Key words
PET/MRI - oncology - consensus recommendations - whole-body imaging - hybrid imaging
Introduction
The aim of this consensus recommendation is to provide guidance to healthcare experts
and physicians regarding clinical indications, execution and interpretation of [18F]-Fluorodeoxyglucose
(FDG) Positron emission tomography/magnetic resonance imaging examinations ([18F]-FDG
PET/MRI) for whole-body staging in oncology [1].
PET is a noninvasive imaging technique that provides quantitative information on 3-dimensional
distributions of radioactively labelled biomolecules (tracer) in tissues. [18F]-FDG
is a tracer composed of radiolabeled glucose, which is the most common tracer for
oncology imaging indications [2]. For the majority of tumors, malignant cells display activated glycolytic pathways
resulting in increased glucose utilization via upregulation of glucose transporter
expression and hexokinase activity [3]
[4]. Thus, more of the glucose analog, [18F]-FDG, is taken up in metabolically active
cancerous cells than in surrounding healthy tissues. Therefore, [18F]-FDG-PET has
been demonstrated to be a sensitive method and well-established imaging modality for
detection, re-/staging as well as for the evaluation of therapy response of solid
tumors [5]
[6].
Magnetic resonance imaging (MRI) is a noninvasive imaging technique that provides
anatomical 3 D visualization of tissues with high spatial resolution based on relative
differences in resonance frequencies of spins following external excitation [7]. In addition, MRI employs multiple imaging sequences and associated soft-tissue
contrasts that yield noninvasive insight into functional and cellular aspects of tissues
and organs [8]. The magnetic field-based excitation and resonance measurement method sets MRI apart
from computed tomography (CT), which is a pure transmission method based on the attenuation
of ionizing radiation. In contrast to CT-based transmission imaging, MRI does not
employ ionizing radiation. Thus, the exposure of patients undergoing PET/MRI to ionizing
radiation originates from the PET portion only and therefore is significantly lower
compared to PET/CT [9].
While attenuation correction is a well-established aspect of PET/CT imaging, it was
a methodologically challenging task to overcome in integrated PET/MRI (please also
refer to the section “Attenuation correction”). Thus, the introduction of MR-compatible
PET detector systems provided the basis for the hardware integration of PET and MRI
components into a single, integrated system [10]
[11]. Prior work of developers of small-animal imaging systems [12] has helped to replace the photomultipliers in PET detectors with semiconductor-based
diodes that are capable of amplifying the scintillation signal in the scintillator
crystals without being affected by the magnetic field [13]. Following further technical and methodological work, fully integrated PET/MRI systems
have been introduced for clinical use with a magnetic field strength of 3 T [14] and MRI sequences were developed that enable a reliable correction of attenuation
artifacts in PET with comparable quality to CT transmission maps for PET/CT imaging
(please also refer to section “Attenuation correction”).
At present, the number of clinical studies with PET/MRI is continuously increasing.
Recent publications comparing the diagnostic accuracy of whole-body PET/MRI demonstrated
equivalence to that of PET/CT (using the same tracers) [15]
[16]. However, a number of potential benefits for PET/MRI have been highlighted with
regards to the high soft-tissue contrast of MRI and consecutively improved delineation
of tumorous lesions [17]
[18]
[19]
[20]. Such studies will benefit from overcoming of the existing variations in the use
of PET/MRI for distinct diagnostic questions [21], and, therefore, consensus recommendations are mandatory to limit these a priori
variations through harmonization and standardization approaches.
Definitions
Similar to the endeavors to establish consensus recommendations for combined PET/CT
imaging protocols and definitions [6]
[22]
[23], we will use the following definitions for combined PET/MRI for easier understanding:
-
A combined PET/MRI system is an integrated PET and MRI system that enables the generation
of PET and MRI data during the same patient acquisition without the need to reposition
the patient between examinations.
-
Fully integrated PET/MRI refers to a hardware combination of both imaging systems
that permits the simultaneous acquisition of PET and MRI data, requiring the use of
MRI-compatible PET detectors.
-
PET/MRI and MR/PET can be used interchangeably. The same is true for PET-MRI and MRI-PET.
-
The information contained in the images from a PET/MRI examination is given by the
tracer-of-choice, the method of acquisition of the emission data (static or dynamic
mode) and the mode of the MR acquisition (T1-weighting, T2-weighting, dynamic contrast-enhanced
sequences, proton density, diffusion-weighted imaging, arterial spin labelling, apparent
diffusion coefficient, etc.).
-
In the clinical routine, PET/MRI examinations do not include a transmission measurement,
and, therefore, alternative means have to be provided to derive attenuation correction
factors (ACF) for the PET data in order to quantify the molecular signals [24].
-
Artifacts comprise all types of PET and MR image distortions that include visually
perceived deviations from typical representations of anatomy and function that may
or may not cause a quantitative bias (e. g., lesion size, tracer concentration, etc.).
These distortions are likely not to arise from a disease process but from methodological
pitfalls or system malfunctions [25].
Indications for/application fields of PET/MRI
The following indications/application fields apply to whole-body [18F]-FDG-PET/MRI
examinations in oncologic imaging.
-
Staging/restaging of known tumors
-
Detection/exclusion of tumor relapse
-
Therapy monitoring
-
Detection of a cancer of unknown primary (CUP)
-
Further differentiation of indeterminate findings in conventional imaging
-
Radiation therapy or biopsy planning
PET/MRI examination
Necessary patient information
-
In preparation for [18F]-FDG-PET/MRI examinations, the following information should
be collected from the patient:
-
History focused on the type and location of the malignant disease,
-
Date of the initial diagnosis,
-
Type of diagnostic confirmation,
-
Treatment prior to the current PET/MRI scan (e. g. biopsy date and results, histology,
surgery, radiotherapy, chemotherapy),
-
Medication at the time of examination, and
-
Any prior examination (particularly imaging studies)
-
History of diabetes mellitus, last food intake, infections or recent colds
-
Ability of the patient to lie still for the duration of the scan (30 – 60 min)
-
Claustrophobia: Ability of the patient to remain in the PET/MRI system for the duration
of the examination
-
Ability to provide informed consent
Patient preparation
The main objectives of patient preparation are the reduction of tracer uptake in normal
tissue (e. g. heart, skeletal muscle) while preserving tracer uptake in the target
structures (tumor tissue). The following is a general-use protocol:
MRI safety
The following points relevant to MRI safety in PET/MRI are to be considered:
-
For patient safety, all patients should be routinely checked and screened with standardized
checklists for potential MR contraindications (e. g. pregnancy, previous contrast
agent reactions, catheters, ports, metallic implants, vascular stents, active implants,
cardiac pacemakers, etc.) [30].
-
All metal objects (e. g. dental prostheses, clothing with zippers and buttons) should
be removed from the patient and cotton clothing without metal should be provided to
the patient.
-
Regarding implants, the specific kind of implant, its location, and its material need
to be investigated beforehand. Information about the MR compatibility and safety of
an implant can be assessed from the implant pass and/or directly from the implant/device
manufacturer (e. g. online sources). The following safety regulations apply and should
be adhered to: “MR unsafe” – absolute contraindication; “MR conditional” – relative
contraindication, conditions apply; “MR safe” – no contraindication. In case of “MR
conditional” implants, all conditions (e. g. max. field strength, SAR limitations,
etc.) as provided by the implant manufacturer and online sources must be reviewed
and applied during the MRI (PET/MRI) examination. In case of “MR unsafe” implants,
the indication for the PET/MRI examination needs to be scrutinized and other imaging
options should be considered.
-
Beyond safety concerns, implants may cause artifacts, large-volume signal voids and
geometric distortions in MR imaging. This may hamper image interpretation.
Attenuation correction
In contrast to CT-based attenuation correction (AC) in PET/CT [31], the attenuation properties of tissue cannot be derived directly from complementary
MR images. Therefore, different concepts of MR-based attenuation correction have been
introduced [24]. The most commonly applied method is based on a two-point Dixon technique, which
facilitates a 4-compartment-model attenuation map (μ-map) to identify air, lung tissue,
fat, and soft tissue [32]
[33]
[34]. Based on this segmentation of MR images into distinct tissue classes, the individual
compartments are assigned a predefined linear attenuation coefficient (LAC) for the
corresponding tissue [33]
[35]
[36].
A number of challenges including the systematic underestimation of PET quantification
related to standard MR-based attenuation correction have been reported, the most prominent
being the lack of consideration of bone tissue and the occurrence of truncation artifacts
[36]
[37] (for further information please refer to the section “Artifacts”). Different compensation
approaches for brain and whole-body imaging have been proposed to account for the
misclassification of bone tissue as soft tissue [38]
[39]. Promising results for whole-body imaging were shown when utilizing a CT-based 3-dimensional
bone-model of major bones as an adjunct to MR-based AC data [34]
[40]
[41]
[42].
Artifacts
Following the introduction of integrated PET/MRI systems, a number of artifacts have
been reported that are related to PET-only, MRI-only or integrated PET/MRI acquisitions.
A selection of the most common artifacts and potential solutions is discussed in the
following paragraph [25].
The most evident artifacts have been shown to be related to MR-based attenuation correction,
causing a systematic underestimation of PET quantification when compared to PET/CT
[43]
[44]. Apart from the misclassification of bone tissue (please refer to the section “Attenuation
correction”), truncation artifacts are a major concern in integrated PET/MRI. Due
to the limited transaxial and lateral field of view (FOV) in MR imaging to a spherical
diameter of about 50 cm, structures beyond these dimensions show geometric distortions
and signal voids, resulting in truncation artifacts alongside the patient arms and
incorrect PET quantification [25]
[45]. In addition to the PET-based MLAA algorithm (maximum likelihood estimation of attenuation
and activity) deriving the patients outer body contours from PET data [46]
[47], a novel purely MR-based truncation correction method was introduced by Blumhagen
et al. [48]
[49]. This method, also referred to as HUGE (B0 homogenization using gradient enhancement),
enlarges the lateral FOV in MR imaging beyond the conventional 50 cm diameter, effectively
eliminating truncation artifacts along the patients arms in MR-based attenuation correction
[48]
[50].
Involuntary patient and organ motion causing a misalignment of emission and attenuation
data is a known challenge in PET/CT imaging that may be further enhanced in PET/MR
imaging due to prolonged examination times. Unlike in PET/CT and owing to simultaneous
PET and MR data acquisition, PET/MRI has potential for MR-based motion correction
of PET data. Different methods for motion correction have been proposed to account
e. g. for respiratory motion artifacts including real-time MR imaging and 4 D MR data
of breathing motion or free-breathing MR imaging to retrospectively perform motion
correction [51]
[52]
[53].
The following points relevant to MR-based attenuation correction and artifacts in
PET/MRI are to be considered:
-
In PET/MRI, AC is based on MR imaging. Thus, artifacts in MR-AC have a direct effect
on PET quantification. Consequently, MR-based AC needs to be accurate and free of
artifacts to provide precise PET quantification. MR-AC images shall be routinely checked
for artifacts, consistency and plausibility during PET/MR image reading. Typical artifacts
are mis-segmentation of air/soft tissue/fat/bone and metal artifacts due to dental
prostheses and due to metallic implants such as stents and surgical clips, etc. Artifacts
may be displayed as signal voids, exceeding the true dimensions of metal inclusions.
Thus, artifacts are mostly easily detectable in MR-AC, indicating regions of potentially
inaccurate PET quantification [45]
[54].
-
While new features for the improvement of MR-AC are constantly developed and implemented
into the commercial software of available PET/MRI systems, including high-resolution
Dixon AC, ultrashort echo time (UTE), zero TE (ZTE) sequences and/or bone models for
bone detection in PET/MRI attenuation correction [17]
[34]
[40]
[41], users need to remain attentive to MR-AC related limitations and artifacts in SUV
quantification.
-
Truncation artifacts along patient arms in MR-AC may affect PET quantification. The
standard method on all available PET/MRI systems for truncation correction is the
PET-based MLAA algorithm [46]. A more recent method for improved MR-based truncation correction in MR-AC is HUGE
[41]
[48]
[50].
-
Only radiofrequency (RF) coils that are labelled for combined PET/MRI use should be
used. Using standard RF coils that are labelled for MR-only use will not be considered
in PET/MRI AC and may, thus, lead to inaccurate PET quantification and artifacts in
PET [32]
[55].
Quality control
Quality control of PET tracers is governed by the “Draft Guidelines for Radiopharmacy”
[56]. Quality control and application recommendations for MR contrast agents are addressed
in the guidelines of the European Society of Urogenital Radiology [26]. Quality control procedures for the PET and MRI subsystems should be set up in accordance
with the published guidelines [57]
[58] but shall at least follow the vendor’s recommendations. In addition, proper cross-calibration
of the PET system with the respective dose calibrator has to be ensured. In routine
operation, daily quality control scans (using a dedicated phantom) shall be conducted
prior to patient scans to ensure correct PET acquisition and quantification.
Imaging workflows
Imaging workflows may vary with the clinical indication. Similar to PET/CT imaging
in oncology, PET/MRI can be performed in whole-body mode, meaning that patients are
scanned over multiple, consecutive bed positions to cover larger co-axial imaging
ranges. Given the extensive variability of MR imaging protocols and the choice of
MR sequences, whole-body PET/MRI examinations have been shown to take longer than
PET/CT examinations of the same co-axial imaging range. Therefore, the need for optimized
and standardized PET/MR imaging workflows has become widely recognized. Over the past
years, a number of proposals have been published [59]
[60]. This document sets out to describe suitable imaging conditions and protocols for
whole-body [18F]-FDG-PET/MRI of oncology patients. Of note, specific protocols and
MR sequences are subject to change depending on the user, vendor and indication for
the examination.
For reasons of simplification and conformity to PET/CT imaging, all workflows mentioned
below apply to whole-body coverage from skull-base to mid-thighs. This coverage is
usually achieved within four to five bed positions (BP) depending on the patient height.
Accordingly, a combination of dedicated (attenuation-corrected) radiofrequency (RF)
head and neck coils and a varying number of phased-array body surface RF coils are
utilized as needed [32]. Imaging is commonly performed in a supine position starting from mid-thigh to skull-base
to ensure minimal impairment of lesions in the vicinity of the bladder due to increased
[18F]-FDG activity in the bladder.
In a first step MRI localizers are acquired to define the axial range for the examination.
Pre-scanning of the shimming and adjustment of the magnetic field are followed by
the attenuation correction (AC) sequence for every BP (for detailed information regarding
MR-based AC please refer to section “Attenuation correction”).
Workflow 1: Ultra-fast PET/MRI
This workflow is based on a 2-min/BP acquisition that facilitates ultra-fast “PET/CT-like”
whole-body staging within a total time of just under 20 min [61]. The reasoning for this specific algorithm is to facilitate ultra-fast whole-body
staging, e. g., in patients with low compliance (e. g. elderly, pediatric) or as a
whole-body coverage adjunct to local dedicated imaging (e. g., local dedicated tumor
staging in head and neck cancer or soft tissue sarcoma + whole-body ultra-fast).
Indications for this ultra-fast workflow include whole-body staging, e. g., for lymphoma
or staging and exclusion of relapse of tumors.
Potential sequences to be obtained within the 2-min PET include: (1) Fast T2-weighted
spin echo sequence (e. g. HASTE) and (2) non-enhanced fast fat-saturated T1-weighted
gradient echo sequence (e. g. VIBE). Contrast media injection and acquisition of post-contrast
fast T1-weighted fat-saturated imaging may be performed subsequent to the non-enhanced
sequences. In case of primary tumors (e. g. malignant melanoma, neuroendocrine tumors)
known to cause hyperarterialized metastases of the parenchymatous organs, additional
dynamic contrast-enhanced imaging of the upper abdomen (e. g. VIBE) can be added.
The combination of the sequences above enables the combined assessment of T2, non-enhanced
and contrast-enhanced features of potential lesions ([Fig. 1]).
Fig. 1 Schematic workflow for the ultra-fast whole-body [18F]-FDG-PET/MRI.
Workflow 2: Fast PET/MRI
This workflow is based on a 4-min/bed acquisition that comprises diffusion-weighted
imaging (DWI) in addition to the above-mentioned sequences listed in the ultra-fast
algorithm ([Fig. 2]) [62]
[63]. The additional diffusion-weighted sequence offers complementary tissue information
to PET and may be applied as a “search” sequence as it is considered useful particularly
in the detection of small lesions, e. g., liver metastases that may be too small to
be picked up by PET. Together with potential post-contrast T1w gradient echo sequences,
this “fast PET/MRI” algorithm should require less than 30 min depending on the total
amount of BP and duration of shimming, etc. ([Fig. 3]).
Fig. 2 Imaging example of a 45-y/o patient with a celiac lymph node metastasis (white arrows)
imaged in ultra-fast and fast [18F]-FDG-PET/MRI.
Fig. 3 Schematic workflow for fast whole-body [18F]-FDG-PET/MRI.
Workflow 3: Dedicated local and whole-body PET/MRI
This workflow comprises dedicated local PET/MRI of the tumor region (e. g., head and
neck, cervical cancer, soft tissue sarcoma of the limbs) and fast sequences for whole-body
coverage. The aim of this workflow is to facilitate a dedicated workup of the primary
cancer and whole-body staging in one examination. The MR protocol for the dedicated
local PET/MRI scan should be selected in accordance with the primary tumor and guideline
recommendations (e. g., cervical cancer [64]). Whole-body imaging can be performed utilizing the above-named ultra-fast or fast
algorithm depending on patient compliance, potential benefit derived from DWI and
desired length of the examination ([Fig. 4], [5]).
Fig. 4 Schematic workflow of dedicated local [18F]-FDG-PET/MRI + whole-body staging.
Fig. 5 Imaging example of a [18F]-FDG-PET/MRI scan of a 52-y/o patient with a large soft
tissue sarcoma of the left lower limb (thick arrows). The figure displays the dedicated
local PET/MRI protocol a for assessment of the primary tumor and the fast protocol for whole-body staging
b, revealing an iliac lymph node metastasis in the left hemipelvis.
Reading and reporting
The following recommendations on reading and reporting are intended to serve as assistance
to novice PET/MRI readers and help standardization. High quality reading and reporting
of PET/MRI examinations is based on expert knowledge of PET and MRI imaging [65]. Hence, PET/MRI reading should be performed jointly by a radiologist and a nuclear
medicine physician or by adequately trained dual-certified physicians (nuclear medicine
and radiology).
It is important to evaluate the “raw” MRI and PET data as well as fused imaging. In
contrast to rather distinct differences in the required expertise and duration of
reading MRI versus CT, PET/MRI reporting can be conducted similar to PET/CT reporting.
After reporting of the definition of the exam, clinical information, examination procedure
and key parameters (including the applied radioactivity, uptake time and amount of
contrast agent), the actual report, in terms of imaging findings and their evaluation,
can be written as an integrated (conjoint description and evaluation of findings in
MRI and PET) or separate report (subsequent description of findings in MRI and PET
and conjoint evaluation).
Conclusion
Since its introduction in 2010, whole-body PET/MRI has become well-established in
scientific and clinical imaging. Still, a number of basic, methodological and professional
challenges have limited its wider general acceptance in the oncologic community as
well as its utilization as a diagnostic alternative to PET/CT. The greatest obstacle
is caused by extensive and heterogenous protocols that have rendered PET/MRI a research
tool that is incompatible with clinical use and is economically challenging. Thus,
we introduced recommendations on workflow options that offer efficient and fast imaging
protocols open for adaptation to meet the purpose of the examination. The three categories
of imaging protocols above allow the standardization and harmonization of PET/MRI,
which is a prerequisite for multi-center trials and the assessment of large patient
cohorts. This may support the future adoption of PET/MRI in clinical routine imaging
and institute reimbursement.