European Nuclear Medicine Guide
In general, any biomedical imaging equipment consists of 1) an imaging acquisition system, 2) a digital computer, and 3) an image display interface. The data acquisition module is responsible for converting the signal that contains diagnostic information (i.e., radiation of interest) to a digital signal. The digital image is represented by a matrix of digital values that can be read by a computer and displayed or modified in order to emphasize the diagnostic information.
Acquisition systems are designed to ensure an optimal 1) spatial resolution, 2) detection efficiency (regarding to dead time), and 3) energy resolution.
Nuclear Medicine image acquisition can be divided in three categories:
Single photon detection/imaging, i.e. planar and tomographic (SPECT)
Positron annihilation detection/imaging, positron emission tomography
Physiologic triggered imaging (gated techniques of the two above)
|
|
annihilation detection/imaging |
Single Photon detection/imaging |
|
Static |
n/a |
Planar projection |
|
Dynamic |
List mode |
Short planar acquisition |
|
Tomography |
Only (PET) |
Yes (SPECT) |
|
Gated |
List mode |
Yes |
The process variables taking part in image acquisition are:
Matrix size
Acquisition time / condition
System resolution
System sensitivity
System dead time resolution
Energy Resolution
The matrix size should be chosen so that the pixel size is less than half the reconstructed spatial resolution. Moreover, it is essential to consider the opinion of the nuclear medicine physician concerning the actual clinical application when determining the optimal matrix size of the diagnostic images
Radionuclides with one or more monoenergetic gamma lines (e.g. Technetium-99m, Indium-111 or Iodine-123) are suitable for planar or tomographic single photon imaging. Pulse height analysis is carried out using a multi-channel analyser (MCA) coupled to the detection unit. All impulses that are within the window of acceptance (typically 10-20%) centred on the nuclides photopeak(s) contribute to the final image. Additional energy windows may be applied to correct the proportion of scatter in the final image.
Single photon imaging includes planar acquisition, that represent a projection from the radioactivity distribution on the patient at a fixed angle (e.g. anterior, posterior, lateral, or oblique) or tomographic acquisition that comprises projections along a 360 degrees arc around the field of view (FOV). The angular frequency must be chosen to ensure the sampling theorem; these values depend on the number of detectors, the arch, and matrix. For example, for a 128x128 matrix, 120 steps are sufficient for a 360 degree arc.
Positron Emission Tomography (PET) is an imaging technique characterized by the simultaneous acquisition of multiplane coincidence 511 keV photons arising from a single positron annihilation. These two photons describe a line of response (LOR) which can be traced back to the coordinates of annihilation. The total number of events detected by the coincidence circuit in a PET scanner are referred to as prompt coincidences. These include true, scattered, random, and multiple coincidences. Only true coincidences carry diagnostic information.
A complete PET system consists of a large number of detectors in a ring geometry placed around the subject to be imaged.
PET imaging is always coupled with computed tomography (CT) which allows for optimal image quality both with anatomical landmark characterization and an attenuation map for PET correction. Given the geometry and detection nature of positron annihilation imaging, only tomography imaging is performed during PET examinations.
An image is acquired for a determined acquisition time or preset number of counts. One should choose a determined acquisition time (in case there is radiopharmaceutical uptake without clinical interest) that ensures significant statistics of the uptake of interest. On the other hand, if the count rate does not vary considerably between patients and there are no significant sources causing artefacts, a preset number of counts can be advantageous and would help to reach optimal scanner throughput.
The output of a single field of view (FOV) static acquisition is a square matrix with the predefined matrix size, typically 128x128, 256x256, 512x512, 1024x1024 or 2048x2048 pixels.
Scanners equipped with two or more detectors, can cover the whole patient in the field of view. When diametrically opposed in 180 degree geometry, both detectors scan simultaneously at a preset velocity in cm/minute. Depending on the detectors sensitivity and the isotope used, whole body imaging can take from 20-45 minutes.
The output of a static acquisition of a gamma camera is usually an array of up to 256x1024 pixels.
In positron emission tomography either system with a standard axial field of view up to about 35 cm (SAFOV) or a long axial field of view are available (LAFOV) up to about some of which scalable in steps of about 30-50cm up to about 2m. Both types of PET imaging systems are capable of capturing the whole body (i.e. including head to toes) of the patient. Using SAFOV-PET systems, this is achieved either by axially moving the subject stepwise with an overlap of bed positions (step and shoot) or slow continuous bed motion (CBM) protocols. LAFOV systems either capture the whole body of the patient at once or likewise step and shoot as well as CBM-protocolls are employed.
Figure x: Principle of SAFOF-PET systems
Figure x: Principle of LAFOV-PET systems
More details to these imaging technologies can be found in chapters A-17 and A-18.
The acquisition of short (1-5 sec) sequential static acquisitions is often referred to as dynamic acquisition. This mode enables the display of the distribution of the actual radiopharmaceutical over time. The data is stored in temporal bins that are predefined by the operator. Depending on the specific kinetic process, shorter images may be acquired which then can be prolonged for a slower process still delivering dynamic information.
Dynamic approaches are particularly interesting if it comes to fully quantitative parametric PET imaging, however in routine clinical settings these are too demanding in terms of time for the investigation it takes and the extend of procedures to undertake. These techniques are crucial if kinetic modeling is desired to characterize the pharmacocinetic and -dynamic behavior of new radiotracers in development, usually in preclinical and clinical research settings. More information on dynamic imaging approaches can be found in chapter A-12, however, a detailed description of cinetic modeling for full parametric quantification of PET is not subject of this guide.
In this acquisition mode, each event is individually written to a file along with information about the coordinates of detection (in the case of PET the LOR). This is available for clinical PET imaging but only for research operations in SPECT. This is advantageous for PET dynamic studies, or if images are prone to motion artefacts for some reason (e.g. patient tremor). List mode acquisition provides fast acquisition speeds at the expense of data volume.
Data acquisition can be synchronized to a physiological signal. The most common gated application uses cardiac ECG monitoring and less often breathing cycles. The process of gating acquisition consists in dividing the physiological process into its constituent parts, and the image acquisition is gated to each of those individual parts. The result is a sequence of frames which can be viewed as a motion picture; each frame represents the counts collected at the corresponding points of the cardiac cycle.
The main limitation of gated acquisition is definitely the low count statistics.
The desired output from a nuclear medicine imaging equipment is, in general, an image or set of images that will allow the medical expert to extract some diagnostic information. It is desired for the final images to have a favourable contrast (signal-to-noise ratio) and high spatial resolution, which will be constrained by the optimal equipment design for a certain imaging procedure.
In nuclear medicine, images are constructed following the line integral of a specific tracer distribution on the imaging target in either planar or three-dimensional imaging. Therefore, in theory, areas with an increased accumulation of tracer will show higher signal. The raw signal will be displayed as a matrix of pixel elements that relate linearly to the number of transmission measurements of the imaged subject. This array is known as a projection. A second process, which may be integrated in the reconstruction method or as post processing, involves a mathematic transformation know as filtering in which a specific frequency on the acquired dataset spectrum is enhanced or minimized. Filtering serves the purpose of increasing the diagnostic value of images, but it can sometimes modify the acquired data.
Planar imaging and tomographic imaging differ only in the fact that planar imaging is the representation of one projection at a fixed angle, while tomography aims at a 360-degree volumetric representation of the imaging target by using sufficient angular sampling. Image reconstruction is in this sense aiming to resolve the inverse problem of the acquisition, in which the final product will be a three-dimensional representation, structurally (dimensions/resolution) and quantitatively (activity concentration) as close as possible to reality (the input imaging subject).