European Nuclear Medicine Guide
Up to now, we saw the multiple advantages of combining molecular imaging with the superior high resolution structural imaging provided by CT. However, since CT also has its limitations in imaging and differentiating between soft tissues, soon after the advent of PET/CT in the clinical arena, it was the aim of technical developments to investigate on the combination of PET and MRI. The combination of those two modalities opened the field of cross-modality imaging to the marriage of the well-known functional PET imaging with the superior soft tissue signal that MRI is able to add. PET and MRI have been in the clinical molecular tomographic imaging arena more than three decades each. While MRI originally aimed at imaging soft tissue structures, PET primarily aimed at imaging physiologic and pathophysiologic processes to image oncologic, neurologic, cardiologic, infectious and inflammatory diseases. After the proof of the operability of specific light detectors such as avalanche photo diodes (APD) and silicon photo multipliers (SiPM) under the conditions of a strong static and changing gradient magnetic fields, the way of combining PET and MRI to hybrid systems was paved. First, PET-detector inserts for preclinical[163–169] and clinical MRI-systems[170,171] were developed to further prove the feasibility of combining the two imaging modalities without intolerably impairing the qualitative and quantitative imaging results of the other. In parallel, by using systems employing one and the same patient handling system/table/bed, the clinical utility of having the image data of the involved modalities inherently co-registered to each other was thoroughly investigated[172–175]. Using existing imaging systems, GE introduced a two-room-setup, a clinical PET/CT- and a MRI-System were installed in rooms right next to each other and a unique patient handling system designed connectable to the patient ports/-bores of both of the systems[176,177]. PHILIPS came up with a comparable design but having the two systems in the same room and using a bi-planar rotating patient table[178,179]. The first system fully integrating the avalanche photodiode (APD) based PET detector system was manufactured by SIEMENS Healthineers and reached market maturity in 2011. Offering the option of a true simultaneous whole body measurement of PET- and MR-signals, the limited timing resolution of the APDs, however, did not allow for time of flight (TOF) PET measurements as well known and utilized in photomultiplier based PET-detector systems in hybrid PET/CT systems[180,181]. In 2014, GE healthcare introduced a TOF-PET-capable fully integrated whole body PET/MRI-system employing SiPMs as light detectors in the PET-detector system. In a comprehensive overview, Quick and Boellaard summarized the performance parameters of the three PET/MRI-systems available for clinical use[182]. Recently, SiPM based PET detector systems are also used in digital PET detector systems that are employed in hybrid PET/CT to further increase timing resolution and reduce the influence of noise by analogue-digital conversion. “Digital” in this context means, that each photon of visible light, generated in the scintillator material (-crystal) is detected by the SiPM and directly converted into a binary output signal (i.e., output=1 → visible light photon detected, output=0 → no light photon detected). This is enabled by the dimensions of the microcells of the SiPM. So, a considerable number of SiPM-microcells read out the scintillation light of one particular detector crystal. Now each individual crystal can be read out individually, and this eliminates the need for localization arrays, -electronics, and arithmetic within a detector block as was widely implemented in state-of-the-art clinical PET/CT systems. The energy discrimination – i.e., determining the energy of the gamma ray causing the light pulse in the crystal material – is achieved by determining the intensity of the scintillation light. As the number of light photons in a pulse represents the light intensity, counting the number of detected photons of visible light – i.e., “fired” SiPM cells with output=1 – enables the determination of the energy of the gamma ray. As there is no analogue to digital signal conversion involved in this process, and a photon of visible light leads to a binary output of a SiPM cell, this method can be called digital photon counting (DPC). More technical details and performance characteristics can be found in the articles by Seifert[183] or Lecomte[184].
For the quantitative validity of the PET measurements, it is essential that the concentrations of activity in respective lesions, volumes, and sub-volumes (Bq/ccm) are determined as accurate as possible. Therefore, the energy degradation of the 511keV photons, as they travel through tissue until they reach the detector system, needs to be involved in the reconstruction of the emission data set. This attenuation and scatter of photons are mainly determined by the electron density of the material through which they travel through and interact. This electron density can be directly obtained by obtaining transmission measurements either of 511keV radioactive sources or the X-rays as emitted by the tubes of CT systems. The reconstructed transmission volume data sets can be used directly if obtained from 511keV transmission measurements or after a (bi-linear) calibration/scaling of the linear attenuation coefficients (LAC) at X-ray energies to those at 511keV, the energy of the photons as emitted after annihilation of a positron with an electron. In contrast, a MR-signal represents the proton density in a tissue[185–187]. This is by no means representative for the attenuation behaviour of material against gamma radiation. Thus, several approaches have been established to identify the different tissue classes using the MR signal. These tissue classes are fat, soft tissue, air, and lung. Depending on imaging circumstances (i.e., the imaged body part, position of the patient in the system, patient comfort, preparation, etc.) the segmentation of these tissue classes succeeds – or not. The segmented tissue classes are assigned a constant linear attenuation coefficient, and the so constituted segmented µ-map is used for attenuation correction of the emission data. Still, these algorithms suffer from being not sufficient enough to detect bone and air. In their recent implementations, most vendors use ultrashort- and zero echo time MR sequences to detect bone and, thus, improve the performance of the tissue class segmentation. These methods are combined with methods of µ-map generation from MR data that use structural (i.e., T1- or T2 weighed) MR data sets in combination with CT-atlas based information of a particular part of the body to generate a more realistic map of linear attenuation coefficients including bone[188–191]. As anatomical atlases are the basis for the assignment of LACs, those methods start to fail if the anatomy of the particular subject under investigation differs from the circumstances in the atlas. Recently, Ladefoged presented a method called RESOLUTE[192]. The method was extensively validated for the head and neck region and proved to be very stable in comparison with the CT as a ground truth[193]. It’s accurate performance could also be shown under more difficult anatomic circumstances and alterations as might be present after surgical interventions[194]. In recent research settings, neuronal network approaches are employed to train algorithms using CT and 511keV data to learn generating continues valued maps of LACs on the basis of structural MR data sets. The newest of this kind are utilizing convolutional neuronal network based deep learning algorithms[195–198].A more indepth view about what artificial intelligence is basically about and what are the potential benefits in the area of Nuclear Medicine will be given in the chapter. Using the aforementioned methods and depending on the body part under investigation, the accuracy of the PET measurement in hybrid PET/MRI settings reach the order of accuracy of that in PET/CT settings. Moreover, all the hardware in the path of the gamma rays needs to be taken into account, because it also degrades/attenuates the PET signal. The (flexible or rigid) MR signal receiver coils and the patient table are either implemented by CT-measured maps of LACs or designed so that the attenuation of the PET signal by this material is negligible. Most of the harmonization procedures of quantitative PET as known from PET/CT are based on the measurement of known phantom structures filled with watery solutions of radioactivity containing different fillable sub-volumes and, thereby, representing known activity concentrations in volumes of different sizes in an either cold or hot background. Firstly, being constructed mainly of plastic, the structure of those phantoms cannot be sufficiently detected by MRI. Secondly, larger volumes of pure water in the MRI field of view causes major distortions of the MR signal. This topic has been addressed by searching for alternative liquids to fill the phantom[199]. Current approaches to use activity fillable phantoms in hybrid PET/MRI, however, employ the implementation of CT-generated µ-maps of the particular phantom to account for the attenuation of the PET signal. Thus, inter-system quantitative comparisons give the comparability of just the quantitative performance of the PET detector system. If the clinical settings for attenuation correction – i.e., the MR-based µ-map – is used for attenuation correction of phantom measurements considerable deviations of accuracy of the PET measurement is found[200,201].
The latest generation of hybrid PET/MRI systems is capable of Time Of Flight (TOF) PET signal detection[202]. This information can be used for simultaneous reconstruction of Activity and Attenuation[203].
Depending on the legislation and organizational requirements and settings, PET/MRI systems should usually be installed in and managed by nuclear medicine, radiology, or a combination of the both departments. If it is integrated in a full hybrid imaging molecular imaging PET centre including a radiopharmaceutical production site with a cyclotron on site, the system should be located in close proximity to this facility. However, satellite PET/MRI units can also be installed at specialized clinical sites such as paediatric, psychiatric, or neurology departments if the aforementioned staffing level is available.