Patient-centric performance and interpretation of positron emission tomography /computed tomography myocardial perfusion imaging: a clinical consensus statement of the European Association of Cardiovascular Imaging of the European Society of Cardiology

Abstract

Positron emission tomography (PET) is the most advanced myocardial perfusion (MPI) technique for the non-invasive assessment of coronary artery disease and its many manifestations, including ischaemia, hibernation, and scar. This comprehensive overview aims to empower clinicians, technicians, and patients with clear, structured knowledge on performing and interpreting PET MPI. This document will describe stress protocols, patient preparation, tracer pharmacodynamics and nuclear properties, camera capabilities, post-acquisition processing, and a comprehensive and clear reporting system for both perfusion and viability imaging.

Introduction

Positron emission tomography (PET) myocardial perfusion imaging (MPI) using radioactive tracers has become a well-established, non-invasive modality for the assessment of coronary artery disease (CAD). Over the course of the last several decades, new tracers, cameras, computerized algorithms, and stress protocols have been introduced and endorsed by various cardiac and imaging societies. Therefore, a novel clinical consensus statement is needed to introduce these topics and highlight European and American procedural differences with a patient-centric educational focus.

We will present a concise and clinically relevant document with visual aids that will cover the performance of PET MPI for patients with suspected or known coronary artery disease. This document sought to integrate knowledge from experts in both Europe and the USA, while incorporating the European Society of Cardiology (ESC) and American College of Cardiology guidelines, as well as best practice guidance from the American Society of Nuclear Cardiology and the European Association of Cardiovascular Imaging of the ESC.^1–3

This document and its figures are meant to be a daily reference for patients and imaging centres to perform these tests, as well as a digestible introduction to new practitioners of PET MPI. In particular, this document aims to be complementary to dense procedural manuals, which are only accessible to experts, as well as larger guidelines on chronic coronary syndromes, which are test agnostic. This document will describe stress protocols, patient preparation, tracer pharmacodynamics and nuclear properties, camera capabilities, post-acquisition processing, and a comprehensive and clear reporting system for both perfusion and viability imaging. We also aimed to enhance a patient-centric approach by using various visual cartoons that can be shown and explained to patients. We have structured this paper to be helpful to a wide range of people, with the recognition that the relevance or comprehension of certain sections may vary depending on level of training. A simplified set of key points has been included in the Supplementary material specifically for patients. Overall, we aim to provide a document that is thorough and accessible for trainees, established providers, and patients.

Stress protocols

Angina can encompass a wide variety of symptoms, ranging from chest, arm, or jaw pain to shortness of breath, nausea, or epigastric discomfort. Angina occurs due to an inadequate supply of blood to the heart, particularly during exercise or stress. Atherosclerosis of the coronary arteries, either in the large epicardial arteries or the microvasculature, is the most common reason for angina. PET MPI is a powerful tool for identifying the location and extent of coronary stenosis. In PET MPI, patients are injected with radioactive tracers which are known to temporarily deposit within myocardial tissue. During stress, either via exercise or a pharmaceutical agent, there will be an imbalanced uptake of the radiotracers in the heart, with less uptake downstream from a significant stenosis. Both qualitatively and quantitatively, clinicians can detect these imbalances of radiotracer to identify areas of heart muscle affected by coronary stenosis.^1,2

Pharmacological stress testing is the standard in PET MPI, as opposed to single photon emission computed tomography (SPECT) where exercise protocols are preferred (Figure 1). Exercise protocols are more challenging in PET MPI due to the short-half-lives of the PET tracers, particularly ⁸²Rb-chloride and ¹⁵O-water which have half-lives of 75 s and 2.1 min, respectively.^5,6 These brief half-lives preclude safely moving a patient from either a treadmill or stationary bicycle to the PET scanner after injection of the tracer at peak exercise. Exercise protocols are more feasible with ¹³N-ammonia and ¹⁸F-flurpidaz due to their longer half-lives. However, exercise protocols typically prevent the measurement of myocardial blood flow (MBF) regardless of tracer, as imaging must be performed at the time of tracer injection to capture first-pass myocardial extraction.

Figure 1.

Methods of stress testing. Stress testing can be performed pharmacologically (Q1, Q2, Q4) or with exercise (Q3). Pharmacological stress testing is more common with PET compared with single photon emission CT, the alternative technology to perform nuclear MPI. Patients undergoing vasodilator stress tests may experience side effects such as flushing, headache, shortness of breath, and chest pain. Similarly, dobutamine may cause palpitations, chest pain, headache, flushing, and shortness of breath. Reproduced with permission from Abadie et al.⁴

Regadenoson, adenosine, and dipyridamole, the available vasodilator stress agents, are adenosine agonists. Areas of myocardium distal to fixed stenoses will have impaired vasodilator response compared with normal vessels, thus causing heterogeneous tracer uptake.

Dobutamine is an adrenergic agonist which increases myocardial oxygen demand via increases in heart rate and contractility. Patients undergoing dobutamine or exercise protocols are expected to achieve 85% of max-predicted heart rate for the test to be considered diagnostic. Atropine, an anticholinergic drug, can be administered if patients have not met max-predicted heart rate despite high doses of dobutamine.⁷

Stress-electrocardiogram (ECG) should be performed with both exercise (treadmill or bicycle) and pharmacologic stress protocols. Stress-ECG may uncover high-risk findings like ischaemic ST changes, such as ST-depressions, or arrhythmia.⁸

Patient preparation

Intake of caffeine, an adenosine antagonist, can interfere with vasodilator agents and therefore caffeine-containing foods, beverages, and medications must be discontinued for at least 12–24 h prior to the exam.⁹ Decaffeinated beverages often contain small amounts of caffeine that may interfere with vasodilation and should also be avoided. Failure to adhere to proper preparation may lead to non-diagnostic studies and/or rescheduling of the exam. Patients are also instructed to take nothing by mouth, except water, for at least 4 h prior to the exam.¹⁰ Common foods, beverages, and medications to avoid can be seen in Figure 2. Methylxanthine-containing medications, such as theophylline, must also be held for five half-lives before a vasodilator exam.

Figure 2.

Examples of medications, foods, and drinks that may interfere with image quality. Some providers may continue beta-blockers or vasodilators during exercise testing to assess the degree of ischaemia while on treatment. Reproduced with permission from Abadie et al.⁴

Referring physicians can consider temporarily discontinuing haemodynamically active drugs, such as calcium-channel blockers, beta-blockers, or vasodilators. For exercise protocols, calcium-channel blockers and beta-blockers may prevent the achievement of maximum predicted heart rate. For vasodilator testing, the decision to hold anti-anginal medications is more controversial. There are mixed data on whether anti-anginal medications, and beta-blockers in particular, attenuate perfusion defects and therefore may reduce the sensitivity of the exam.^11–13 However, discontinuation of chronic medications may result in worsening of underlying tachyarrhythmias and hypertension. The most recent American Society of Nuclear Cardiology procedural guidelines recommend discontinuation of haemodynamically active medications, whereas the European Association of Nuclear Medicine procedural guidelines do not have an explicit recommendation.^7,10

Decisions on haemodynamically active medications should be tailored to the patient and the indication of the exam. For example, referring physicians may wish to hold anti-anginal medications when trying to diagnose ischaemia in patients without established CAD. Conversely, with exercise protocols, referring physicians may continue anti-anginal medications in patients with established CAD to assess their effects on the relief of symptoms, the reduction of ischaemia, and the amount of myocardium at risk. Beta-blockers can attenuate perfusion heterogeneity in both exercise and pharmacologic protocols but should not have significant effects on quantitative MBF. If haemodynamically active medications are to be interrupted, they should ideally be stopped for three to five half-lives.^13–15

Patients should be positioned in a supine position in a comfortable manner, with arms above their heads out of the camera’s field-of-view. A comfortable position should be ensured with supports below the knees and arms and with appropriate belts. Furthermore, patients should be informed about the common symptoms caused by vasodilators to avoid movement during image acquisition. Common side effects of vasodilators are flushing, headache shortness of breath, and chest pain.⁷ Aminophylline, an adenosine receptor antagonist, can be administered to shorten the duration of adverse effects of adenosine receptor agonists after image acquisition.⁷ Dobutamine may cause palpitations, chest pain, headache, flushing, and shortness of breath, or exacerbate tachyarrhythmias.

More detailed information regarding patients’ preparation, as well as general (contra)indications to cardiac stress tests have been covered in dedicated imaging guidelines, to which the reader is referred.^7,10

Clinical Pearl #1

One size does not fit all—PET MPI should be tailored to answer the clinical question. For each case, it is important to select the appropriate stress modality (exercise vs. pharmacologic) and decide on whether to withhold haemodynamically active medications.

Imaging protocols

The short-half-lives and high energy of PET tracers simplify imaging protocols. As compared with SPECT protocols, all PET protocols can be performed in 1 day and often in less than an hour. For the tracers with very short-half-lives (⁸²Rb-chloride and ¹⁵O-water), the same dose of tracer may be given for both rest and stress imaging, reducing radiation exposure. Imaging must be performed immediately after tracer injection to capture the arterial input function needed to quantify MBF. The interval between rest and stress imaging varies based on which tracer is used; for example, imaging protocols with ⁸²Rb-chloride, with the shortest half-life of 75 s, are typically quicker than those with ¹³N-ammonia, with a longer half-life of 10 min.^16,17 Examples of protocols with ⁸²Rb-chloride, ¹³N-ammonia, the newly approved, ¹⁸F-flurpiridaz, and ¹⁵O-water can be seen in Figures 3–6.

Figure 3.

Example protocol with ⁸²Rb-chloride.

Figure 6.

Example of protocol with ¹⁵O-water.

Figure 4.

Example protocol with ¹³N-ammonia (¹³N-NH₃). Protocols with shorter wait times can also be performed, using subtraction techniques to adjust for the residual activity.^18,19

Figure 5.

Example protocol with ¹⁸F-flurpiridaz. Stress-first imaging is preferred, as normal stress perfusion obviates the need for rest imaging. This may reduce radiation exposure, which is particularly important for ¹⁸F-flurpiridaz given its higher radiation dose compared with ¹³N-ammonia and ⁸²Rb-chloride. Rest-first protocols are also feasible/acceptable. Protocols with shorter wait times can be performed, using subtraction techniques to adjust for the residual activity, although these techniques have not specifically been validated with ¹⁸F-flurpiridaz.^18,19

Tracers

PET MPI has traditionally been performed using either ⁸²Rb-chloride, ¹³N-ammonia, or ¹⁵O-water.^{16,17  18}F-Flurpiridaz was recently approved in the USA.²⁰ Other F-18-based tracers are in earlier clinical-phase development.²¹ A summary of the properties of the current PET perfusion tracers is shown in Table 1.

Table 1.

Tracer properties for PET MPI

	82Rb-chloride	13N-ammonia	15O-water	18F-tracersa
Isotope production method	Generator	Cyclotron	Cyclotron	Cyclotron
Isotope half-life (min)	1.3	10	2.0	110
Mean estimated positron range (mm)b	6.8	1.8	3.0	0.6
Effective dose (mSv/GBq)	1	2	1	≈20
Peak stress/rest extraction (%)c	35/70	95/100	100	95/100
Peak stress/rest retention (%)c	25/70	50/90	0	55/90
Dose range/typical dose	10–30 MBq/kg	370–740 MBq	300–500 MBqd	≈100/220–250 MBqe

^aCurrently data for ¹⁸F-flurpiridaz and preliminarily data for ¹⁸F-SYN2 available.^16,22

^bData from Conti and Eriksson²².

^cData from Murthy et al.¹⁶

^dThree-dimensional scanner.

^eRest/stress.

Because of their short physical half-lives, the use of ¹³N-ammonia and ¹⁵O-water require an on-site cyclotron. ⁸²Rb-chloride is produced from an ⁸²Sr/⁸²Rb-generator, with supply typically lasting 4–8 weeks, depending on the initial activity and desired radiotracer activity. ¹⁸F-labelled tracers, because of the longer isotope half-life (∼2 h), can be produced at regional cyclotron or radiopharmacy and distributed as a unit dose. The short-half-lives of ⁸²Rb-chloride and ¹⁵O-water enable fast rest–stress imaging protocols, whereas the use of ¹³N-ammonia or ¹⁸F-flurpiridaz require either a delay of four to five half-lives or a higher dose for the second scan (typically stress). The effective radiation dose (mSv/GBq) is an order of magnitude lower for the short-lived isotopes than for ¹⁸F-labelled tracers, which can be balanced by reducing the total injected activity (at the expense of longer imaging times) and/or performing stress-first protocols. ⁸²Rb-chloride has a longer positron range than other perfusion tracers, but this has only limited effects on spatial resolution in daily practice.^16,17

The physiological properties of perfusion tracers vary in terms of myocardial extraction from blood and tissue retention, which may have an impact on image quality and choice of kinetic model for MBF quantification. ¹⁵O-water is a freely diffusible tracer with a high initial tissue extraction over a wide range of MBF values, resulting in a tracer uptake rate (K₁) that is close to the true MBF. However, ¹⁵O-water washes out rapidly and there is effectively no tracer retention in the myocardium above the blood background level. Thus, interpretation is based on parametric images of quantitative MBF rather than of qualitative and semi-quantitative retention images. ¹³N-ammonia, like ¹⁵O-water, has high initial tracer extraction along with tissue retention of ∼50–60% at peak stress. ⁸²Rb-chloride has substantially lower extraction fraction and tracer retention at peak stress than ¹³N-ammonia (Table 1). The ¹⁸F-labelled tracers with available data show high extraction fraction and retention^16,17,23,24 (Figure 7).

Figure 7.

Comparison of myocardial extraction between different commercially available PET and SPECT tracers.²⁵

Incomplete extraction and retention of tracer can reduce the contrast between high- and low-MBF regions, leading to an underestimation of regional perfusion abnormalities when assessed visually. When MBF is quantified, suboptimal extraction is corrected using a tracer-specific equation. The reliance on tracer-specific correction factors in tracers with lower extraction, such as ⁸²Rb-chloride, may have increased variability, particularly at higher blood flow ranges.^16,17 However, these differences are largely negligible in clinical practice.^26,27

The appropriate tracer for an institution must factor in real-world factors, such as cost, availability, and patient population. ¹³N-ammonia and ¹⁵O-water, for example, require an on-site cyclotron with close interval between production and patient-injection. An on-site cyclotron may not be feasible for many institutions. Alternatively, ⁸²Rb-chloride is created via an ⁸²Rb-generator via the decay of ⁸²Strontium (⁸²Sr); generators are typically placed adjacent to the PET scanner with autoinjectors in the patient. A sufficient volume of PET MPI must be performed to justify the expense of an ⁸²Rb-generator before the end of its lifespan. ¹⁸F-Flurpiridaz, the most recently approved PET tracer, has the longest half-life and can be created at a local radiopharmacy and delivered on the day of testing. Lastly, if an institution has a specific population for which exercise testing is critical, ¹³N-ammonia and ¹⁸F-flurpiridaz are the only current options. Each tracer comes with unique benefits and challenges; the optimal tracer will vary from institution to institution.^22,24

Clinical Pearl #2

Tracer production, half-life, protocols, extraction, and artefacts differ. These factors must be considered on both the patient level (i.e. image interpretation) and institutional level (i.e. setting up a nuclear practice).

Image acquisition

Contemporary PET scanners operate in 3D acquisition mode, as opposed to older 2D systems. Three-dimensional systems are more sensitive, requiring lower injected tracer activity, and therefore, lower patient radiation exposure. Careful consideration should be given to optimizing injected dose to avoid detector saturation during the blood pool first-pass intake (when MBF quantification is acquired), while preserving sufficient activity in the myocardial tracer retention phase is needed for perfusion images.^5,16 Detector saturation may result in falsely elevated MBF due to underestimation of the blood input function. The shorter-lived tracers, ⁸²Rb-chloride and ¹⁵O-water, are particularly challenging in this regard. The maximally tolerated activity varies greatly between 3D PET systems. New digital PET systems integrated with silicon photomultiplier detectors have increased the dynamic range of 3D PET systems, potentially reducing the need to trade-off perfusion image quality for MBF accuracy.^5,6,28

Photons emitted from the heart can be heterogeneously deflected from their linear path by surrounding tissues, reducing count rates at the detector. This phenomenon, called attenuation, can cause attenuation artefacts. PET cameras are commonly combined with computed tomography (CT), allowing for attenuation correction. In addition, hybrid PET/CT systems enable the combination of MPI with coronary artery calcium assessment, and, with scanners containing ≥64 CT detector rows, coronary CT angiography.^29–31 The use of state-of-the-art methodology enables integration of CT and PET imaging with an acceptable radiation exposure to the patient; the addition of a non-contrast CT scan also allows for non-obstructive, as well as obstructive plaque, to be detected via the detection of coronary calcium.

Image reconstruction

Similar to SPECT imaging displays, analytic reconstruction (typically with filtered back projection) and iterative reconstruction are the two techniques used for PET image reconstruction. Early cardiac PET imaging used analytical approaches for image reconstruction due to the excessive computing time required for iterative algorithms. As a result of significant improvements in computing technology and more efficient iterative algorithms, iterative reconstruction now is widely available and is the standard for PET image reconstruction.

In iterative reconstruction, the fundamental concept is to make a guess about the location of activity distribution. The projection data produced by this activity distribution are calculated and compared with the actual data acquired. If the two data sets differ, then the guess is adjusted based on the difference and the entire process is repeated. This process repeats until the data sets match. The crucial components of iterative reconstruction are the method(s) used to update the estimated activity distribution based on the differences in data sets and the calculation of projection data from the estimated activity distribution. The most commonly used approach in iterative reconstruction is the maximum-likelihood expectation maximization (MLEM) algorithm. MLEM requires much iteration to generate a clinically reasonable image and is computationally demanding.³² A modification to the MLEM algorithm, ordered subset expectation maximization (OSEM), allows for iterative construction to be clinically feasible. In OSEM, the full data set is not needed to update the image estimate. Rather, data from a few projections (subsets) are used to perform updates, thereby significantly decreasing computation time.³³

The advantage of iterative reconstruction over filtered back projection (analytic approaches) is the incorporation of many characteristics of the data acquisition, including attenuation, scatter, randoms, spatial resolution, and dead time, into the reconstruction model. This allows for the camera acquisition process to be more accurately represented in the data and yields higher fidelity images. Generally, a higher number of iterations results in more accurate image estimates with the trade-off of increased noise. Noise can be reduced by post-reconstruction smoothing (Figure 8). Most of the iterative reconstruction challenges encountered in the early days of clinical PET adoption have been resolved with modern computational advances. Images can now be reconstructed and displayed in <1–2 min.

Figure 8.

Iterative reconstruction and filtering. Cardiac PET horizontal long axis images reconstructed using OSEM algorithm. (A) 1 iteration, 5 subsets; (B) 3 iterations, 5 subsets; (C) 7 iterations, 5 subsets; (D) 10 iterations, 5 subsets; (E) 3 iterations, 5 subsets with an 8 mm Gaussian smoothing filter.

Imaging display and interpretation

Interpretation of PET-MPI studies should integrate all available clinical data including patient history, stress testing haemodynamics, electrocardiographic data, qualitative and semi-quantitative myocardial perfusion, gated left and right ventricular size and function, quantitative MBF, and CT findings, in particular the presence or absence of coronary artery calcifications with or without dedicated coronary artery calcium scoring (Figure 9). There are several commercially available software packages that allow for the display and interpretation of these data and allow for quantitative analysis of MPI studies.^34,35 Additionally, it is recommended that PET-MPI reports contain structured data elements, as previously described in American Society of Nuclear Cardiology (ASNC) guidelines and European Association of Cardiovascular Imaging (EACVI) position papers.^36–38

Figure 9.

Elements of a cardiac PET myocardial perfusion study. (A) Gated images at stress (top row) and rest (bottom row) with quantitative analysis of left ventricular ejection fraction and volumes. (B) Static perfusion images at stress (top row) and rest (bottom row) displayed in short axis, horizontal long axis, and vertical long axis planes. Perfusion defects are scored using a semi-quantitative scale on a bulls-eye plot, with summed stress, summed rest, and summed difference scores. (C) Coronary artery calcium score. (D) MBF analysis using tracer kinetic modelling, reporting rest, stress, and reserve MBF. ⁸²Rb-chloride is the tracer used in this study.

MPI data are displayed in standard tomographic planes: short axis, vertical long axis, and horizontal long axis (Figure 10). From these views, the left ventricle can be displayed in a 17-segment model or polar map, colloquially known as a bull’s eye. The polar maps are able to communicate the size, severity, and anatomical distribution of a defect³⁹ (Figure 11, Table 2). In this system, scores of 0 through 4 correspond to normal, mildly abnormal, moderately abnormal, severely abnormal, and absent uptake of radiopharmaceutical, respectively, in a particular left ventricular segment. Perfusion defects can be reported as fixed (present on rest and stress imaging) or reversible (only present on stress). Reversible defects typically represent ischaemia; fixed defects could represent scar or hibernating myocardium.

Figure 10.

Standard views of the left ventricle. Traditionally, stress images are placed above rest images. SA, short axis that runs from the apex (left) to the base (right); HLA, horizontal long axis that runs from the inferior (left) to anterior wall (right), VLA, vertical long axis that runs from the septum (left) to the lateral wall (right).

Figure 11.

Standard 17-segment model and polar map for reporting semi-quantitative perfusion. Each segment corresponds to a particular vascular distribution. Reproduced with permission from Abadie et al.⁴

Table 2.

Criteria for semi-quantitative assessment of defect size

Summed score	≤3	Normal or equivocal
	4–8	Mild defect
	9–12	Moderate defect
	>12	Severe defect

The total perfusion score at rest is the summed rest score (SRS), at stress is the summed stress score (SSS), and the difference between rest and stress, the summed difference score (SDS). The percentage of myocardium at risk can be calculated by dividing the SSS or SDS by 68 and multiplying by 100.³⁹

Left ventricular ejection fraction and left ventricular end diastolic and end systolic volumes can be accurately calculated by commercially available software using the gated images with 8 or 16 frames/cardiac cycle. Ejection fraction and left ventricular volumes should be reported for both rest and stress imaging data^40–42 (Table 3). Dyssynchrony, such as in the setting of left bundle branch blocks or ventricular pacing, can also be assessed qualitatively and quantitatively (Figure 12).

Table 3.

Categories of left ventricular function by ejection fraction

Categories of left ventricular function	Ejection fraction (%)
Hyperdynamic	≥70
Normal	55–70
Low normal	50–55
Mildly	45–50
Moderately	35–45
Severely reduced	≤35

Figure 12.

Myocardial dyssynchrony can be accessed via gated PET images. (1A) The patient has a narrow QRS on electrocardiogram with a narrow range of R-R intervals (1B). Panel 1C demonstrates that all the myocardial segments are reaching peak displacement at the same time. Conversely, patient B has a left bundle branch block and premature ventricular contraction (2A) with a wide range of R-R intervals (2B). Panel 2C shows a delayed and dyssynchronous peak displacement of the myocardial segments.

Quantitative MBF analysis is a central component of PET MPI studies and should be reported whenever available. Validated kinetic models used to calculate MBF are available in commercially available software, making MBF analysis clinically feasible, accurate, and reproducible.^16,37,43 Rest MBF and stress MBF, in addition to MBF reserve (MFR, the ratio of stress MBF to rest MBF), should be included in PET MPI reports and integrated into the overall interpretation of the PET-MPI study. Quantitative MBF data should be additive to the relative myocardial perfusion and left ventricular functional data and should factor into the overall impression of the PET-MPI study (Figure 13). The most commonly used threshold for an abnormal MFR is <2.0.⁴⁴ Of note, however, different stress modalities augment MBF differently, with exercise typically a two-fold increase, and vasodilators a three- to four-fold increase; dobutamine can augment greater than vasodilator agents.⁴⁵

Figure 13.

Example of quantitative MBF. Top left: retention images are used to ensure the myocardial border is properly registered and aligns with tracer uptake. Top right: figure demonstrates degree of patient motion during the exam which may interfere with accuracy of blood flow data. Bottom left: time-activity curves showing MBF at stress and rest. Note the early blood pool peak (green) that quickly drops to lower levels of activity. Conversely, the myocardial flow persists throughout the curve (other colours). Bottom right: resultant stress, rest, and reserve flow by vascular distribution, with adjacent global value. ⁸²Rb-chloride is the tracer used in this study.

Quantitative MBF has shown good agreement with invasive measures of stenosis severity, such as fractional flow reserve, and therefore may be able to predict the need for revascularization.⁴⁶ Quantitative MBF is particularly helpful in cases where the relative myocardial perfusion images are normal with abnormal myocardial flow reserve. In such cases, the first step is to ensure that there are no technical factors to account for this discordance (Figure 11). Assuming there are no other causes for abnormal MFR (for example, lack of vasodilator response, low MFR due to high resting MBF, technical issues with dynamic image acquisition), one needs to consider multi-vessel obstructive coronary artery disease and/or coronary microvascular dysfunction among other causes as possible aetiologies for low MFR⁴⁴ (Figure 14). Because perfusion imaging relies on the heterogeneous uptake of radiotracer in the left ventricle, severe obstruction in all coronary vessels may be missed, a phenomenon called balanced ischaemia. Quantitative MBF increases the sensitivity of PET in this high-risk population. Acknowledging the presence or absence of coronary artery calcium, the presence or absence of left ventricular systolic dysfunction, a decrement in left ventricular systolic function at the time of stress, and clinical history can assist in providing an assessment to why there are abnormalities in MFR despite visually normal perfusion.

Figure 14.

Normal myocardial flow reserve is typically defined as ≥2.0. In the absence of obstructive epicardial disease, an abnormal myocardial flow reserve (<2.0) can indicate microvascular dysfunction. Two main categories of microvascular dysfunction have been described, classical/structural with low stress flow, and functional/endogen with high rest flow.

Clinical Pearl #3

The ability to quantify myocardial blood flow is a key advantage of PET MRI compared with traditional SPECT imaging and should be readily incorporated into reading and reporting protocols.

Artefacts and non-diagnostic studies

High-quality scans and precise image interpretation are crucial for fully leveraging the capabilities of cardiac PET in patient care. Imagers should be able to distinguish between low and high-quality PET-MPI scans, identify reasons for non-diagnostic studies, understand what causes these artefacts, and how to effectively correct them.⁴⁷ Artefacts are the most frequent reason for non-diagnostic PET-MPI studies, followed by non-response to pharmacological stressors.

Discrepancies between visual interpretation and quantitative flow occur. With vasodilator protocols, when perfusion is normal, stress, and reserve flows are abnormal, it is vitally important to ensure that the patient had an adequate response to the vasodilator. Common reasons for an inadequate response are consumption of adenosine antagonists prior to the examination (e.g. caffeine-containing food and drink) or extravasation of the vasodilating agents. Patients adequately exposed to vasodilators should have an increase in their heart rate and a drop in blood pressure; however, haemodynamic changes are not always reliable. Splenic switch-off, the relative decrease in tracer activity in the spleen during stress compared with rest, can be used to demonstrate that a patient had an adequate vasodilatory response⁴⁸ (Figure 15). This finding was first described with adenosine protocols, but has more recently been demonstrated with regadenoson.⁴⁹ A change in quantitative MBF from rest to stress can also be used to ensure the patient was adequately stressed. Abnormal visual perfusion with normal quantitative flow can also occur.

Figure 15.

The PET/CT fusion images at stress (A) show a decrease in tracer uptake in the spleen compared with the rest images (B). This sign, termed splenic switch-off, can be used to ensure the patient had an adequate response to the vasodilator stress agent, which in this case was regadenoson. ⁸²Rb-chloride is the tracer used in this study.

Common artefacts, independent of radiotracers and vasodilators, include attenuation-correction-artefacts, mis-registration of emission and transmission scans, and patient motion; these artefacts can often be easily recognized and, to a certain extent, improved in post-processing.^50,51 Left ventricular hypertrophy, particularly in hypertrophic cardiomyopathy, can cause an increase in tracer uptake, which can cause the relative difference in the contralateral segments to be mistaken for hypoperfusion.

Quantitative PET-MPI artefacts affecting or rendering MBF assessment uninterpretable often arise from inaccurate attenuation correction, patient movement, and suboptimal tracer-bolus delivery.^17,52 Suboptimal tracer-bolus delivery can be detected via abnormalities in the myocardial time-activity curves⁵³ (Figure 16). Importantly, for both qualitative and quantitative analysis, reduced image count density, for example due to extravasation, use of i.v. ports located close to the heart, or extreme obesity (without injected dose adjustment), can directly affect the diagnostic quality of the study.⁵⁰

Figure 16.

(A) The patient had a poorly functioning intravenous catheter and extravasation of the radiotracer. As a result, the curves showing the uptake of the tracer at rest (yellow and blue) are delayed compared with stress (red and green). (B) Correct time-activity curve where the rest and stress tracer activity peaks at the same time.

Careful scheduling of the scan based on patient characteristics and radiotracer availability, coaching the patients prior to and during the examination, careful review of the fused images, and the use of modern software that enables correction of cardiorespiratory motion and myocardial creep (stress-induced, non-periodical repositioning of the heart in the mediastinum) are critical to mitigate artefacts and minimize non-diagnostic studies.^47,54–56  Table 4 summarizes the most frequent artefacts with suggested solutions and Figure 17 demonstrates examples commonly encountered in clinical practice.

Table 4.

Summary of the most common PET-MPI artefacts, how they can be recognized, and potential strategies for mitigating their impact

Artefact	Cause	How to Recognize	How to Tackle
Motion artefacts	Patient movement or heavy breathing.	Blurring of images, observed on static and/or dynamic images. Atypical fixed or reversible perfusion defects (butterfly appearance). Erroneous myocardial blood flow estimates. Most prominent usually during stress.	Patients must stay as still as possible during the scan. Coaching patients prior to the scan. Instruct patients during the scan. Challenging to correct post-acquisition. Deleting frames with motion and re-summation of motion-free frames might helpful if artefact limited. Use software that corrects for motion during post-processing.
Attenuation artefacts	Improper alignment/superimposition of PET and attenuation-correction CT scans. Commonly during co-registration of transmission and emission scans at the scanner console. Patient motion/breathing during transmission or emission scans.	Might be challenging to identify. Fixed or reversible defect commonly in the lateral (anterior and/or inferior) wall. Discordance between static and dynamic (motion and breathing-corrected) images.	Training/feedback of technicians is paramount. Quality control of fused images at interpretation workstation. Repeat fusion and reconstruction of images. Less important if only at rest and in the case of normal stress perfusion.
Suboptimal bolus delivery/detector saturation	Improper manual tracer administration. High-dose injections of short-half-life and low extraction fraction tracers (commonly 82Rb-chloride). Risk greater in higher sensitivity 3D PET scanners and systems with photomultiplier tube technology.	Careful examination of the time-activity curves. Sharp peaks or troughs in the time-activity curve, suggesting rapid changes in tracer concentration. Plateau-like peaks occurring at the first-pass of the tracer-bolus Sudden drops in detected counts during the first-pass phase, leading to gaps or incomplete data. Erroneous myocardial blood flow estimates (usually increased MBF values).	Optimize injection protocols to ensure smooth and consistent tracer delivery. Avoid manual injection of tracer. Use fast-controlled automated tracer injection. Ensure image acquisition at least 5 s prior to radiotracer injection. Adjust the tracer dose or optimize the scanner settings to prevent overload of the detection system.
Injection site artefacts	Tracer injection by port or central venous line near the heart Local issues with tracer administration (i.v. line).	High tracer uptake at non-target sites visible on the scan, usually near the injection site.	Ensure proper injection techniques and inspect the injection site before scanning. If possible, avoid using central venous lines.
Myocardial creep	Shift in the position of the heart over the course of a dynamic scan, particularly during stress.	Visual mis-registration of at least one third of the left ventricular wall width, commonly (but not limited) in the downward (inferior) direction.	Monitor dynamic dataset frame by frame and correct for any shifts using real-time imaging feedback and post-processing alignment tools.
Spillover effects	Activity from adjacent organs (gastrointestinal tract, lungs) or blood pool, masks, or distorts the actual myocardial signal.	Might be challenging to identify. Artificially elevated counts or apparent lower counts in myocardium. Discordance between static and dynamic (motion and breathing-corrected) images.	Adjust regions of interest on both static and dynamic images to exclude if possible, the extracardiac activity. Interpret accordingly in cases where complete exclusion is not feasible (e.g. lung uptake adjacent to lateral wall)

Figure 17.

Demonstrative examples of the most common artefacts encountered at PET MPI. (A) Typical appearance of a motion artefact (butterfly appearance) evident at both stress and rest images of a restless patient with claustrophobia undergoing PET MPI. (B) Typical fixed lateral wall defect in a patient undergoing rest/stress ¹³N-NH₃ PET MPI. Note that the defect is more evident at stress due to higher tracer dose applied. (C) Mis-registration of the stress PET images with the CT transmission images resulting in a reversible perfusion defect at the inferolateral basal wall. (D) Typical apical thinning defect at a patient undergoing PET MPI with normal wall motion and thickening. (E) Increased lung uptake at a patient undergoing rest/stress ¹³N-NH₃ PET MPI directly after cigarette smoking. Note the more prominent extracardiac uptake during rest. Asterisks depict the artefact at the slices, and arrow heads depict the corresponding artefact appearance at the polar plots.

In addition to those general pitfalls, the different radiotracers used in PET MPI might be presented with specific artefacts. ⁸²Rb-chloride, with its rapid decay, might cause detector saturation during tracer first-pass, while ⁸²Sr generator age-related decrease in activity per volume might impact the blood first-pass curve.^57,58 Gastrointestinal tract uptake of the tracer, particularly in patients taking proton pump inhibitors and following gastric bypass surgery, and increased blood pool activity, commonly seen in patients with severely depressed left ventricular function, may limit diagnostic image quality.^59,60

¹³N-ammonia, while generally offering high-quality uptake images, is challenged by attenuation artefacts from respiratory movements, such as lung uptake impacting proper visualization of the lateral segments of the left ventricle, myocardial creep, and apical thinning.^17,61–63 The latter is observed with all tracers but is in general more pronounced with ¹³N-ammonia. A fixed basal lateral/inferolateral wall perfusion defect with normal wall motion and wall thickening can be also observed.^64–66 Finally, increased lung uptake (primarily in rest and less under stress) has been reported in patients with decompensated heart failure, inflammatory lung disease, and recent smoking just prior to the exam.^67,68

¹⁵O-water PET is less susceptible to myocardial creep due to the absence of uptake images; however, it requires meticulous correction for high activity in the blood pool and spill over from the left and right ventricles, a process typically integrated into kinetic modelling software.^69,70

Clinical Pearl #4

Diagnostic confidence can only be as high as test quality. Ensure you are attempting to pre-empt The introduction of artefacts into images, as well as properly recognizing and reporting them when they occur.

Viability imaging

Viable myocardium encompasses a broad spectrum of conditions and includes stunned and hibernating myocardium⁷¹ (Figure 18). The primary focus of this section will be the detection of hibernating myocardium, which is often indistinguishable from scar on traditional perfusion imaging.^72–74 Viability imaging can identify dysfunctional myocardium that has the potential for recovery with revascularization and can therefore guide therapy.⁷⁵

Figure 18.

The spectrum of myocardial ischaemic dysfunction.

Viability imaging can be performed with both SPECT, with thallium or technetium (^99mTc), and PET, with ¹⁸F-fluorodeoxyglucose (FDG). Delayed thallium imaging is an outdated and rarely used method to assess viability and will not be discussed in this position paper. ^99mTc tracers passively diffuse across membranes where uptake and mitochondrial retention are conditioned by the electrochemical gradient. To assess viability, acquisition of rest images is performed under maximal vasodilatation with nitrates.⁹

PET imaging with FDG is preferred over SPECT for the assessment of viability.⁶⁷ FDG has a relatively short-half-life of 109.8 min, or slightly <2 h. Nevertheless, this half-life is sufficiently long to allow shipping of the tracer to remote PET scanning facilities where it cannot be produced with a local cyclotron. Common tracer doses and imaging protocols can be found in Table 5 and Figure 19.

Table 5.

Common radiation dose to patients by different PET stress protocols and tracers

Study protocol	Radiotracer	Typically administrated activity (mCi/MBq)	Estimated radiation dose (mSv)
3D PET
Rest + stress perfusion	82Rb	20 + 20/740  +  740	2
Rest + stress perfusion	13N-NH3	10 + 10/370 + 370	2
Rest + stress perfusion	15O-water	10 + 10/370 + 370	0.8
Rest perfusion + viability	82Rb/18F-FDG	20 + 5/740 + 185	2 + 3.5
Rest perfusion + viability	13N-NH3/18F-FDG	10 + 5/370 + 185	2 + 3.5
2D PET
Rest + stress perfusion	82Rb	50 + 50/1480  +  1480	4
Rest + stress perfusion	13N-NH3	20 + 20/740 + 740	4
Rest + stress perfusion	15O-water	30 + 30/1110 + 1110	2.4
Rest + stress perfusion	18F-flurpiridaz	2.5–3.0 + 9.0–9.5/93–111 + 333–352a	7a
		2.5–3.0 + 6.0–6.5/93–333 + 222–241b	6.3b
Rest perfusion + viability	82Rb/18F-FDG	50 + 20/1480 + 375	4 + 7
Rest perfusion + viability	13N-NH3/18F-FDG	20 + 10/375	4 + 7

Radiation dose does not include contribution from CT, which may vary based on type of scanner and whether concurrent calcium scoring is performed.^23,76–78

^aFlurpiridaz reported rest/stress doses for rest/exercise stress.

^bFlurpiridaz reported rest/stress doses for rest/pharmacological stress.

Figure 19.

Examples of viability imaging protocols for PET with ¹⁸F-FDG and SPECT with ⁹⁹m-technetium (^99mTc). For FDG-PET (top), patients consume a low-fat meal the evening prior. Rest imaging is performed with traditional perfusion tracers. If there are fixed perfusion defects, the patient can proceed with viability imaging. Patients undergo glucose loading followed by injection of IV insulin to target goal serum glucose. FDG is then injected and viability imaging performed. Hibernating myocardium will have uptake of FDG. For SPECT (bottom), no dietary prep is necessary. After rest imaging demonstrates fixed perfusion defects, patients can undergo viability imaging. Nitroglycerine is administered prior to the second injection of technetium to augment MBF and tracer uptake. SL NTG, sublingual nitroglycerine.

In chronically ischaemic myocardium, there is a relative increase in glucose utilization compared with long-chain fatty acids. Areas of myocardium with resting hypoperfusion on traditional perfusion imaging, but with the uptake of FDG, are considered viable/hibernating (Figure 19). The assessment of viability with FDG-PET requires careful patient preparation to promote glucose utilization by the myocardium.

Viability imaging—patient preparation and imaging protocols

To ensure that the myocardium will preferentially use glucose instead of fatty acids, a strict preparation is required. The evening preceding the investigation, a low-fat meal should be consumed. Prior to the exam, non-diabetic patients typically receive a glucose load along with intravenous insulin to promote myocardial metabolism of glucose over fatty acids.⁷⁹ This can be achieved via oral glucose loading or a glucose infusion.

Oral glucose loading can be achieved with 25–100 g glucose dose, typically followed by intravenous insulin. The dose of intravenous insulin varies based on the blood glucose level. Studies have demonstrated that with proper protocols and monitoring, the administration of IV insulin has an excellent safety profile.⁸⁰ Oral glucose loading is less labour intensive and time consuming, but can result in suboptimal images and up to 10% uninterpretable scans. In diabetic patients, this method will not lead to sufficient cardiac FDG uptake.⁸¹

With euglycaemic hyperinsulinaemia clamping, there is continuous and simultaneous infusion of glucose and insulin preceding, during, and after FDG administration. This method maximizes glucose, and thus FDG, uptake in the myocardium. This method results in high insulin levels and consequently excellent image quality. However, it can be labour intensive and time consuming. Less extensive versions of this method have also been reported with excellent image quality.⁸¹

Acipimox, a niacin derivative, can be used to lower free circulating fatty acid levels and thereby promote the use of glucose by the myocardium. Several protocols exist for the use of acipimox and the timing of its administration prior to tracer injection. In these protocols, imaging is started 45–60 min after FDG injection. Flushing, a common side effect of acipimox, can be prevented by orally administering 500 mg acetylsalicylic acid (aspirin). In diabetic patients, a short acting insulin can be added shortly before FDG administration.⁸¹

Viability imaging—image display and interpretation

FDG-imaging should be interpreted in the context of the resting, and when appropriate, stress perfusion imaging. In cases with hibernation, there will typically be a fixed perfusion defect on the rest MPI imaging with corresponding areas of FDG uptake (Figure 20). Conversely, in cases with scar, there will be no FDG uptake in the areas of fixed perfusion defects (Figure 21). Of note, studies can have areas of both hibernating and scarred myocardium. When stress images are also obtained, FDG uptake can be seen in areas of reversible perfusion defects, a phenomenon called ischaemic memory (Figure 22). Since ¹⁵O-water relies on quantitative, rather than qualitative, assessment, it cannot be used for viability imaging, which relies exclusively on the qualitative comparison of perfusion and FDG-imaging. Studies are underway to determine whether measures of MBF can be used to differentiate scar vs. hibernation, but this is largely pre-clinical.⁸²

Figure 20.

(A) Large fixed perfusion defect in the inferior and inferolateral segments on resting ⁸²Rb-chloride imaging. (B) Viability imaging shows FDG uptake in the inferior and inferolateral wall demonstrating hibernating myocardium.

Figure 21.

(A) Large fixed perfusion defect in a left anterior descending distribution on resting ⁸²Rb-chloride imaging. (B) Viability imaging shows no FDG uptake in these segments, demonstrating scar.

Figure 22.

Stress/rest/viability imaging with ⁸²Rb. Stress (A) and rest (B) imaging shows a small, fixed defect in the inferior segments along with ischaemia in the lateral wall. (C) FDG images show scar in the inferior segments. FDG uptake in the lateral wall, corresponding to the ischaemic territory, is an example of ischaemic memory.

Viability imaging—artefacts and non-diagnostic studies

Artefacts due to soft tissue attenuation, patient motion and mis-registration can occur, emphasizing the need for adequate attenuation correction and quality control. Patient preparation for viability PET can be problematic in patients with Type 2 diabetes, who have underlying insulin resistance.^9,47 Lastly, the threshold of glucose uptake and number of segments that predict functional improvement is not well defined, as the literature is heterogeneous with mixed results. Decisions to pursue further therapy based on the results of viability testing require integration of other imaging modalities and the clinical context.

Viability imaging—hybrid PET and cardiac magnetic resonance imaging

PET and cardiac magnetic resonance imaging (CMR) are the two most advanced modalities for assessing hibernation. Rather than measuring metabolic activity, as in PET, contrast-enhanced CMR can assess for viability by measuring the degree of fibrosis and extracellular expansion where gadolinium accumulates to a greater extent than normal myocardium.⁷² On CMR, myocardial segments for which there is >50% late gadolinium enhancement is considered scar and unlikely to recovery. Since PET and CMR rely on different techniques to assess viability, hybrid PET/CMR may provide a more accurate assessment of viability and aid in resolving discrepancies when traditional PET/CT and CMR are discordant.^83,84 However, further research is warranted to better understand the additive value of combining these two modalities.

Clinical Pearl #5

PET is more sensitive than SPECT for the detection of hibernating myocardium and is therefore the preferred modality when available.

CT for PET including incidental findings

Attenuation correction is required for interpretation of cardiac PET.⁸⁵ A CT scan is obtained either pre- and/or post the radiotracer emission scan with a field-of-view that includes the chest to avoid truncation artefacts. Patients must remain still to avoid mis-registration artefacts (Figure 23). To reduce radiation dose, CT performed for attenuation correction typically uses lower tube current and larger slice thickness than dedicated cardiac imaging, such as is used for calcium scores or coronary CT angiograms. The CT is performed without intravenous contrast, ECG-gating, or breath-holds. Consequently, this CT is not optimized for diagnostic chest imaging.

Figure 23.

CT transmission-emission misalignment. Misalignment of CT transmission and ¹³N-NH₃ emission scan inferiorly and inferolateral causes an apparent perfusion defect (A) that disappears with proper alignment (B).

Despite the lower resolution, clinically actionable findings, both cardiac and non-cardiac, can be present and should be reported.⁸⁶ Common cardiac findings include coronary or valve calcification, which can be reported as either binary (present or absent) or according to qualitative categories (mild, moderate, and severe), and pericardial effusions (Figure 24). CT in cardiac PET includes the thoracic aorta and may identify patients with abnormal dilations (Figure 25). Hypoattenuation of the liver (<40 Hounsfield units) on non-contrast CT may signify non-alcoholic fatty liver disease which can provide additional prognostic information and change medical therapy.⁸⁷ Lung or breast findings are also common and can include masses, infiltrates, or pleural effusions (Figure 24). Many risk factors for cardiovascular disease also increase the risk of cancer, emphasizing the need for a thorough assessment of the CT data. Prior dedicated chest or cardiac CT can be reviewed to assess for interval change.

Figure 24.

Incidental findings on CT performed for attenuation correction. (A) Coronary artery calcification (dotted line) and bilateral pleural effusions (solid line). (B) Breast mass (dotted line) and severe mitral annular calcification (solid line).

Figure 25.

Incidentally found, aneurysmal dilation of the ascending aorta.

A dedicated coronary artery calcium score can be performed in patients without known CAD. Coronary calcium scoring requires prospective, ECG-gating with inspiratory breath-holds and therefore cannot be used for attenuation correction. Adding a coronary artery calcium score may improve specificity, as patients with low coronary artery calcium scores are less likely to have flow-limiting CAD. Coronary artery calcium scoring can provide additional prognostic value and should be integrated with the patient’s age, sex, and traditional risk factors. The prognostic value of calcium scoring should not be applied to those with prior revascularization.⁸⁸

Calcium scoring can also be used to guide medical management with the initiation of lipid-lowering therapies, such as statins, or antiplatelet medications.^89,90 Statin therapy slows atherosclerosis, but can convert non-calcified plaque to calcified plaque and therefore worsen calcium scores.⁹¹ While calcium scoring continues to have prognostic value in patients on statins, the relationship is weaker compared with non-statin users.⁹² Patient education is critical to ensure patients understand how their therapy may affect calcium scoring on serial MPI studies.

Clinical Pearl #6

Incidental CT findings can have significant impact on patient management, particularly the presence of coronary calcium. This may prompt changes in medical management of coronary artery disease, even in the absence of ischaemia.

Artificial intelligence and nuclear imaging

Artificial intelligence (AI) is being increasingly introduced into many aspects of cardiovascular imaging, including nuclear myocardial perfusion. The primary areas of ongoing advancement are image reconstruction, image interpretation, and prognostication.⁹³ Like iterative reconstruction, AI may be used in post-image processing to enhance signal to noise ratio and improve image resolution.⁹⁴ Retrospective studies have also shown promising AI models for direct image interpretation and prediction of major adverse cardiovascular events.^95,96 Prospective research is needed to best determine the generalizability and cost-effectiveness of various nuclear MPI AI models. A more detailed position paper on AI applications in nuclear imaging has been previously published by the EACVI.⁹⁷

Generating a clinically meaningful report

A clinically meaningful report includes an assessment of image quality, interpretation of perfusion and gated images, quantification of MBF, evaluation of extracardiac findings, and finally, a clinical impression with risk-stratification (Figure 26).

Figure 26.

Pillars of a comprehensive PET-MPI report.

Image quality should be reported, giving an impression of the readers’ diagnostic confidence and artefacts that may limit the interpretation of the exam.

Perfusion imaging should be reported in standard approach with comment on the coronary distribution, defect size, and defect severity. Markers of high-risk, including transient ischaemic dilatation or a drop in left ventricular ejection fraction, should be noted. Chamber size and function from the gated images should be included. MBF is a key strength of PET MPI and can improve the sensitivity of the exam. MBF should be reported when data are available and of high quality.

Associated symptoms or ECG-changes during the exam may be clinically meaningful. A comment on coronary calcification by CT can also improve the diagnostic and prognostic value of the exam.

Finally, after combining the MPI, MBF, and non-MPI results, the reader should provide a personalized and actionable report for clinicians to integrate into their practice.

Conclusion

PET MPI is an advanced nuclear molecular imaging tool for the diagnosis and risk-stratification of CAD, as well as the assessment of scar and hibernation. This document provides a comprehensive overview of the technical aspects of PET MPI, with a focus on patient preparation, tracer characteristics, image acquisition and interpretation, and common pitfalls and artefacts that are encountered in daily practice.

Supplementary data

Supplementary data are available at European Heart Journal—Imaging Methods and Practice online.

Data availability

No new data were generated or analysed in support of this research.

Lead author biography

Dr Bryan Abadie is an Associate Staff at the Cleveland Clinic Heart, Vascular, and Thoracic Institute. His clinical and research interest is multimodality imaging, including nuclear cardiology, echocardiography, cardiac magnetic resonance imaging, and cardiac computed tomography angiography.