New full-counts phase-matched data-driven gated (DDG) PET/CT

Abstract

Background:

Data-driven gated (DDG) PET has gained clinical acceptance and has been shown to match or outperform external-device gated (EDG) PET. However, in most clinical applications, DDG PET is matched with helical CT acquired in free breathing (FB) at a random respiratory phase, leaving registration and optimal attenuation correction (AC) to chance. Furthermore, DDG PET requires additional scan time to reduce image noise as it only preserves 35 to 50% of the PET data at or near the end-expiratory phase of the breathing cycle.

Purpose:

A new full-counts, phase-matched (FCPM) DDG PET/CT was developed based on a low-dose cine CT to improve registration between DDG PET and DDG CT, to reduce image noise, and to avoid increasing acquisition times in DDG PET.

Methods:

A new DDG CT was developed for three respiratory phases of CT images from a low dose cine CT acquisition of 1.35 mSv for a coverage of about 15.4 cm: end-inspiration (EI), average (AVG), and end-expiration (EE) to match with the three corresponding phases of DDG PET data: −10 to 15%; 15 to 30% & 80 to 90%; and 30 to 80%, respectively. The EI and EE phases of DDG CT were selected based on the physiological changes in lung density and body outlines reflected in the dynamic cine CT images. The AVG phase was derived from averaging of all phases of the cine CT images. The cine CT was acquired over the lower lungs and/or upper abdomen for correction of misregistration between PET and FB CT as well as DDG PET and FB CT. The three phases of DDG CT were used for AC of the corresponding phases of PET. After phase-matched AC of each PET dataset, the EI and AVG PET data were registered to the EE PET data with deformable image registration. The final result was FCPM DDG PET/CT which accounts for all PET data registered at the EE phase. We applied this approach to 14 ¹⁸F-FDG lung cancer patient studies acquired at 2 min/bed position on the GE Discovery MI (25-cm axial FOV) and evaluated its efficacy in improved quantification and noise reduction.

Results:

Relative to static PET/CT, the SUV_max increases for the EI, AVG, EE, and FCPM DDG PET/CT were 1.67±0.40, 1.50±0.28, 1.64±0.36 and 1.49±0.28 respectively. There were 10.8% and 9.1% average decreases in SUV_max from EI and EE to FCPM DDG PET/CT, respectively. EI, AVG, and EE DDG PET/CT all maintained increased image noise relative to static PET/CT. However, the noise levels of FCPM and static PET were statistically equivalent, suggesting the inclusion of all counts was able to decrease the image noise relative to EI and EE DDG PET/CT.

Conclusions:

A new FCPM DDG PET/CT has been developed to account for 100% of collected PET data in DDG PET applications. Image noise in FCPM is comparable to static PET, while small decreases in SUV_max were also observed in FCPM when compared to either EI or EE DDG PET/CT.

INTRODUCTION

PET/CT data are normally acquired in free breathing (FB) due to the limited system sensitivity of the PET scanner. The influence of respiratory motion is unavoidable, posing a challenge for quantification, localization, and treatment response assessment of the functional PET data.¹ Artifacts from respiratory motion are blurry tumor appearance in PET images and/or misregistration between PET and FB CT. Attempts to acquire PET data in deep inspiration breath-hold (DIBH)^2–5 gained little traction in the clinic due to patient compliance issues and the challenging selection of patients to benefit from DIBH. To acquire CT at mid-expiration⁶ or end-expiration^7,8 breath hold to match with (static or average) PET also has mixed results. Coaching a patient to breath-hold during the short CT scan may be complicated by patient compliance and operator dependence before and during the CT scan.^6,9 In addition, many patients are covered by a warm blanket due to a low temperature setting in the PET/CT scanner room, so coaching patients to breath-hold at mid-expiration or end-expiration can be difficult. In our clinic, most patients breathe shallowly without any breathing instructions and as a result, most of the PET/CT scans are of good quality albeit some do suffer from either motion blur in PET data and/or misregistration between PET and FB CT.

Data-driven gated (DDG) PET is a promising technique to reduce motion blur in PET. It can derive the respiratory signal based on the calculations of the center of mass,¹⁰ principal component analysis (PCA)¹¹ or spectral analysis¹² of a series of short-time frames of dynamic PET data for respiratory-gated PET. DDG PET removes the major issues of patient selection and long setup time associated with external device-gated (EDG) PET or 4D PET.¹³ DDG PET has been shown to outperform EDG PET due to a higher failure rate in identifying the end-inspiration (EI) triggers in EDG PET.^14,15 Despite the encouraging results of DDG PET, there is still misregistration between DDG PET and FB CT just like there is between static PET and FB CT.^14,16 The CT scan in PET/CT is typically a helical acquisition obtained under FB, while DDG PET data is usually derived from the expiratory phase of the breathing cycle. There is still a chance of misregistration between DDG PET and the FB CT for attenuation correction (AC) and localization. This issue has been previously reported in studies on DDG PET, as misregistration between DDG PET and helical CT can occur in certain cases.^14,17 The use of incorrect CT phases for AC can lead to a significant change in PET quantification, with the impact more severe for DDG PET than static PET.^18,19

At this time, the only approach to providing phase-matched CT for DDG PET is 4D CT, which requires a respiratory monitoring device (RMD).^8,20 Combining DDG PET and 4D CT is impractical because most patients are covered by a warm blanket in PET/CT imaging, preventing direct contact between the skin and an RMD such as the optical Real-time Position Management device (Varian Medical Systems, Palo Alto, California, USA) on the patient’s abdomen.¹³ Use of the AZ-733V strain gage (Anzai Co., Tokyo, Japan) or the air bellows belt (Philips Health Care, Andover, MA, USA) can improve the contact; however, it requires a technologist to wrap the device around the patient’s waist, increasing radiation exposure to the technologist, and decreasing the acceptance of EDG or DDG PET in the clinic.

Limited clinical acceptance of EDG PET was primarily supported by the study of 130 patients comparing EDG PET embedded in or separated from a whole-body PET scan, in which challenges were reported including long acquisition and reconstruction times and requiring a multidisciplinary team with a thorough understanding and execution of the process, from patient prep through scan acquisition.²¹ This can also be supported by the use of DDG PET over EDG PET to maintain a simple and “practical” implementation in the clinic.²² Putting or wrapping a respiratory monitoring device on the patient and increasing radiation exposure to the technologist were part of the patient prep and both limited the application of EDG PET.^21,23,24

We have designed a new DDG CT to match with DDG PET.^17,25 The new DDG CT was derived from a low-dose cine CT scan of about 1.35 mSv for a coverage of about 15.4 cm.¹⁷ In clinical practice, addressing significant PET/CT misregistration often involves rescanning the patient over a limited range of one to two PET bed positions with PET/CT, which typically yields better registration. However, in our approach, rather than performing a repeat PET/CT, we acquire only a cine CT over the area of misregistration and without any repeat PET. When comparing cine CT to a repeat PET/CT, cine CT demonstrates a significantly lower scan coverage (15.4 ± 4.7 cm compared to 32.5 ± 11.5 cm; P < .001) and a reduced effective radiation dose (1.3 ± 0.6 mSv compared to 3.7 ± 2.6 mSv; P < .01).^17,25 Although the CT radiation dose to the patient was increased, the low dose cine CT could save 50% CT radiation dose and several minutes of PET scan time compared to the repeat PET/CT scan protocol for misregistration correction.¹⁷ Both average CT and DDG CT can be derived from the low-dose cine CT to correct for static PET and DDG PET, respectively.^17,25

DDG PET has been shown to have a higher standardized uptake value (SUV) and smaller uptake volume for tumors than static PET data.²⁶ However, DDG PET increases image noise as it is made up of only 35 to 50% of the PET data at or near the EE phase, or the so-called quiescent phase of respiration for best image quality.^14,15 The purpose of this work was to design a new full-counts, phase-matched (FCPM) DDG PET/CT to provide optimal AC of each DDG PET phase and improve the noise characteristics of DDG PET relative to traditional EE phase-based DDG PET methods.

MATERIALS AND METHODS

Patient Data

This study was approved by the institutional ethics review board at the University of Texas, M.D. Anderson Cancer Center. Fourteen patients whose PET/CT data showed a misregistration of at least one lung tumor between CT and PET were recruited in the study. Determination of misregistration was made by the technologist before the completion of the PET scan after surveying the quality of the PET/CT fusion. In the case of misregistration, a cine CT scan (detailed later) for misregistration correction was acquired before releasing the patient from the PET/CT scan. This workflow is only used for misregistration correction. All patients were injected with ~370 MBq ¹⁸F-FDG and scanned on a 5-ring GE Discovery MI PET/CT scanner with a 25 cm axial field of view (AFOV).²⁷ There were 14 lung cancer patients recruited in the study, yielding a total of 25 tumors for SUV analysis.

PET/CT Scan Protocol with Misregistration Correction

The helical CT scan protocol was 120 kVp, 0.984 pitch, 0.5 sec gantry rotation time, 64×0.625 mm x-ray collimation, and tube current modulation (TCM) based on anterior to posterior (AP) scout view at noise index = 30, maximum mA = 560 and minimum mA = 60 and 100 for without and with injection of iodinated contrast, respectively. The choice of AP scout was to account for the patient normally positioned lower than the iso-center of the gantry to avoid claustrophobia to the patient. As a result, the patient size appeared smaller than when the patient was positioned higher for TCM, which lowers radiation exposure to the patient due to the smaller patient size in the AP scout. If CT contrast is used, the helical CT scan will start 56 sec after injection of 100 cc of iohexol (Omnipaque-300) or iodixanol (Visipaque-320). The PET acquisition time per bed position during the scan of the torso was 2 min for body mass index (BMI) < 35, 2.5 min for BMIs ≥ 35 but < 40, and 3 min for BMIs ≥ 40. The overlap between two consecutive bed positions was 25 slices or 28%. When the scan reached the legs, acquisition time was reduced to 1.5, 2, and 2 min for BMIs of < 35, 35 to 40, and ≥ 40, respectively. The scan time per bed position was determined in part by an American College of Radiology (ACR) phantom acquisition of a simulated 70 kg patient and 370 MBq ¹⁸F-FDG injection and patient data.²⁷

To correct misregistration between PET and CT data right after the PET scan, a cine CT scan of less than 1 min acquisition time for about 14–16 cm axial coverage over the misregistration area was acquired under FB. The cine CT protocol was 120 kVp, 0.8 sec gantry rotation time, 8×2.5 mm x-ray collimation, TCM of noise index = 70, maximum mA = 20, and minimum mA = 10, and 5-sec cine CT duration, selected to cover 97.5% of the normal respiration rates of patients age > 65 years.²⁸ The scan coverage of cine CT was a multiple of 20 mm (8×2.5 mm) x-ray collimation. The cine CT scan has been shown to improve registration of PET and CT data without requiring a repeat PET scan and at 50% of the CT radiation dose relative to repeat PET/CT.¹⁷

DDG PET/CT

DDG PET on the GE PET/CT calculates the weighting factors of the first three principal components in each sinogram (projection data) of about 0.3 sec for a single bed-position acquisition.^11,29 The weighting factors were Fourier transformed for the peak value in the frequency range of 0.1 to 0.4 Hz corresponding to the breathing cycles of 2.5 to 10 sec. The ratio between the peak value to the mean value is defined as the strength of the signal being respiratory-like. The larger the number, the more the signal is respiratory-like, and vice-versa. Respiratory gating can be prospectively activated once the strength exceeds a set threshold or retrospectively activated after data acquisition. The respiratory-like signal in the frequency domain is then converted to the time domain. A peak detection algorithm is then applied to derive the EI triggers from each cycle.^11,30 Real-time detection of the strength of respiratory-like signal allows for prospective gating and extends the acquisition time to compensate for the loss of coincidence counts outside the quiescent or near the EE phase of the respiration during gating. Retrospective rather than prospective gating was used in this study.

DDG CT was developed in-house to process the cine CT data for the three respiratory phases of CT images:²⁵ average (AVG), end-inspiration (EI), and end-expiration (EE). The AVG CT of each slice location was averaged from all the cine CT images at the same slice location. For the EI and EE phases, after segmentation of the lung and body regions, both the density from average Hounsfield units (HU) in the lung regions and the body outline of each cine CT image were derived from the cine CT images including the lung region, the EE and EI phases were chosen based on the largest and smallest density in the lungs, respectively. From the cine CT images not including the lung region, the largest and smallest expansions of the body outline were selected for the EE and EI phases, respectively. A consistency check was performed to ensure the images of both the EE and EI phases in collimation of 20 mm x-rays or 8 slices of 2.5 mm were acquired at the same time. This DDG CT was implemented on a server to provide DDG CT to the GE Discovery PET/CT scanners on the hospital network (4 Discovery MI of 25 cm, 2 Discovery DR of 15 cm, and 1 Discovery 690 of 15 cm). The server can serve more than 7 GE PET/CT scanners and it was constructed on an Ubuntu desktop computer, which can provide the DDG CT images in 3 min to the requesting scanner. The PET data for AC with the DDG CT images can be performed retrospectively during the scan of the next patient without any impact on the performance of the PET/CT scanners.

Full Counts DDG PET/CT

A workflow diagram of the methodology for creating FCPM DDG PET/CT is outlined in Figure 1. The three phases of EI, AVG and EE of DDG CT were used for AC of 25% (−10 to 15%), 25% (15 to 30% and 80 to 90%), and 50% (30 to 80%) DDG PET data, respectively. The negative sign (−) in the percentage indicates that this portion of data was derived from the preceding respiratory cycle. After AC of each PET dataset, the EI and AVG PET data were deformed and summed to the EE PET data with a deformable image registration (DIR). The DIR method was based on non-local spatial-temporal constraints with group-wise non-rigid-registration (NRR)³¹ in the GE research toolbox. For 3 PET phases ${\left\{I_k\right\}}_{k=1}^3$ defined on a vector space Ω and reference image μ, the deformation fields ${\left\{u_k\right\}}_{k=1}^3$ were optimized to minimize the following objective function E:

$$E\left[\mu ,\mathit{u}\right]=\sum _{k=1}^3{\int }_{\mathrm{\Omega }}{\left|I_k\left(.+u_k\right)-\mu \right|}^{2}dx+{\lambda }_s{\int }_{\mathrm{\Omega }}{\int }_{\mathrm{\Omega }}w\left(x,y\right){\left|u_k\left(x\right)-u_k\left(y\right)\right|}^{2}dx\mathrm{ }dy+{\lambda }_t{\int }_{\mathrm{\Omega }}{\int }_{\mathrm{\Omega }}\widehat{w}\left(x,y\right){\left|v_k\left(x\right)-v_k\left(y\right)\right|}^2\mathrm{ }dx\mathrm{ }dy,$$

where ${ I}_k\left(.+u_k\right)$ is the image $I_k$ after deformation field $u_k.$ Both ${\lambda }_s$ and ${\lambda }_t$ are scalars that weight the two regularization terms. The first regularization term following ${\lambda }_s$ is a spatial penalty, on the motion fields ${\left\{u_k\right\}}_{k=1}^3$, which captures correlations in the motion field between non-local regions. The second regularization term following ${\lambda }_t$ is a temporal penalty which captures the non-local spatial correlations of velocity $v_k=u_{k+1}-u_k$. It captures temporal trends of motion across non-local pixels which would help NRR to reduce the impact of noise, large motion, low contrast etc. The temporal penalty is less restrictive on the motion fields than the spatial penalty since only spatial variations in velocity are penalized. $w(x,y)$ and $\widehat{w}(x,y)$ are weight functions to set the active pixels y correlated with the current pixel x.

Figure 1.

Flowchart of the proposed method for full-counts phase-matched DDG PET/CT

The result was a registered, full counts, phase matched (FCPM) DDG PET/CT at the EE phase that accounts for all PET data. This approach was applied on all 14 lung cancer patients with a total of 25 tumors. A 3-cm diameter volume of interest (VOI) in the liver was used to assess if FCPM DDG PET/CT could achieve a comparable noise level to the baseline (BL) static PET/CT and a comparable performance of DDG PET in quantification. Another 3-cm VOI, located near the lesions, was selected as the background reference for the calculation of Contrast-to-Noise Ratio (CNR) and Signal-to-Noise Ratio (SNR).

Statistical Analysis

Statistical analysis on the SUV and noise of the different PET/CT datasets was conducted with one-way ANOVA including the Geisser-Greenhouse correction with GraphPad Prism 10.0.3.

RESULTS

Figures 2 and 3 show two patient study examples. In both patients, all four DDG PET/CT methods improved registration as expected. Both also showed a consistently lower SUV_max for AVG and FCPM DDG PET/CT than for EI and EE DDG PET/CT. This is expected as there was more motion unaccounted for in the AVG phase reconstruction relative to the EI and EE phases. The presence of the liver dome in AVG and EE phases of DDG CT in the last three columns of Figure 3 shows a limited coverage in the cine CT scan for misregistration correction of the targeted tumor. This discontinuity in anatomy is normal when the CT for AC needs to be as large as the PET in the scan range, and the cine CT of a limited coverage needs to borrow the FB CT images outside of the cine CT coverage for AC of the full PET dataset. Figure 4 illustrates the comparison in SUV, CNR, and SNR ratios between the various PET/CT methods. Relative to static PET/CT, SUV_max ratios for the other four PET/CT reconstructions were 1.65±0.41, 1.50±0.28, 1.65±0.37 and 1.50±0.29 (mean ± standard deviation) for the EI, AVG, EE and FCPM DDG PET/CT, respectively (Figure 4a). The results suggested that motion and misregistration corrections increased SUV_max for all four DDG methods, but there were distinctions between some of the methods. Combining all three phases of EI, AVG, and EE to FCPM DDG PET/CT decreased SUV_max relative to the EI, AVG, and EE phases by 10.1%, 0.4%, and 10.1%, respectively. FCPM DDG PET/CT was lower in SUV_max when compared to the EI and EE phases of DDG PET/CT (both statistically significant) and about the same as the AVG DDG PET/CT (not statistically significant). Additionally, the FCPM method achieved a substantial 60–70% improvement in CNR and SNR, as illustrated in Figures 4b and 4c.

Figure 2.

An example of (from left to right columns) the baseline (BL) static PET/CT, EI, AVG, EE, and FCPM DDG PET/CT. The top row is the fusion of PET and CT showing the SUV_max of a lung tumor, and the bottom row is PET only. A 3-cm volume of interest is in both fusion and PET-only images for noise measurement. The standard deviations in SUV are shown in the PET-only images.

Figure 3.

An example of (from left to right columns) the baseline (BL) static PET/CT, EI, AVG, EE, and FCPM DDG PET/CT. The top row is the fusion of PET and CT showing the SUV_max of a lung tumor, and the bottom row is PET only. A 3 cm volume of interest is in both fusion and PET-only images for noise measurement. The standard deviations in SUV are shown in the PET-only images. The presence of the liver dome in the AVG and EE phase of DDG CT in the last three columns shows a limited coverage in the cine CT scan for misregistration correction of the targeted tumor.

Figure 4.

The SUV_max, CNR, and SNR ratios of end-inspiration (EI), average (AVG), end-expiration (EE), and full-counts phase-matched (FCPM) to the baseline (BL) static PET/CT. All four methods passed the Kologorov-Smirnov normality test of α=0.05. Statistically significant differences for paired comparisons with the FCPM DDG PET/CT were selectively indicated (****P< 0.0001, ***0.001, **P<0.01, *P<0.1).

Background noise measurements in the liver for all PET/CT methods are in Table 1. The summary statistics were 0.17±0.05, 0.3±0.11, 0.31±0.10, 0.23±0.07 and 0.17±0.05 for BL, EI, AVG, EE, and FCPM DDG PET/CT, respectively. Figure 5 illustrates the noise comparisons along with their statistical significance. There was no statistical difference between BL PET/CT and FCPM DDG PET/CT or between EI and AVG DDG PET/CT in the noise level. For all patients, FCPM DDG PET/CT had about the same noise level as the static PET/CT (within 0.01 SUV), suggesting that both FCPM DDG PET/CT and static PET/CT had a similar image noise level. However, there were significant differences (P< 0.0001) in the noise level between any of EI, AVG, and EE DDG PET/CT and either static PET/CT or FCPM DDG PET/CT; and between EI (or AVG) and EE phases of DDG PET/CT (P=0.001 for EI and P<0.0001 for AVG). These results are consistent with the EE phase of using more data (50%) than either the EI or AVG phase (both 25%). There was no statistical difference between EI and AVG phase of DDG PET/CT (P>0.9999).

Table 1 –

Noise measurements of the liver in a 3-cm diameter volume of interest.

Patient	Baseline	End-Inspiration	Average	End-expiration	Full-counts
1	0.10	0.21	0.16	0.14	0.10
2	0.18	0.34	0.28	0.24	0.18
3	0.18	0.31	0.28	0.23	0.18
4	0.27	0.62	0.56	0.39	0.28
5	0.14	0.21	0.26	0.20	0.13
6	0.10	0.22	0.24	0.13	0.10
7	0.25	0.34	0.41	0.3	0.25
8	0.13	0.25	0.25	0.19	0.14
9	0.15	0.28	0.28	0.19	0.16
10	0.14	0.25	0.25	0.19	0.14
11	0.20	0.37	0.42	0.31	0.19
12	0.13	0.21	0.3	0.19	0.12
13	0.20	0.37	0.35	0.28	0.22
14	0.14	0.26	0.27	0.21	0.14

BL PET/CT and FCPM DDG PET/CT are not statistically significant (P>0.9999), and so are the DDG PET/CT of EI and Average (P=0.9999). FCPM DDG PET/CT is statistically significant from EI, AVG, or EE DDG PET/CT (P<0.0001), so is BL PET/CT different from EI, AVG, or EE DDG PET/CT (P<0.0001). ‘Baseline’ denotes ‘baseline (BL) static PET/CT’; ‘ End-Inspiration’, ‘Average’, ‘End-expiration’ and ‘Full-counts’ denote End-Inspiration (EI), Average (AVG), End-expiration (EE), and Full-counts phase-matched (FCPM) DDG PET/CT.

Figure 5.

The noise in a 3-cm volume of interest of baseline (BL) static PET/CT, and four DDG PET/CT of full-counts phase-matched (FCPM), end-inspiration (EI), average (AVG), end-expiration (EE). There was no statistical (ns) difference between BL and FCPM and between EI and AVG. However, there were statistically significant differences (****P< 0.0001, ***P<.001) between any of EI, AVG, EE and either BL or FCPM.

DISCUSSIONS

We used 25% (−10 to 15%), 25% (15 to 30% and 80 to 90%) and 50% (30 to 80%) of the total PET data for the EI, AVG and EE phases of DDG PET/CT in this study to demonstrate the feasibility of FCPM DDG PET/CT to achieve a similar noise level of static PET/CT and an improved registration of DDG PET/CT in tumor quantification. Prior studies on matched PET/CT reconstructions have focused more on CT than PET to avoid any additional radiation dose from 4D CT as in 4D PET/CT.¹³ Lu et al. proposed a respiratory motion compensated PET/CT with the motion information derived from matched attenuation-corrected gated PET data, which first found a FB-CT matched PET phase as reference and deformed the other phases of PET data to the reference PET phase for full counts according to the correlation of internal organ motion in PET and external respiratory signals.³² Hamill et al. improved alignment of PET and CT in PET/CT in cases of misregistration based on the respiratory signals collected in both PET and CT acquisitions.³³ Our approach used no external gated device and focused on finding a CT from the cine CT to match with PET and leveraged the successful application of DDG PET available from multiple vendors,^17,25 which can turn every PET scan into a gated PET with a penalty of increased noise if the PET acquisition time remains the same. To our knowledge, FCPM DDG PET/CT was the first DDG PET/CT (DDG PET and DDG CT) to include all PET counts without any external gated device. The selection of 50% was based on the observation of overall registration in DDG PET/CT from the default 50% (30 to 80%) DDG PET and the EE phase of DDG CT in our clinical data.^17,25,26 The division of the remaining 50% of PET data to 25% each for EI and AVG could be improved in a future study by including analysis of the respiratory waveform derived from DDG PET. It was evident that SUV_max was reduced by FCPM DDG PET/CT when compared to either EI or EE phase DDG PET/CT because noise reduction would reduce SUV_max, based on the measurement of a single voxel. Use of SUV_peak may not show the decreases of 10.8% and 9.1% in SUV_max from EI and EE to FCPM DDG PET/CT, respectively.³⁴

All patient data have some degree of irregular respiration in FB. There was no noticeable degradation in the performance of FCPM DDG PET/CT by irregular respiration in this study. DDG PET data was averaged from 2 min of data, which spans 30 to 40 respiratory cycles of 3 to 4 sec, and which is not likely to be impacted by some irregular respiratory cycles. However, DDG PET will not work if there is no periodicity in the respiration. On the other hand, an irregular respiratory cycle could have a significant impact on DDG CT, derived from a 5-second cine CT scan. We have reported that the EE or quiescent phase of DDG CT was less impacted by irregular respiration than the average CT.²⁵ As we spend more time at EE than at EI during respiration and the EE position is more stable than the EI position, DDG CT at EE should outperform DDG CT at EI. If irregular respiration renders DDG CT at EI or average CT less effective than DDG CT at EE, we could fall back to DDG PET/CT at EE (DDG PET at EE and AC by DDG CT at EE). As a reminder, all our data were collected when an irregular respiration occurred during helical CT of PET/CT which caused misregistration between PET and CT and were rectified by DDG CT.^17,25

FCPM DDG PET/CT could also be applied to cardiac PET/CT. Cardiac DDG PET imaging based on the heart motion during respiration has been investigated on PET/MR in simulations, phantom studies, and clinical patients with or without cardiac gating. Four-bin amplitude-based gating plus motion correction was able to improve the uptake ratio between the myocardium and background and maintain similar noise levels as the non-gated study.³⁵ Cardiac PET/CT normally requires two separate imaging of rest and stress states. Although rest imaging is similar to ¹⁸F-FDG oncological imaging, respiratory motion in stress imaging could be impacted by the introduction of stress agents. In a randomized study of 48 patients, dipyridamole was better than adenosine as a stress agent for more uniform respiration and resulted in a higher frequency of successful respiratory gating and superior image quality.³⁶

CONCLUSIONS

A new FCPM DDG PET/CT has been developed to correct for tumor motion without misregistration between CT and PET, while also maintaining a similar noise level to static PET/CT without an increase in scan time.