Physics, Techniques and Procedures
(or positron emission tomography) is a tomographic nuclear imaging procedure, which uses positrons as radiolabels and positron - electronannihilation reaction-induced gamma rays to locate the radiolabels. Current clinical indications for PET imaging include the location of epileptic foci, the grading of brain tumours, the assessment of cerebral and cardiac perfusion, the identification of hibernating myocardium and tumour staging (Fig.1). PET is an important research tool for the assessment of cerebral function, metabolism and receptor ligand systems. These applications as well as the use of PET for the detection of foci of acute inflammation are being intensively investigated at all major PET centres. The principal radiopharmaceuticals used are fluorodeoxyglucose FDG , O 15 water and N 13 ammonia.
The PET principle is as follows. A low dose of a radiopharmaceutical labelled with a positron emitter such as C-11, N-13, O-15 or F-18 is injected into the patient, who is scanned by the tomographic system. Scanning consists of either a dynamic series or a static image obtained after an interval during which the radiopharmaceutical enters the biochemical process of interest. The scanner detects the spatial and temporal distribution of the radiolabel by detecting gamma rays during the so-called emission scan. The events leading to this detection are (Fig.2):
1. the positron is emitted by a beta decay,
2. it is slowed down to small speeds which are necessary for
3. the annihilation reaction between the positron and a shell electron of a neighbouring atoms to occur. The distance the positron travels (mean free path) depends on the energetics of the beta decay but is typically one or a few millimeters.
4. The annihilation reaction produces two 511 keV gamma rays which travel in almost exactly opposite directions (this is due to the conservation of energy and momentum laws).
5. The two gamma rays are detected by a coincidence counting detection system (see below).
6. After proper filtering the collected raw data sinograms are reconstructed into a cross-sectional image.
In a coincidence counting system, the detection of one gamma ray by one detector results in the opening of an electronic time "window", during which detection of a gamma ray in another detector results in the counting of a coincidence event (Fig.2). It is obvious from the geometry of the gamma ray emission that the radioactive decay leading to the detection of such a coincidence event must have occurred along a straight line connecting the two detectors. Since the probability of absorption of the two gamma rays is independent of the position of the event along that imaginary line, PET is an inherently quantitative imaging method allowing the measurement of regional concentrations of the radiopharmaceutical injected after proper calibration. To achieve quantitative detection, several problems have to be overcome. First, a correction has to be introduced for elastic scattering (scatter correction), while Compton scattering can be corrected for to some extent by using an energy discriminator on the gamma ray detectors. Second, a correction for random coincidences has to be introduced. The frequency of such random coincidences (two gamma rays detected which come from different decay events) depends on the imaging system. They are more frequent the higher the concentration of the activity in the patient and are more frequent during three-dimensional data acquisition (see below). Third, an absorption correction matrix has to be acquired to correct for tissue absorption. This is done by acquiring a transmission PET scan prior to or after the injection of the radiopharmaceutical. Typically, a germanium (see periodic table of elements) radiation source is rotated around the patient and the attenuation of this radiation by the patient recorded and reconstructed into a transmission scan, which is then used to correct the data of the emission scan. Alternatively, the correction is made by assuming that the attenuation values for the imaged pixels is homogeneous, an assumption which is only reasonable for brain scans. The reconstructed data are transaxial slices of a few millimetres in thickness. Depending on the application they are reconstructed into coronal (Fig. 1), sagittal or oblique sections for clinical reading.
Dynamic data acquisition is performed when the data are to be quantified. In this technique, scans are acquired over times as short as 30 seconds. The most important technique for analysis of dynamic data is input deconvolution. Using this mathematical technique, quantitative information on tissue perfusion (see perfusion imaging) and other parameters can be extracted (see compartment modelling).
Usually data are acquired in a two-dimensional (2D) acquisition mode with a PET scanner consisting of several detector rings (Fig.3). In 2D imaging the detectors are separated by lead septa which permit coincidences to occur between opposing detectors (red lines, Fig. 3 left) or adjacent detectors (green lines, Fig. 3 left). In the three-dimensional (3D) mode the septa are retracted and coincidences between all possible detector banks are admitted (Fig. 3, right). While data acquisition in 3D mode has become standard, whole-body imaging is currently done in 2D or 3D mode and cardiac imaging in 2D mode. The major problems with 3D data acquisition are the large number of random coincidences counted and the long data reconstruction time, but the latter problem is alleviated by increasing computer performance.