Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Apr 1.
Published in final edited form as: Int J Radiat Oncol Biol Phys. 2020 Nov 10;109(5):1638–1646. doi: 10.1016/j.ijrobp.2020.11.014

New Data-Driven Gated Positron Emission Tomography and Computed Tomography Free of Misregistration Artifacts

Tinsu Pan *, Yang Lu †, M Allan Thomas *, Zhongxing Liao ‡, Dershan Luo §
PMCID: PMC7965243  NIHMSID: NIHMS1652435  PMID: 33186619

Abstract

Purpose:

We developed a new data-driven gated (DDG) positron emission tomography (PET)/computed tomography (CT) to improve the registration of CT and DDG PET.

Methods:

We acquired 10 repeat PET/CT and 35 cine CT scans for the mitigation of misregistration between CT and PET data. We also derived end-expiration phase CT as DDG CT for attenuation correction of DDG PET. Radiation exposure, body mass index (BMI), scan coverage, and effective radiation dose were compared between repeat PET/CT and cine CT. Of the 35 cine CT patients, 14 (capturing 59 total tumors) were compared among average PET/CT (baseline PET attenuation correction by average CT), DDG PET (DDG PET attenuation correction by baseline CT), and DDG PET/CT (DDG PET attenuation correction by DDG CT) for registration and quantification without increasing the scan time for DDG PET.

Results:

Compared with repeat PET/CT, cine CT had significantly lower scan coverage (32.5 ± 11.5 cm vs 15.4 ± 4.7 cm; P < .001) and effective radiation dose (3.7 ± 2.6 mSv vs 1.3 ± 0.6 mSv; P <.01). Repeat PET/CT and cine CT did not differ significantly in BMI or radiation exposure (P > .1). Cine CT saved the scan time for not needing a repeat PET. The SUV ratios of average PET/CT, DDG PET, and DDG PET/CT to baseline PET/CT were 1.14 ± 0.28, 1.28 ± 0.20, and 1.63 ± 0.64, respectively (P < .0001), suggesting that the SUVmax increased consecutively from baseline PET/CT to average PET/CT, DDG PET, and DDG PET/CT. Motion correction with DDG PET had a larger impact on quantification than registration improvement with average CT did. The biggest improvement in quantification was from DDG PET/CT, in which both registration was improved and motion was mitigated.

Conclusion:

Our new DDG PET/CT approach alleviates misregistration artifacts and, compared with DDG PET, improves quantification and registration. The use of cine CT in our DDG PET/CT method also reduces the effective radiation dose and scan coverage compared with repeat CT. Published by Elsevier Inc.

Introduction

The clinical use of positron emission tomography (PET)/computed tomography (CT) can be diminished by respiratory motion–related degradation artifacts such as tumor blurring1 and misregistration between CT and PET data.2 Tumor blurring and misregistration can both compromise tumor quantification, and misregistration can also make it difficult to assess nodal involvement.3 This is an especially relevant issue in the application of 68Ga-DOTATATE for neuroendocrine tumor imaging due to high uptake in the liver and the spleen, which are likely targets of misregistration.

The conventional approach to reduce tumor blurring due to respiration is to use external device-based gated (EDG) PET/CT. In EDG PET/CT, the PET data are binned into multiple phases of respiration according to a respiratory signal produced by an external sensor that tracks the patient’s breathing.4,5 Several issues impede the widespread clinical application of this approach, however. First, EDG PET/CT requires a gating device. Although some PET/CT scanners have gating devices, such devices are not often used owing to the effort required to set them up and the uncertainty regarding which patients can benefit from EDG PET/CT. Second, the approach requires a technologist to physically place, on or around the waist of the patient, an infrared-reflector box, strain-gage sensor, or pneumatic belt bellows that interfaces with the gating device. This process exposes the technologist to additional radiation. Third, the approach requires an additional 10 minutes of PET scan time in addition to the 20 to 25 minutes required for a whole-body PET/CT scan.6 Fourth, it may require EDG CT7 to match with EDG PET, an essential step for radiation therapy treatment planning applications of PET/CT (but not for diagnostic imaging owing to concerns about radiation exposure).

In contrast to EDG PET, data-driven gated (DDG) PET uses principal component analysis to extract a respiratory signal from the PET data.8 DDG PET, which was recently commercialized by GE Healthcare,9 outperformed EDG PET in a study of 144 patients.10 The DDG PET method was able to retain 50% of the full PET data by identifying and then selecting the end-expiratory, or quiescent, phase of respiration for specific PET data extraction. Quiescent phase gating was proposed to increase the statistics of the gated PET data in the expiratory phase, as the exhalation duration is normally longer than the inhalation duration.11,12

Compared with EDG PET, DDG PET offers an improved workflow without any setup time and less radiation exposure to the technologist. Furthermore, DDG PET can be applied to all patient data either prospectively or retrospectively, eliminating the difficulty in assessing what patients may benefit from EDG PET. However, maximizing the potential of DDG PET still requires a registered CT. Registration between CT and gated PET can be improved by matching breath-hold CT data captured near the end-expiratory phase with PET data also captured in the quiescent phase.11 Additional methods include using rigid or deformable registration to match PET data with CT data13,14 or matching the correct portion of the PET data with CT data.15,16

The realization of DDG PET will make it impractical to use EDG CT to match PET and CT data. The quantification of PET without an associated CT could be achieved by applying maximum likelihood activity and attenuation reconstruction17,18 or by using machine learning to generate a pseudo CT for PET data attenuation correction.19 Approaches used without a CT or with a pseudo CT for PET quantification may underestimate the importance of CT in the assessment of nodal involvement, which relies on accurate registration between CT and PET.

Free-breathing CT does not represent a specific respiratory phase, but rather a combination of all respiratory phases repeated throughout the duration of the CT scan. Each CT slice corresponds to a specific respiratory phase, but a stack of CT slices may not correspond to any respiratory phase. A practical way to overcome clinically relevant PET/CT misregistration is simply to rescan the patient in a limited scan range PET/CT. Often, the repeat scan has better registration, and the problem is resolved. Misregistration is mostly a CT, rather than PET, problem. This is because CT images are produced from subseconds of data, which can represent any phase of respiration, whereas PET images are produced from minutes of data, which represent an average respiration. Repeat PET in a repeat PET/CT is therefore unnecessary and may cause patient fatigue or discomfort. A repeat PET/CT may also prolong the uptake time of the next patient, complicating clinical workflows and patient throughput.

Herein, we propose a new DDG PET/CT approach that uses DDG PET9 and accounts for the need of a repeat CT scan to optimize the clinical use of PET/CT. In this approach, instead of acquiring a repeat PET/CT, we acquire a cine CT over the misregistration area to produce 2 key components: (1) an average CT for attenuation correction of the PET data and (2) a DDG CT for attenuation correction of the DDG PET. We applied this novel technique to correct for misregistration between CT and PET around the diaphragm, a common site of misregistration artifacts. Our results demonstrate for the first time that, without prolonging the PET scan time for DDG PET, cine CT can result in a lower radiation dose to the patient than the use of a repeat PET/CT, and DDG PET/CT can achieve better motion correction than DDG PET.

The novel DDG CT design outlined in this work has specific potential applications in radiation therapy. The majority of PET/CT scanners are associated with diagnostic imaging and therefore are without a respiratory gating device. The new DDG PET/CT method described in this work makes it feasible for radiation therapy to use such a PET/CT scanner to simulate a patient when gating is deemed necessary for tumor motion corrections and misregistration between CT and PET. Average CT has already been adopted in dose calculation for radiation therapy planning20 and in image alignment with cone beam CT in image guided radiation therapy.21 In addition, maximum-intensity projection CT images from cine CT provide the tumor motion information for delineation of internal gross tumor volumes.22,23 All of these benefits can be derived from the same cine CT scan for the new DDG PET/CT without any hardware respiratory gating.

Materials and Methods

Patient selection

The study was approved by an institutional review board. The need for informed written consent was waived. Ten repeat PET/CT scans and 35 cine CT scans were collected for the mitigation of misregistration between CT and PET data. The cine CT scan protocol was implemented at our institution as an alternative option for PET/CT cases involving clinically significant misregistration. It saves overall scan time because a repeat PET scan is not needed because CT, not PET, is the cause of misregistration. As such, the 10 repeat PET/CT scans that were analyzed in this work occurred before the cine CT protocol was implemented. The repeat PET/CT and cine CT scans analyzed all came from different patients, as no patient underwent both repeat PET/CT and cine CT.

PET/CT acquisition protocol

The scans were acquired using 4 different GE Discovery PET/CT scanners: D690, D710, DR, and DMI. The injection activities were targeted at 370 MBq of 18F-FDG and 185 MBq of 68Ga-DOTATATE for general and neuroendocrine tumor imaging, respectively. On the DMI scanner, the PET scan times per axial field of view (AFOV) were 2 minutes for patients whose body mass index (BMIs) were <35, 2.5 minutes for patients whose BMIs were ≥35 but <40, and 3 minutes for patients whose BMIs were ≥40. On the D690, D710, and DR scanners, the PET scan times per AFOV were 2 minutes for patients whose BMIs were <25, 3 minutes for patients whose BMIs were ≥25 but <40, and 4 minutes for patients whose BMIs were ≥40. The PET scan time was not increased for DDG PET.

The whole-body PET/CT and repeat PET/CT protocols were identical except that the repeat PET/CT scans were limited to the axial extent relevant to the misregistration. The same CT protocol was used for all CT scanners: 120 kVp, pitch factor 0.984, gantry rotation time 0.5 sec, x-ray collimation 64 × 0.625 mm, noise index = 30, maximum mA = 560, and minimum mA = 60 without and minimum mA = 100 with iodinated contrast injection. The repeat CT scan coverage was 15.4 cm for the first AFOV and an additional 11.8 cm for each additional AFOV on the D690, D710, and DR scanners, and then 24.9 cm for the first AFOV and an additional 17.9 cm for each additional AFOV on the DMI scanner. The cine CT scan protocol was 120 kVp, 5 seconds cine scan duration, gantry rotation time 0.8 seconds, x-ray collimation 8 × 2.5 mm, noise index = 70, minimum mA = 10, and maximum mA = 20. The 5-second cine duration was chosen to cover 97.5% of the normal respiration rate of patients age >65 years.24 The scan coverage of cine CT was a multiple of 2 cm (or x-ray collimation of 8 × 2.5 mm).

Average CT and DDG CT

The cine CT data were used to derive an average CT, which was then used for attenuation correction of the whole-body PET data.2,25 End-expiration (EE) CT, or DDG CT, was also derived from the same cine CT data and then used for attenuation correction of the DDG PET data. DDG PET data were derived from 50% of the PET data in the quiescent phase at 30% offset from the end-inspiration phase of each respiratory cycle.9 Because it contained only 50% of the PET data, the DDG PET data were noisier than the whole-body PET data.

The derivation of EE CT for DDG CT from the cine CT data was achieved as follows. The cine CT data of a 2-cm axial extent at the diaphragm level was reviewed to identify 1 of the 7 phases of the cine CT data as the initial EE CT. This identification was based on the property of highest CT Hounsfield units in the lungs due to compression of the lung tissues at end-expiration or the largest liver presence at the diaphragm level. Once identified, the EE CT at the diaphragm level was combined with the CT images at the neighboring superior and inferior locations based on a data consistency or cross-correlation criterion between the EE CT images and the neighboring CT images. The most similar CT images to the original EE CT images were then added, increasing the spatial extent of the total image set identified as EE CT. This process was repeated until all the cine CT images at the neighboring superior and inferior locations were correlated for inclusion to the EE CT images. The cross-correlation criterion used in this process came from an automatic 4D CT sorting algorithm by Carnes et al.26

Quantitative measurements

Of the 35 cine CT patients, 14 were compared for registration and quantification—the other cine CT patients either did not have tumor in the misregistration area or were acquired before the availability of DDG PET. These 14 cases captured 59 total tumors in 10 liver cancer patients and 4 lung cancer patients. Of the 14 patients, 9 were injected with 18F-FDG and 5 were injected with 68Ga-DOTATATE. We used MIM version 6.9.6 (MIM Software Inc, Beachwood, Ohio) to measure the maximum standardized uptake value (SUVmax) of each tumor on the following 4 data sets: (1) whole-body, or baseline, PET/CT, (2) average PET/CT (whole-body PET, attenuation correction by average CT), (3) DDG PET (DDG PET, attenuation correction by baseline CT), and (4) DDG PET/CT (DDG PET, attenuation correction by DDG CT).

Statistical analysis

To assess differences in radiation exposure (CTDIvol), patient BMI, scan coverage, and effective radiation dose between repeat CT and cine CT, we used an unpaired t test with Welch correction without the assumption of the same standard deviation. The SUVmax values of the average PET/CT, DDG PET/baseline CT, and DDG PET/CT data sets were normalized to those of the baseline PET/CT data set, and differences between data sets were assessed using 1-way ANOVA with Greenhouse-Geisser correction. All statistical analyses were performed with GraphPad Prism 8.0.0 (GraphPad Software, San Diego, California) and statistical significance was considered true for P < .01.

Results

Comparison of repeat PET/CT and cine CT

The radiation exposure, patient BMI, scan coverage, and CT effective dose data for repeat CT compared with cine CT are shown in Figure 1. Findings are reported as (means ± standard deviations). The mean radiation exposure of repeat PET/CT (7.6 ± 5.8 mGy) and that of cine CT (5.9 ± 1.7 mGy) did not differ significantly (P = .29). The mean BMI of patients who underwent repeat PET/CT (30.9 ± 9.5) and that of patients who underwent cine CT (26.9 ± 6.0) also did not differ significantly (P = .30), indicating that neither group of patients received a larger radiation dose owing to a higher BMI. The scan coverage of repeat PET/CT (32.5 ± 11.5 cm) was significantly larger than that of cine CT (15.4 ± 4.7 cm; P < .001). The effective CT radiation dose of repeat PET/CT (3.7 ± 2.6 mSv) was also significantly higher than that of cine CT (1.3 ± 0.6 mSv; P <.01). Effective CT radiation dose was based on effective dose per dose-length-product of 0.0145 mSv/mGy−1 cm−1, averaged from 0.014 mSv/mGy−1 cm−1 for the chest and 0.015 mSv/mGy−1 cm−1 for the abdomen from the AAPM Report 9627 as the misregistration areas normally cover both the chest and the abdomen.

Fig. 1.

Fig. 1.

Open in a new tab

Plots comparing the (A) radiation exposures, (B) patient body mass index (BMI), (C) scan coverages, and (D) effective radiation doses of 10 repeat positron emission tomography/computed tomography (CT) and 35 cine CT scans. **P < .01; ***P < .001. Abbreviation: ns = not significant.

The CT effective dose was lower for cine CT than for repeat PET/CT because cine CT had a smaller scan coverage at a similar radiation exposure for a similar patient population. In addition, cine CT saved about 6 minutes of PET scan time for a scan coverage of more than 2 AFOVs (35.7 cm) in repeat PET/CT. Two of the 10 repeat PET/CT patients were repositioned from the arms-up position in baseline PET/CT to the arms-down position in repeat PET/CT. On the other hand, all 35 cine CT patients were able to keep their arms-up position in cine CT as in baseline PET/CT, suggesting that fatigue from holding arms up could be an issue in repeat PET/CT. Although most of the repeat PET/CT and cine CT improved registration over the baseline PET/CT, average CT was more reproducible than repeat CT by visual inspection. An example of 2 consecutive average CT scans derived from the cine CT7 of a lung cancer patient not in this study who received CT simulation for radiation therapy is shown alongside a baseline CT followed by a repeat CT of another patient in Figure 2.

Fig. 2.

Fig. 2.

Open in a new tab

Superpositions of sagittal and coronal views of the baseline computed tomography (CT) and repeat CT scans of 1 patient (A) and of 2 average CT scans in the CT simulation of radiation treatment for a lung tumor (cross-hairs) of another patient (B). Spatial agreement in the helical CT for the baseline and repeat CT scans is worse than it is for the 2 average CT scans.

Comparison of average PET/CT, DDG PET, and DDG PET/CT

For the 59 tumors in the 14 scans selected for registration and quantification comparison, the SUV ratios of average PET/CT, DDG PET, and DDG PET/CT to baseline PET/CT were 1.14 ± 0.28, 1.28 ± 0.20, and 1.63 ± 0.64, respectively (P < .0001). These results suggest a trend of the SUVmax increasing from baseline PET/CT to average PET/CT to DDG PET to DDG PET/CT (Fig. 3). Overall, motion correction with DDG PET resulted in a larger increase in SUVmax than registration improvement with average CT did. Then in DDG PET/CT, in which both misregistration and tumor motion were mitigated, the biggest increase in SUVmax among all comparisons was observed. For most tumors, the SUVmax increased consecutively from baseline PET/CT to average PET/CT, to DDG PET, and to DDG PET/CT. Two of the patient studies showing this trend of increase, 1 each with 68Ga-DOTATATE and 18F-FDG, are shown in Figure 4. For a few tumors, the SUVmax did not follow such a consistent trend owing to increased misregistration between the baseline CT and DDG PET data. There were also 2 notable cases worthy of further detailed analysis. One exception was a lung tumor in a 68Ga-DOTATATE study in which misregistration was more dominant than tumor motion. This tumor was falsely avid in baseline PET/CT and DDG PET (SUVmax = 4.9 for both) in which misregistration was clearly present. Then the tumor was nonavid in average PET/CT (SUVmax = 1.1) and DDG PET/CT (SUVmax = 1.6) where misregistration was addressed (Fig. 5A). Avidity was based on a threshold for SUVmax of ≥2.5. The uptake of the lung tumor was influenced by the high uptake of the liver and was artificially high, owing to misregistration between the baseline CT and PET data caused by a deeper-than-normal inspiration during the baseline CT acquisition. The other exception was a liver tumor in an 18F-FDG study (Fig. 5B). This case was falsely nonavid in baseline PET/CT (SUVmax = 1.1) and DDG PET (SUVmax = 1.7), but then avid in average PET/CT (SUVmax = 2.9) and DDG PET/CT (SUVmax = 5.4).

Fig. 3.

Fig. 3.

Open in a new tab

Plot of standardized uptake value (SUV) ratios of average positron emission tomography (PET)/computed tomography (CT), data-driven gated (DDG) PET, and DDG PET/CT to baseline PET/CT. The bottom 10% and top 10% ratios are shown in circles. For most tumors, there was a trend of the SUVmax increasing from average PET/CT to DDG PET to DDG PET/CT. There were two exceptions: one was only 23% and 32%, and the other one was 270% and 500%, from average PET/CT and DDG PET/CT, respectively (both cases are shown in Fig. 5). *P < .0001.

Fig. 4.

Fig. 4.

Open in a new tab

Baseline positron emission tomography (PET)/computed tomography (CT), average PET/CT, data-driven gated (DDG) PET, and DDG PET/CT (from left to right) of (A) a 68GaDOTATATE scan and (B) an 18F-FDG scan. The fusion is at the top, and PET is at the bottom. The numbers are standardized uptake value (SUV)max values.

Fig. 5.

Fig. 5.

Open in a new tab

Baseline positron emission tomography (PET)/computed tomography (CT), average PET/CT, data-driven gated (DDG) PET, and DDG PET/CT (from left to right) of (A) a 68GaDOTATATE scan showing a lung tumor that is falsely avid in the misregistered baseline PET/CT and DDG PET but not avid in the registered average PET/CT and DDG PET/CT; and (B) an 18F-FDG scan showing a liver tumor that is avid in the registered average PET/CT and DDG PET/CT but not avid in the misregistered baseline PET/CT and DDG PET. Avidity was based on a threshold of standardized uptake value (SUV)max ≥2.5.

An example of the SUVmax decreasing from baseline PET/CT to DDG PET but increasing from baseline PET/CT to average PET/CT to DDG PET/CT in an 18F-FDG study is shown in Figure 6A. DDG PET at expiration may not benefit quantification when CT data are dominated by inspiration. Registration can also improve the confidence of lymph node localization. One example of nodal involvement identified in average PET/CT but not baseline PET/CT is shown in Figure 6B.

Fig. 6.

Fig. 6.

Open in a new tab

(A) Baseline positron emission tomography (PET)/computed tomograpy (CT), average PET/CT, data-driven gated (DDG) PET, and DDG PET/CT (from left to right) of an 18FFDG scan. Owing to increased misregistration from baseline PET/CT to DDG PET, the standardized uptake value (SUV)max of the lung tumor decreased from baseline PET/CT to DDG PET. Both average PET/CT and DDG PET/CT showed better registration than baseline PET or DDG PET did. (B) A 65-year-old man with duodenal carcinoid tumor underwent 68Ga-DOTATATE PET/CT for initial staging. The baseline PET/CT (fusion and CT; top row) showed 2 avid foci (arrows) in the portal hepatic region. These 2 foci were not correlated with any corresponding CT abnormalities but were correlated with 2 nodular wall thickenings in the proximal duodenum, corresponding to the known primary cancer, in the average PET/CT (bottom row). Thus, the average PET/CT provided important information for the clinical management of the patient to become a candidate for surgery.

Discussion

In this work we have demonstrated for the first time that the cine CT data used to generate average CT can also be used to produce DDG CT to help maximize the use of DDG PET. The results presented here show that DDG PET/CT improves both registration and quantification relative to other techniques previously used clinically. DDG PET/CT achieved better registration than DDG PET did. DDG PET was more effective than average PET/CT in terms of increasing SUVmax. Notably, the largest increases in SUVmax were achieved with DDG PET/CT, followed by DDG PET and then average PET/CT. The improvements in both registration and quantification indicate that DDG PET, whose data are captured at expiration, may not benefit from a CT taken at inspiration. Furthermore, accurate registration is uniquely important for the identification of nodal involvement, and we observed several examples improved by average or DDG PET/CT.

The present study had a data selection bias because it included only misregistered PET/CT cases. However, these cases were chosen precisely because of their compromised image quality owing to misregistration. All of these cases benefited from more accurate attenuation correction by average CT generated from cine CT data. Cine CT acquisitions also resulted in a smaller effective radiation dose to the patient than repeat CTs did. In addition, cine CT did not require a repeat PET scan, which would require approximately 6 minutes, not counting the time required for patient transport and arm position changes. Therefore, addressing misregistration with average CT derived from cine CT was better than doing so with repeat PET/CT in terms of clinical workflow and patient throughput.

The decision to initiate a cine CT scan can be made while the patient is still on the imaging table and before scan completion. Cine CT data are acquired in about 1 minute, and then the patient is released from the scanner. The computation time to derive average CT from the cine CT data is also about 1 minute. Applying average CT for PET attenuation correction can be completed while the next patient’s data are being acquired. In this study, cine CT and the related average CT computations were implemented on a network of 7 GE PET/CT scanners across multiple campuses.

Several studies have demonstrated methods for mitigating misregistration and/or motion correction,11,14–16,28 but few methods that address both misregistration and motion correction have been implemented in the clinic. The novel clinical protocol presented in this work, in which cine CT is used to generate average CT for attenuation correction and DDG CT to combine with DDG PET for motion correction, is designed particularly for PET/CT studies that are compromised by misregistration and would normally require a repeat PET/CT. Our approach may be among the first to be implemented in the clinic to tackle both misregistration and motion correction together.

In the present study, we did not increase the scan time for DDG PET. However, DDG PET can be used with increased scan time to compensate for the increased image noise due to the selective data extraction from quiescent phase gating. Registration plays an important role in the comparison of DDG PET and DDG PET/CT, as both share the same DDG PET data but have 2 different sets of CT data. The clinical benefits of improved registration and motion correction achievable with DDG PET/CT may outweigh the drawbacks of the increased noise. Further study of this tradeoff is warranted and will be pursued in future work.

Because the scan coverage of repeat CT was more than twice as large as that of cine CT, cine CT resulted in a lower radiation dose to the patient than repeat CT, even at similar radiation exposure levels. One possible explanation for the increased coverage in repeat CT is that the technologists treated repeat PET/CT as a separate PET/CT study requiring the acquisition of multiple AFOVs. Another possible explanation is that the technologists were trying to maintain uniform PET sensitivity in the middle of multiple AFOVs of data acquisition, due to the triangular shape of 3D PET imaging sensitivity with peak performance at the center.

A conceptually easy yet difficult-in-practice solution to mitigate misregistration would be to conduct a repeat CT rather than a combined repeat PET/CT. However, the CT coverage would have to be larger than 1 to 2 AFOVs and accurately cover the misregistration area. The current user interface of the GE scanner does not allow a graphical prescription of a repeat CT scan of 1 or 2 AFOVs associated with the previous PET scan. It is a more complicated process involving the review of PET images and recording the range of each AFOV in the reconstruction parameters. In contrast, cine CT can be performed in multiples of 2 cm without a direct match to the discrete PET AFOV acquisitions, making it a convenient remedy to address misregistration.

Current vendor options for average CT require EDG CT.7 Average CT can be implemented in either cine CT (GE) or helical CT (Siemens, Philips, Canon, etc) scan modes. Our implementation is different in that DDG CT is produced from the same cine CT that is used for average CT. Fortunately, the cine CT for average CT is available on all GE PET/CT scanners, thus making our implementation, which embeds a limited scan coverage of average CT into a whole-body CT, possible. Siemens has also recently prototyped a cine CT for its prospective 4D CT.29 In the future, vendors may want to consider allowing any scan coverage of CT data to be used for attenuation correction of the PET data, as we did in this work for average CT and DDG CT.

Similar to that of the CT data in whole-body PET/CT, the quality of average CT data can be affected by a sudden irregular respiratory cycle. However, because we acquired the cine CT data for 5 seconds per location, we had a better chance of obtaining reliable DDG CT data should the average CT data be affected by an irregular breath cycle.

This approach has great potential to be adopted for radiation therapy planning in which immobilization and a flat tabletop can be introduced and a larger scan range of the whole lung or abdomen is required. There is no constraint in the scan range of the cine CT in DDG PET/CT, which is identical to the cine CT used in GE’s 4D CT. Our design establishes the possibility that a diagnostic PET/CT scanner can be used for radiation therapy simulation when registration of CT and PET is important, tumor motion information is critical, and a respiratory gating device is not available on the PET/CT scanner.

Conclusion

A new approach for addressing both misregistration and motion correction in PET/CT was implemented and compared with the previous techniques in clinical use. Cine CT was used to produce both average CT and DDG CT for attenuation correction of average PET and DDG PET, respectively. This novel DDG PET/CT method resulted in better quantification and registration than either DDG PET alone or average PET/CT without increasing the scan time of PET. An increased scan time is unavoidable in the alternative solution of repeat PET/CT for cases with misregistration. The effective radiation dose of cine CT was also lower than that of repeat PET/CT. Cine CT can improve the clinical workflow and patient throughput compared with repeat PET/CT, and it can be used to maximize the benefits of accurate attenuation correction and motion correction from DDG PET for treatment response assessment and radiation therapy applications.

Acknowledgments—

Removed for review purpose.

Disclosures: This research was supported in part by the NIH Grants 1R21CA222749-01A1 and R03EB030280, GE Healthcare, and a ROSI grant from the University of Texas MD Anderson Cancer Center, Division of Radiation Oncology. T.P. was a consultant for Bracco Medical Systems, LLC. The contents are solely the responsibility of the authors.

Footnotes

All patient data acquired in this study are available for reuse upon internal review board approval and material transfer agreement. All analyzed data are included in the paper.

References

  • 1.Chi A, Nguyen NP. 4D PET/CT as a strategy to reduce respiratory motion artifacts in FDG-PET/CT. Front Oncol 2014;4:205. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Pan T, Mawlawi O, Nehmeh SA, et al. Attenuation correction of PET images with respiration-averaged CT images in PET/CT. J Nucl Med 2005;46:1481–1487. [PubMed] [Google Scholar]
  • 3.Schmidt-Hansen M, Baldwin DR, Hasler E, et al. PET-CT for assessing mediastinal lymph node involvement in patients with suspected resectable non-small cell lung cancer. Cochrane Database Syst Rev 2014;2014:CD009519. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Nehmeh SA, Erdi YE, Pan T, et al. Four-dimensional (4D) PET/CT imaging of the thorax. Med Phys 2004;31:3179–3186. [DOI] [PubMed] [Google Scholar]
  • 5.Nehmeh SA, Erdi YE, Pan T, et al. Quantitation of respiratory motion during 4D-PET/CT acquisition. Med Phys 2004;31:1333–1338. [DOI] [PubMed] [Google Scholar]
  • 6.Werner MK, Parker JA, Kolodny GM, et al. Respiratory gating enhances imaging of pulmonary nodules and measurement of tracer uptake in FDG PET/CT. AJR Am J Roentgenol 2009;193:1640–1645. [DOI] [PubMed] [Google Scholar]
  • 7.Pan T, Lee TY, Rietzel E, et al. 4D-CT imaging of a volume influenced by respiratory motion on multi-slice CT. Med Phys 2004;31:333–340. [DOI] [PubMed] [Google Scholar]
  • 8.He J, O’Keefe GJ, Gong SJ, et al. Device-less gating for PET/CT using PCA. IEEE Trans Nucl Sci 2008;55:2557–2565. [Google Scholar]
  • 9.Khamis H, Wollenweber S. MotionFree: Device-less digital respiratory gating technology, seamlessly integreated in PET imaging routine. Available at: https://www.gehealthcare.de/-/jssmedia/3d354ee989204b1883317512029a37e7.pdf?la=de-de. Accessed November 22, 2020.
  • 10.Walker MD, Morgan AJ, Bradley KM, et al. Data driven respiratory gating outperforms device-based gating for clinical FDG PET/CT. J Nucl Med 2020;61:1678–1683. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Daouk J, Fin L, Bailly P, et al. Improved attenuation correction via appropriate selection of respiratory-correlated PET data. Comput Methods Programs Biomed 2008;92:90–98. [DOI] [PubMed] [Google Scholar]
  • 12.Liu C, Alessio A, Pierce L, et al. Quiescent period respiratory gating for PET/CT. Med Phys 2010;37:5037–5043. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Kalantari F, Li T, Jin M, et al. Respiratory motion correction in 4D-PET by simultaneous motion estimation and image reconstruction (SMEIR). Phys Med Biol 2016;61:5639–5661. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Lu Y, Fontaine K, Mulnix T, et al. Respiratory motion compensation for PET/CT with motion information derived from matched attenuation-corrected gated PET data. J Nucl Med 2018;59:1480–1486. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Chang G, Chang T, Pan T, et al. Implementation of an automated respiratory amplitude gating technique for PET/CT: Clinical evaluation. J Nucl Med 2010;51:16–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Hamill JJ, Meier JG, Betancourt Cuellar SL, et al. Improved alignment of PET and CT images in whole-body PET/CT in cases of respiratory motion during CT. J Nucl Med 2020;61:1376–1380. [DOI] [PubMed] [Google Scholar]
  • 17.Boellaard R, Hofman MB, Hoekstra OS, et al. Accurate PET/MR quantification using time of flight MLAA image reconstruction. Mol Imaging Biol 2014;16:469–477. [DOI] [PubMed] [Google Scholar]
  • 18.Bousse A, Bertolli O, Atkinson D, et al. Maximum-likelihood joint image reconstruction/motion estimation in attenuation-corrected respiratory gated PET/CT using a single attenuation map. IEEE Trans Med Imaging 2016;35:217–228. [DOI] [PubMed] [Google Scholar]
  • 19.Liu F, Jang H, Kijowski R, et al. A deep learning approach for 18F-FDG PET attenuation correction. EJNMMI Phys 2018;5:24. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Khamfongkhruea C, Thongsawad S, Tannanonta C, et al. Comparison of CT images with average intensity projection, free breathing, and mid-ventilation for dose calculation in lung cancer. J Appl Clin Med Phys 2017;18:26–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Amoush A, Abdel-Wahab M, Abazeed M, et al. Potential systematic uncertainties in IGRT when FBCT reference images are used for pancreatic tumors. J Appl Clin Med Phys 2015;16:5257. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Riegel AC, Chang JY, Vedam SS, et al. Cine computed tomography without respiratory surrogate in planning stereotactic radiotherapy for non-small-cell lung cancer. Int J Radiat Oncol Biol Phys 2009;73:433–441. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Wang JZ, Li JB, Qi HP, et al. Effect of contrast enhancement in delineating GTV and constructing IGTV of thoracic oesophageal cancer based on 4D-CT scans. Radiother Oncol 2016;119:172–178. [DOI] [PubMed] [Google Scholar]
  • 24.Rodriguez-Molinero A, Narvaiza L, Ruiz J, et al. Normal respiratory rate and peripheral blood oxygen saturation in the elderly population. J Am Geriatr Soc 2013;61:2238–2240. [DOI] [PubMed] [Google Scholar]
  • 25.Pan T, Mawlawi O, Luo D, et al. Attenuation correction of PET cardiac data with low-dose average CT in PET/CT. Med Phys 2006;33:3931–3938. [DOI] [PubMed] [Google Scholar]
  • 26.Carnes G, Gaede S, Yu E, et al. A fully automated non-external marker 4D-CT sorting algorithm using a serial cine scanning protocol. Phys Med Biol 2009;54:2049–2066. [DOI] [PubMed] [Google Scholar]
  • 27.American Association of Physicists in Medicine. Report No. 096—The measurement, reporting, and management of radiation dose in CT. Available at: https://www.aapm.org/pubs/reports/detail.asp?docid=97. Accessed November 22, 2020.
  • 28.Fin L, Daouk J, Morvan J, et al. Initial clinical results for breath-hold CT-based processing of respiratory-gated PET acquisitions. Eur J Nucl Med Mol Imaging 2008;35:1971–1980. [DOI] [PubMed] [Google Scholar]
  • 29.Werner R, Sentker T, Madesta F, et al. Intelligent 4D CT sequence scanning (i4DCT): First scanner prototype implementation and phantom measurements of automated breathing signal-guided 4D CT. Med Phys 2020;47:2408–2412. [DOI] [PubMed] [Google Scholar]

RESOURCES