Hybrid multimodality systems

Functional imaging of the biodistribution of radiopharmaceuticals is, by its own nature, limited in supplying detailed anatomical information, being aimed to reproduce biochemical function of tissues rather than some anatomy related descriptor, such as tissues composition or density. Moreover, accurate anatomic localization of functional abnormalities imaged by emission scans, is hindered by the limited spatial resolution achievable in SPECT and also, even if at a different extent in PET.

For most of the molecular tracers used, some anatomic information can be inferred from non-specific uptake in muscles, brain, heart, liver, colon and other organs, or from excretion through the urinary system. Even if localization relative to such low resolution anatomic landmarks may help image interpretation, a detailed anatomic framework such as that provided by CT represents clearly a major improvement.

Hybrid multimodality imaging is one of the most rapidly growing imaging modalities. The combination of nuclear medicine imaging (SPECT or PET) with CT is considered to be an evolution in imaging technology where fusion of two established modalities offers more than the sum of the parts. Both modalities have their strength: CT scanners image anatomy with high spatial resolution;

nuclear medicine imaging (SPECT or PET) provides the metabolic and functional information. Modern hybrid imaging modalities have the ability to provide, in a single imaging session, detailed anatomical and metabolic and functional information. PET–CT has revolutionized the care of cancer patients in developed countries and is increasingly being adopted in emerging economies. Similarly, the use of SPECT–CT is rapidly increasing and probably this hybrid imaging technology will become the gold standard for conventional nuclear medicine.

In addition to this substantial clinical benefit of registering anatomical and functional images, the coupling of CT with SPECT and PET systems provides additional technical benefits, by enabling CT to be used for attenuation correction, thus reducing the duration of the procedure, reducing motion artefacts, increasing quality of the corrections and, as a consequence, making possible accurate quantitative imaging.

4.3.1. The problem of attenuation correction

Self-absorption of photons emitted within the body of a patient administered with a radiopharmaceutical is a well known effect that significantly degrade both the quality of reconstructed images, and the capacity of collecting quantitative information, in terms of absolute quantification (Bq/g of tissue) or of semiquantitative indexes, such as the standardized uptake value (SUV).

Owing to differences in the principle of detection, specific issues appears in SPECT and in PET in developing strategies for attenuation correction. Despite these differences, a common approach can be identified since in both modalities to properly take into account self-absorption and correct for its effect, an accurate description of the distribution of the attenuation coefficient of radiation (strictly correlated with the distribution of densities of tissues) is necessary for the imaged volume. Techniques for attenuation correction were then developed in order to assume or to measure the distribution of the linear attenuation coefficient μ in imaged tissues.

In the first category is the Chang method, widely used in SPECT, or similar geometrical method used in PET in which, once a description of the shape of the imaged volume is defined, a uniform value for the effective μ is assigned to each voxel in the volume. While these techniques are computationally efficient, they cannot reach a good level of accuracy, since they do not take into account differences in tissues density. In particular, these techniques fail when marked variations in densities are present, like in the case of highly inhomogeneous districts (thorax) or when bone is involved (lack of proper characterization of in-homogeneity within bones, age related variations in bone density).

This has led to the first attempts to acquire patient specific information of the density distribution, that were based on simultaneous transmission acquisition during simultaneous emission studies, making use of radionuclide sources (e.g. ¹⁵³Gd sources emitting at 100 keV in the case of SPECT imaging and rotating ⁶⁸Ge/⁶⁸Ga sources, emitting 511 keV annihilation photons, in PET).

While these methods can allow for sufficiently accurate collection of tissues density distribution, they typically require prolonged acquisition time and increased running costs for acquiring calibrated transmission sources, their management and disposal, once their operational life is terminated due to physical decay. This complex series of problems was the initial motivation for exploring the use of multimodality systems, including a CT scanner coupled to an emission camera, with the aim of acquiring high spatial resolution and low noised images of the density distribution within the imaged volume.

4.3.2. SPECT–CT

A multimodality system, integrating a SPECT scanner with a CT system, in which the patient can be positioned in the same bed that automatically moves in the two imaging positions with a simple translation, provides a huge advance in technology. The two imaging datasets can be acquired in a close sequence, so that they can be practically considered simultaneous, and a simple rigid re-alignment allows for the registration of corresponding slices. The CT data can then be used both to correct for tissue attenuation in the SPECT scans and to display metabolic data on an accurate morphological context, allowing for efficient, simultaneous

‘navigation’ in function and anatomy.

To accomplish this, since the two datasets are acquired at different level of spatial resolution, resampling is necessary in order to match voxel sizes. In early systems, this was typically made by degrading the spatial resolution of SPECT images, while in more modern systems, CT images are used as the basis to be superimposed with properly interpolated and scaled SPECT data.

Early SPECT–CT designs coupled SPECT with low performance CT, based on low power X ray tubes and slow rotation speeds, where the CT component was by no means optimized for diagnostic quality imaging. The aim was focused in providing attenuation correction and a basic but satisfactory anatomical context for the SPECT, while saving the running costs of attenuation correction radionuclide sources and maintaining a relatively low cost of the CT component.

The success of these systems, the increased experience in proper clinical use of the information gathered by both modalities and the parallel experience with PET–CT, has led to the introduction of SPECT–CT systems that incorporate a high performance CT component with capabilities comparable to dedicated CT scanners. With this development, SPECT–CT now benefits from substantially improved CT image quality, faster data acquisition and a broader range of CT protocols. A full range of different multidetector CT slice configurations are now available, as well as alternative designs including those based upon flat panel detectors, reflecting the vitality of this technique and its continued relevance in clinical practice.

4.3.3. PET–CT

PET imaging is affected by several physical effects, like scattered and random coincidences, photon attenuation, detector efficiency variations, spatial resolution non-uniformity, scanner dead time and others. Of these, the most important is by far photon attenuation within the body of the patient that influences image quality as well as quantitative accuracy. The CT component of a PET–CT system allows for an accurate attenuation correction, with a marked improvement

in both image quality and quantification. This is per se an extraordinary result that could justify its use. However, the primary purpose of multimodality imaging is the precise, high spatial resolution anatomical localization of regions identified on the PET tracer uptake images.

CT and PET have been for long time used sequentially in the diagnosis and staging of disease and in monitoring the effects of therapy, with PET acquiring an increasing role, particularly when the CT scan was equivocal. For years, visual comparison of the separate anatomic and functional image sets has been the standard approach adopted to synthesize additional information, eventually using, where appropriate, software to fuse and align the two sets of images.

This situation changed dramatically with the introduction of the multimodality PET–CT scanners. These devices solve the problem of image registration and fused display through hardware rather than software, providing the capability to acquire accurately aligned anatomic and functional images for a patient, within a single scanning session. Since the patient remains positioned on the same bed for both imaging modalities, temporal and spatial differences between the two sets of images can be neglected.

Initially, PET–CT systems were a PET scanner and a CT scanner combined together, either under a common gantry cover or effectively separate, with different acquisition systems and whose results were combined in one or the other workstation. Instead, modern PET–CTs are designed and engineered as an integrated system (see Fig. 12), controlled by the same common platform, including a common patient database containing both PET and CT data. They include all the technological improvements achieved in both methodologies.

New PET scintillation crystals, together with innovative iterative reconstruction algorithms, make possible an increased spatial resolution and count sensitivity, as well as exploiting TOF information. The CT component uses multidetectors, making possible very fast, spiral acquisition at 4, 16, 64 and even more slices, not differently from stand alone CT scanners. The fusion and display software included in the last generations of workstations allows for an easy, intuitive navigation in the data sets and immediate measurement of SUV in the areas of interest. High quality fused images have supported the extraordinary diffusion of PET and particularly its widespread acceptance in oncology.

The possibility of acquisition of gated PET scans, standard feature of all new systems when equipped with proper physiological signals generators (e.g. respiratory gating triggers), can reduce image blurring due to physiological motion and has particular relevance in oncological studies, as a support for accurate treatment planning, and in nuclear cardiology studies.

The high end CT component makes modern PET–CT scanners indicated for all types of study, including applications in nuclear cardiology, given the extended coverage of multislices detectors. However, for the majority of

oncological studies, a full diagnostic CT scan is not necessary: in this case, low dose protocols with limited current at the X ray tube are adopted, allowing also an extended tube life.

In any case, the fast acquisition of the CT component in a multimodality scan enables to obtain in a short time transmission data for attenuation correction with high statistics, significantly decreasing the total time for completion of a scan needed with ‘classical’ PET scanners using ⁶⁸Ge transmission sources, thus optimizing the throughput of patients and the usage of the equipment.

4.3.4. PET–MR

A multimodality imaging system combining PET and MR makes possible the in vivo assessment of biochemical processes while granting anatomical information with excellent soft tissues contrast. However, integrating PET and MRI is a complex technological task. First of all, photomultiplier tubes, typically adopted by the majority of PET detectors, cannot operate within a strong magnetic field. For this reason, the first clinical PET–MR scanners were based on separate PET–CT and MR scanner, installed at a relatively close distance and using a special ‘shuttle’ bed, to move the patient from one equipment to the other.

Alternative photodetectors have then been introduced, like avalanche photodiodes and, more recently, silicon photomultipliers. These can operate in the

FIG. 12. PET–CT scanner (courtesy of M. Marengo).

presence of a magnetic field, in particular, fast response silicon photomultipliers have been used in TOF capable PET–MR scanners. All this made it possible to integrate the two systems to allow for simultaneous acquisition, thus preventing the small differences of time and positioning unavoidable in PET–CT. The achievements in this field have led to the development of PET detectors based of fast scintillating crystals associated with highly efficient digital photodetectors that are now adopted by the last generation of PET–CT scanners.

Compared to PET–CT, the use of MR information in the acquisition of multimodality studies presents specific differences and raises new issues. As regards attenuation correction, the basic approach in PET–MR is to segment the images from different MR sequences and classify data into a limited number of component (air, lung, fat, soft tissue, bone), each characterized by a uniform attenuation value in order to obtain an attenuation map. This approach has proven to be sufficiently accurate, even if artefact can still arise in specific cases, like very large patients or metallic prostheses.

With regard to image reconstruction of emission data, PET–MRI systems use the most advanced iterative reconstruction algorithms developed for PET–CT. These include accurate modelling of the system response, improving noise characteristics and spatial resolution of the images. However, considering quantitative aspects, this can result in SUV values overestimated compared to standard PET–CT results (see Refs [21–30]).