Management of Respiratory Motion Artefacts in 18F-fluorodeoxyglucose Positron Emission Tomography using an Amplitude-Based Optimal Respiratory Gating Algorithm

Summary

Amplitude-based optimal respiratory gating (ORG) effectively removes respiratory-induced motion blurring from clinical ¹⁸F-fluorodeoxyglucose (FDG) positron emission tomography (PET) images. Correction of FDG-PET images for these respiratory motion artefacts improves image quality, diagnostic and quantitative accuracy. Removal of respiratory motion artefacts is important for adequate clinical management of patients using PET.

Abstract

Positron emission tomography (PET) combined with X-ray computed tomography (CT) is an important molecular imaging platform that is required for accurate diagnosis and clinical staging of a variety of diseases. The advantage of PET imaging is the ability to visualize and quantify a myriad of biological processes in vivo with high sensitivity and accuracy. However, there are multiple factors that determine image quality and quantitative accuracy of PET images. One of the foremost factors influencing image quality in PET imaging of the thorax and upper abdomen is respiratory motion, resulting in respiration-induced motion blurring of anatomical structures. Correction of these artefacts is required for providing optimal image quality and quantitative accuracy of PET images.

Several respiratory gating techniques have been developed, typically relying on acquisition of a respiratory signal simultaneously with PET data. Based on the respiratory signal acquired, PET data is selected for reconstruction of a motion-free image. Although these methods have been shown to effectively remove respiratory motion artefacts from PET images, the performance is dependent on the quality of the respiratory signal being acquired. In this study, the use of an amplitude-based optimal respiratory gating (ORG) algorithm is discussed. In contrast to many other respiratory gating algorithms, ORG permits the user to have control over image quality versus the amount of rejected motion in the reconstructed PET images. This is achieved by calculating an optimal amplitude range based on the acquired surrogate signal and a user-specified duty cycle (the percentage of PET data used for image reconstruction). The optimal amplitude range is defined as the smallest amplitude range still containing the amount of PET data required for image reconstruction. It was shown that ORG results in effective removal of respiration-induced image blurring in PET imaging of the thorax and upper abdomen, resulting in improved image quality and quantitative accuracy.

Introduction

Positron Emission Tomography (PET) in combination with X-ray computed tomography (CT) is a widely accepted imaging tool in clinical practice for accurate diagnosis and clinical staging of a variety of diseases¹. The advantage of PET imaging is the ability to visualize and quantify a myriad of biological processes in vivo with high sensitivity and accuracy². This is achieved through intravenously administering a radioactively labelled compound, also known as a radiotracer, to the patient. Depending on the radiotracer being used, tissue characteristics such as glucose metabolism, cellular proliferation, degree of hypoxia, amino acid transport, and expression of proteins and receptors, can be visualized and quantified².

Although several radiotracers have been developed, validated, and used in clinical practice, the radioactive glucose analogue ¹⁸F-fluorodeoxyglucose (FDG) is the most widely used radiotracer in clinical practice. Given that FDG predominantly accumulates in cells with an elevated glycolytic rate (i.e., cells with elevated glucose uptake and conversion to pyruvate for energy production), it is possible to discriminate tissues with different metabolic states. Similar to glucose, the first step of FDG uptake is transport from the extra-cellular space over the plasma membrane to the intra-cellular space, which is facilitated by glucose transporters (GLUT)³. Once the FDG is in the intra-cellular space, phosphorylation by hexokinases will result in the generation of FDG-6-phosphate. However, in contrast to glucose-6-phosphate, FDG-6-phosphate cannot enter the Krebs cycle for further aerobic dissimilation due to the absence of a hydroxyl (OH) group at the second (2’) carbon position. Given that the reverse reaction, the dephosphorylation of FDG-6-phosphate back to FDG, hardly occurs in most tissues, the FDG-6-phosphate is trapped intracellularly³. Therefore, the degree of FDG uptake is dependent on the expression of the GLUT (in particular GLUT1 and GLUT3) on the plasma membrane, and the intracellular enzymatic activity of hexokinases. The concept of this continuous uptake and trapping of FDG is referred to as metabolic trapping. The fact that FDG preferentially accumulates in tissues with an elevated metabolic activity is shown in Figure 1a, demonstrating the physiological distribution of FDG in a patient. This FDG-PET image shows higher uptake in heart, brain, and liver tissues, which are known to be metabolically active organs under normal conditions.

The high sensitivity for detecting differences in the metabolic state of tissues makes FDG an excellent radiotracer for discriminating normal from diseased tissues, given that an altered metabolism is an important hallmark for many diseases. This is readily depicted in Figure 1b, showing an FDG-PET image of a patient with stage IV non-small cell lung cancer (NSCLC). There is increased uptake in the primary tumor as well as in metastatic lesions. In addition to visualization, quantification of radiotracer uptake plays an important role in clinical management of patients. Quantitative indices derived from PET images reflecting the degree of radiotracer uptake, such as the standardized uptake value (SUV), metabolic volumes, and total lesion glycolysis (TLG), can be used to provide important prognostic information and measure treatment response for different patient groups^4,5,6. In this regard, FDG-PET imaging is increasingly being used to personalize radiotherapy and systemic treatment in oncology patients⁷. Furthermore, the use of FDG-PET for monitoring acute treatment induced toxicity, such as radiation induced esophagitis⁸, pneumonitis⁹ and systemic inflammatory responses¹⁰, has been described and provides important information for making image-guided treatment decisions.

Given the important role of PET for clinical management of patients, image quality and quantitative accuracy is important for appropriately guiding treatment decisions based on PET images. However, there are numerous technical factors that can compromise quantitative accuracy of PET images¹¹. An important factor that can significantly influence image quantification in PET is related to the longer acquisition times of PET compared to other radiological imaging modalities, typically several minutes per bed position. As a consequence, patients are usually instructed to breath freely during PET imaging. The result is that PET images suffer from respiratory induced motion, which can lead to significant blurring of organs located within the thorax and upper abdomen. This respiratory-induced motion blurring can significantly impair adequate visualization and quantitative accuracy of radiotracer uptake, which can affect clinical management of patients when using PET images for diagnosis and staging, target volume definition for radiation treatment planning applications, and monitoring of therapy response¹².

Several respiratory gating methods have been developed in an attempt to correct PET images for respiratory motion artefacts¹³. These methods can be categorized into prospective, retrospective, and data-driven gating strategies. Prospective and retrospective respiratory gating techniques typically rely on the acquisition of a respiratory surrogate signal during PET imaging¹⁴. These respiratory surrogate signals are used to track and monitor the patient’s respiratory cycle. Examples of respiratory tracking devices are detection of chest wall excursion using pressure sensors¹² or optical tracking systems (e.g., video cameras)¹⁵, thermocouples to measure the temperature of breathed air¹⁶, and spirometers to measure airflow and thereby indirectly estimating volume changes in the patient’s lungs¹⁷.

Respiratory gating is then typically accomplished by continuously and simultaneously recording a surrogate signal (designated S(t)), with the PET data during image acquisition. Using the surrogate signal acquired, PET data corresponding to a particular respiratory phase or amplitude range (amplitude-based gating) can be selected^12,13,18. Phase-based gating is performed by dividing each respiratory cycle into a fixed number of gates, as depicted in Figure 2a. Respiratory gating is then performed by selecting data acquired at a particular phase during the patient’s respiratory cycle to be used for image reconstruction. Similarly, amplitude-based gating relies on defining an amplitude range of the respiratory signal, as shown in Figure 2b. When the value of the respiratory signal falls within the set amplitude range, the corresponding PET listmode data will be used for image reconstruction. For retrospective gating approaches, all data is collected and re-binning of the PET data is performed after image acquisition. Although prospective respiratory gating methods use the same concepts as retrospective gating approaches for re-binning of PET data, these methods rely on collecting data prospectively during image acquisition. When a sufficient amount of PET data is collected, image acquisition will be finalized. The difficulty of such prospective and retrospective gating approaches is maintaining acceptable image quality without significantly prolonging image acquisition times when irregular breathing occurs¹³. In this regard, phase-based respiratory gating methods are particularly sensitive to irregular breathing patterns^13,19, where significant amounts of PET data can be discarded due to rejection of inappropriate triggers, resulting in considerable reduction of image quality or unacceptable lengthening of image acquisition time. Additionally, when inappropriate triggers are accepted, the performance of the respiratory gating algorithm and thereby the effectiveness of motion rejection from the PET images can be reduced due to the fact that respiratory gates are defined at different phases of the respiratory cycle, as depicted in Figure 2a. Indeed, it has been reported that amplitude-based respiratory gating is more stable than phase-based approaches in case of irregularities in the respiratory signal¹³. Though amplitude-based respiratory gating algorithms are more robust in the presence of irregular breathing frequencies, these algorithms are more sensitive to baseline drifting of the respiratory signal. Drifting of the baseline signal can occur due to numerous reasons when the patient’s muscle tension (i.e., transition of a patient into a more relaxed state during image acquisition) or breathing pattern changes. In order to prevent such baseline drifting of the signal, care should be taken to securely attach tracking sensors to the patient and perform regular monitoring of the respiratory signal.

Although these problems are known, traditional respiratory gating algorithms only allow limited control over image quality and usually require significant lengthening of image acquisition time or increased amounts of radiotracer to be administered to the patient. These factors resulted in limited adoption of such protocols in clinical routine. In order to circumvent these problems related to the variable quality of the respiratory gated images , a specific type of amplitude-based gating algorithm, also known as optimal respiratory gating (ORG), has been proposed¹⁸. Respiratory gating with ORG permits the user to specify image quality of the respiratory gated images by providing a duty cycle as input to the algorithm. The duty cycle is defined as a percentage of the acquired PET list-mode data that is used for image reconstruction. In contrast to many other respiratory gating algorithms, this concept permits the user to directly determine image quality of the reconstructed PET images. Based on the duty cycle specified, an optimal amplitude range is calculated, which takes the specific characteristics of the entire respiratory surrogate signal into account¹⁸. The optimal amplitude range for a specific duty cycle will be calculated by starting with a selection of different values for the lower amplitude limit, designated (L), of the respiratory signal. For each selected lower limit, the upper amplitude limit, designated (U), is adjusted in such a way that the sum of the selected PET data, defined as data acquired when the respiratory signal falls within the amplitude range (L<S(t)<U), is equal to the specified duty cycle. For example, for a duty cycle of 50% and six minutes of acquired PET listmode data, the amplitude range is adapted to include three minutes (50%) of PET data. The optimal amplitude range (W) is defined as the smallest amplitude range used for respiratory gating that still contains the required amount of PET data (i.e., ArgMax([U-L])), as depicted in Figure 2c¹². Thus, by specifying the duty cycle, the user makes a trade-off between the amount of noise and the degree of residual motion residing in the ORG PET images. Lowering the duty cycle will increase the amount of noise, though this will also reduce the amount of residual motion in the PET images (and vice versa). Although the concepts and effects of ORG have been described in previous reports, the purpose of this manuscript is to provide clinicians with details on the specific protocols when using ORG in clinical practice. Therefore, the use of ORG in a clinical imaging protocol is described. Several practical aspects, including patient preparation, image acquisition and reconstruction protocols will be provided. Furthermore, the manuscript will cover the user interface of the ORG software and specific choices that can be made when performing respiratory gating during PET imaging. Lastly, the effect of ORG on lesion detectability and image quantification, as shown in previous studies, are discussed.

Protocol

All procedures performed involving human participants were in accordance with the ethical standards of the internal review board (IRB) of the Radboud university medical center and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. The ORG algorithm is a vendor specific product and is available on the Siemens Biograph mCT PET/CT scanner family and newer PET/CT models.

1. Patient preparation

1. Patient anamnesis
   1. Check the patient’s name and date of birth. Inclusion criteria are similar to routine non-gated PET scanning. No additional in- or exclusion criteria are required.
   2. Check the label delivered with the syringe containing the radiotracer (name, date of birth and amount of activity).
      NOTE: The amount of activity administered to the patient is dependent on the patient’s body mass and can vary between institutions (in this protocol an amount of 3.2 MBq/kg is suggested).
   3. Ensure that the clinical information on the application form is correct by interviewing the patient. Ask the patient whether there were any recent relevant changes in treatment or medication.
   4. Ask the patient whether he or she has diabetes mellitus (DM). In case the patient has DM, ask whether he or she followed appropriate preparation (i.e., no administration of short working insulin less than 4 hours prior to the PET scan, or use of blood glucose lowering agents (such as metformin).
   5. Ask the patient whether he or she has any allergies or uses anticoagulants.
   6. Measure the patient’s blood glucose by applying a drop of blood obtained by pricking the side of the fingertip of the patient on a dedicated test strip (the serum glucose should not exceed 11.0 mmol/L).
   7. Explain the patient preparation and imaging procedures to the patient.
2. Administration of the radiotracer
   1. Secure venous access to the patient by inserting a peripheral venous cannula in one of the antecubital veins.
   2. Attach a three way stop cock system with Luer lock to a 20 mL syringe containing saline (this is the secondary syringe).
   3. Flush the three way stop cock system with saline (for the purpose of deaeration).
   4. Attach the three way stop cock with syringe to the end of the venous cannula.
   5. Check whether the venous cannula is patent by carefully flushing 10 mL of saline through the cannula (ask the patient whether he or she has any complaints during flushing).
   6. Attach the syringe containing the radiotracer (primary syringe) to the three way stop cock. Turn the valves of the three way stop cock so that the flow direction of fluid through the system runs from the syringe containing the radiotracer to the peripheral venous cannula. Administer the radiotracer by slowly pushing the plunger of the syringe (the syringe containing the tracer is placed in a special lead shielded container).
   7. Turn the valves of the three way stop cock in such a way that the syringe containing saline is connected to the primary syringe (that contained the radiotracer) and flush the syringe to rinse any residual radiotracer from the syringe.
   8. Turn the valves of the three way stop cock and push the plunger of the primary syringe to administer any residual radiotracer remaining in the syringe to the patient.
   9. Repeat step 1.2.7. and 1.2.8. three times.
   10. Turn the three way stop cock (to prevent backflow of blood from the patient’s vein) and detach the primary syringe. Attach a third syringe containing furosemide, turn the three way stop cock again and administer 0.5 g/Kg of furosemide (with a maximum amount of 10 mg) by pushing the plunger of the syringe. Remove the peripheral  venous cannula and apply pressure to the puncture site using a sterile bandage. Check whether there is no significant bleeding and from the puncture site and fix the bandage using medical tape.
3. Patient incubation
   1. Let the patient rest in a comfortable position, preferably in a dimly lit room, for 50 minutes.
   2. After 50 minutes, instruct the patient to void their bladder.
   3. At 55 minutes, escort the patient to the scanner and position the patient supine with the arms up on the scanner bed. Use appropriate arm support to make it as comfortable as possible for the patient. If the patient is not able to elevate his or her arms, scanning can be performed with the arms position alongside the patient.
   4. Observe the patient’s breathing pattern and secure the respiratory belt around the patient’s thorax (usually, the position just beneath the rib cage is optimal). Ensure that the sensor is placed at a location where abdominal wall excursion is identified after visual inspection (usually 5-7 cm from the midline). Secure the belt around the patient by using the Velcro-based closing system.
   5. Check on the scanner display whether the respiratory signal remains within bounds of the minimum and maximum range (if the respiratory signal is clipping, fasten or tighten the belt appropriately).
   6. Tip: Make sure the belt is fastened tight enough around the patient’s chest. Given that patients enter a more relaxed state after some time, the respiratory signal tends to drop (baseline drift of the signal). This prevents the signal from going out of bounds, thus maintaining a high quality of surrogate signal that is being used for respiratory gating.
   7. Start scanning at 60 minutes after incubation time.

2. Image acquisition and reconstruction

1. Protocol selection
   1. Select the whole-body protocol on the scanner. This can be done by moving the cursor over the appropriate protocol category (indicated by the circles next to the patient icon in the examination card), and click on the appropriate protocol (Figure 3).
   2. The ORG acquisition protocol will start with a scout scan (topogram) of the patient. To initiate acquisition of the topogram, press the scanner start key (yellow round key with a radiation sign) on the scanner control box (Figure 4). To halt or abort acquisition of the topogram, press the suspend or stop key respectively.
   3. Start by planning the PET bed positions on the topogram. This can be done by clicking the left mouse button on the topogram and setting the scan range.
   4. Select the bed positions which to be corrected for respiratory motion (Figure 5).
      NOTE: These are the ‘gated’ bed positions that cover the thorax. The ‘gated’ bed positions are recorded in listmode. Depending on the clinical indication, bed positions covering the upper abdomen can also be gated (for example when imaging is indicated for liver or pancreatic lesions). For the non-gated bed positions, it is only necessary to record the sinograms for image reconstruction.
   5. Set image recording time for the PET bed positions (Figure 5).
      NOTE: Depending on the amount of injected activity, scan duration of the non-gated bed positions has to be adapted to yield sufficient image quality. Additionally, the recording time of the non-gated bed positions in combination with the duty cycle being used for image reconstruction of the gated bed positions, the recording time of the gated bed positions is determined. For example, for a duty cycle of 35%, lengthening the scan by factor 3 yields approximately similar statistics for gated and non-gated bed positions. Suggested imaging protocol at the Radboud University Medical Center is a recording time for non-gated bed positions of 2 minutes, whilst for gated bed positions recording time is 6 minutes using a duty cycle of 35%
   6. After setting up the acquisition parameters, press and hold the start key (yellow round button with a radiation sign) on the scanner control box and wait until the scanner bed has moved back to the starting position. Press the start key again to acquire a low dose CT scan from the patient (head to feet). After acquiring the CT scan, press the start key to initiate the PET scan.
   7. During image acquisition, regularly check on the patient and the quality of the respiratory signal (adjust the respiratory belt if required).
      NOTE: Adjustment of the belt should only be performed when no respiratory gated bed positions are acquired. Therefore, adjustments should be done before or after these bed positions are acquired. Adjustment of the belt during acquisition of the gated bed position will affect the quality of the ORG images. Careful observation of the respiratory signal and possible adjustment of the respiratory belt before acquisition of the gated bed positions is required to counteract any significant baseline drifting of the signal during PET scanning.
2. Image reconstruction
   1. Review the respiratory signal that has been acquired and select the appropriate duty cycle for the gated bed positions (Figure 6).
      NOTE: The amplitude range used for respiratory gating is superimposed on the respiratory signal). Check for inconstancies or baseline drifts in the respiratory signal that can influence quality of the respiratory gating.
   2. Select image reconstruction protocol optimized for viewing (Figure 7). This is usually a high-resolution image reconstruction protocol with smaller voxel sizes for detection of small lesions. It is of importance to realize that the ORG algorithm will calculate the optimal amplitude range by using the entire respiratory signal of the selected bed positions. Though different duty cycles can be used for different bed positions (for example to correct for a varying quality respiratory signal), using different duty cycles for different bed positions is not advised given that this will introduce variations in image quality between different bed positions.
      NOTE: Here is an example image reconstruction protocol for viewing:
      • Algorithm: TrueX + TOF (UltraHD PET)
      • Number of iterations:3
      • Number of subsets: 21
      • Matrix size: 400 × 400
      • Post-reconstruction filtering, kernel (3D Gaussian), full width half maximum (FWHM): 3.0 mm
      • Duty cycle 35%
   3. Furthermore, reconstruct the PET images with an protocol compliant to the Research4Life (EARL) initiative for quantitative PET imaging. These are usually lower resolution images with specific post-reconstruction filtering applied.
      NOTE: Here is an example image reconstruction protocol for image quantification:
      • Algorithm: TrueX + TOF (UltraHD PET)
      • Number of iterations: 3
      • Number of subsets: 21
      • Matrix size: 256
      • Post-reconstruction filtering, kernel (3D Gaussian), full width half maximum (FWHM): 8.0 mm
      • Duty cycle 35%
   4. Send the reconstructed images to the PACS archive. The images are now ready to be evaluated by the nuclear medicine physician

Results

The use of ORG in PET results in an overall reduction of respiratory-induced blurring of the images. For example, in a clinical evaluation of patients with non-small cell lung cancer (NSCLC), ORG resulted in detection of more pulmonary lesions and hilar/mediastinal lymph nodes²⁰. This is readily demonstrated in Figure 8 and Figure 9, showing non-gated and ORG PET images of patients with NSCLC.

In particular, OR...

Discussion

In the nuclear medicine community, the deteriorating effects of respiratory motion artefacts in PET imaging have been well-recognized for a long time. It has been shown in many studies that the blurring effect of respiratory motion artefacts can significantly influence image quantification and lesion detectability. Although several respiratory gating methods have been developed, respiratory gating is currently not widely being used in clinical practice. This is particularly due to a resulting variable image quality, unac...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Sensor Port, sensor, black box, wave deck, elastic band, load cell sensor (complete set)	anzai medical co.	respiratory gating system AZ-733V	http://www.anzai-med.co.jp/en/product/item/az733v