators. There are currently more than 15 approved or pilot projects underway at the PET Center. A very brief description of some of these projects and their objectives are listed below. The PET Center is still expanding and extending the scope of studies both in research into the workings of the human brain and into clinical applications for reliable diagnosis and treatment management of cancer, epilepsy and cardiac disease.
(1) Ageing and Dementia
| To Study: | Deficits in cognition with ageing and dementia | Measure: | rCBF, rCMR(O2), rCMR(glu) | Objective: | To correlate changes in function (intellect, memory, judgement) with physiological changes in both normal ageing and dementias. Cross correlated with anatomic changes (MRI scans) |
(2) Schizophrenia: cognition and emotional function
| To Study: | Deficits in language memory. Abnormal emotional responses. | Measure: | rCBF, rCMR(glu) | Objective: | To correlate language and memory performance with pattern of physiological activation as a function of stimulus complexity and of context. To measure emotional response to faces and prose with and without emotional content and physiological abnormalities compared to controls |
(3) Schizophrenia: neuropathy vs abnormal activation
| To Study: | Neuropathy (type II) vs abnormal activation (type I) in schizophrenics. | Measure: | D2 receptors, CMR(O2), CMR(glu) | Objective: | To correlate physiological responses to cognitive tasks with schizophrenia syndromes to test the hypothesis that schizophrenia is of 2 types. Type I: normal anatomy but higher D2 receptor density giving overactive subcortex and higher left hemispheric rCBF with delusions/hallucinations. Type II: anatomic atrophy, lower metabolism with hypofrontality. |
(4) Seizure
| To Study: | Seizure and surgery as a model for brain injury and brain plasticity | Measure: | rCBF | Objective: | To use temporal lobe epilepsy as a model of brain injury to see redistribution of function in the brain (plasticity) following injury (lesion responsible for seizures) or insult (surgical intervention) to the brain as a function of age of injury and time after insult. |
(5) Parkinsons Disease
| To study: | Subcortical dementia in Parkinsonism (deficits in integrative function) | Measure: | rCBF | Objective: | To investigate the subtle subcortical dementias accompanying Parkinsonism and resulting from loss of integrative function. Present simple and complex tasks (attention and integration mediated tasks) to follow any physiological deficits. |
(6) Cancer (renal, breast, lung)
| To study: | Glucose uptake in metabolically active tumors | Measure: | FDG uptake | Objective: | To determine the ability of whole body scanning to detect and assess metabolically active tumor sites pre and post treatment. |
(7) Cardiac
| To study: | Myocardial ischaemia and viability | Measure: | rMMR(glu), rMBF (with 13NH3, 82Rb) | Objective: | To detect and measure ischaemia and necrosis in the myocardium |
Figures:
