Impact of Respiratory-gated 4D PET/CT Scan for Motion Correction in Characterizing Lesions Adjacent to the Diaphragm – A Cross-sectional Study at a Tertiary Care Institute

Introduction

F-18 fluorodeoxyglucose positron emission tomography (F-18 FDG PET) is widely used in oncological imaging for diagnosis and metastatic workup because the proliferating cancer cells have a higher rate of glucose metabolism (measured as standard uptake value-SUV). Hybrid imaging by integrating 18F-FDG PET with computed tomography (CT) has incremental value, combining metabolic and anatomical information. The metabolic information of FDG PET is useful in characterizing indeterminate lung nodules on CT. Accurately localized PET information helps to diagnose and manage patients with malignant disease.

In sequential whole-body PET/CT acquisition, PET acquisition takes around a few minutes to get enough signal for image reconstruction with positron emission (time-averaged image over many breathing cycles), whereas CT acquisition takes only a few seconds (snapshot of the lesion at one distinct phase of the respiratory cycle). Respiratory motion can cause underestimation of the metabolic activity by about 30% and overestimation of lesion volumes by almost two times or more due to smearing or mismatching.[1]

In case of lung tumors, the magnitude of the motion depends on its dimension (smaller tumor, high mobility) and its location (mobility increases on passing from the upper to the middle, down to the lower lobe).[2] Motion direction is mostly craniocaudal, with magnitude ranging from a few millimeters to 1–3 cm for the lungs and around 1.5–2.5 cm for the liver.[34] Respiratory motion blur the images and degrade its contrast due to smearing, leading to overestimation of lesion volume and underestimation of SUV (as mismatching of PET and CT data affects attenuation, proved by phantom studies).[56]

Hence, the knowledge of organ and lesion motion is an important factor for diagnosis and management in tumor imaging, as well as for radiotherapy planning, requiring accurate safety margin. Assessment of tumor response also becomes unreliable due to respiratory motions, especially for lung lesions.[78]

Phase-based respiratory-gated (RG) PET/four-dimensional (4D) PET is one of the methods to correct the respiratory motion effects on PET-CT. Here, PET data are acquired into discrete bins in synchrony with the breathing phase over a few respiratory cycles using motion sensors. The final image is formed by combining the discrete bins of each respiratory cycle phase by phase. Such short-time bins allow the near-instantaneous tracking of the lesion, thereby resulting in approximately motion-free images using various software reconstructions. Investigators have used different respiratory tracking systems to monitor the respiratory motion, which can be hardware or software systems, that simultaneously generate a trigger during image acquisition at a user-predefined phase.[9]

Different types of external respiration sensors are pressure sensors, spirometers, temperature sensors, and video camera systems. Anzai belt, AZ-733V marketed by Anzai Medical (Tokyo, Japan), is one such pressure-sensing device used for gating.[10111213]

This study aims to analyze the impact of phase-based motion-corrected gated PET/CT on lesions adjacent to the diaphragm using Anzai pressure sensors and various reconstruction software.

Materials and Methods

This was a cross-sectional study, done in a tertiary care institute within a study period of 2 years involving adult cancer patients with liver and lung lesions on PET/CT. This study was conducted after getting institutional ethics committee approval and patient consent.

Patients excluded in this study: pediatric age group, patients with respiratory distress or on any artificial ventilator support, very sick, disoriented, uncooperative or unwilling patients.

Imaging protocol

After at least 4 h of fasting, with measured blood glucose level < 160 mg/dL, patients received an intravenous injection of about 148–185 MBq (4–5 mCi) of F-18 FDG. Whole-body imaging was performed using a combined PET/CT scanner (Discovery PET/CT IQ, GE Healthcare, WI, USA) as per the standard guidelines using the conventional protocol about 55–75 min later, and images were reconstructed.[14]

About 80 min to 120 min after F-18 FDG administration, gated PET and CINE mode CT images (with low radiation dose-10 mA, 120 KVp) of the desired area (chest or upper abdomen) were acquired using a pressure sensor system (AZ7337, Anzai Real-time Position Management, Anzai Medical Systems, Tokyo, Japan). The sensor part consists of an elastic belt, which can be fastened around the patient’s abdomen or thorax. During inspiration, increased stress results in increased pressure to the belt. This pressure is then measured by a small load cell, an electronic force sensor placed in a pocket inside the belt. The belt is usually interfaced to a computer, where the change in the pressure, and thus breathing amplitude, is displayed as a function of time. To synchronize the recorded respiratory signal with the list-mode PET acquisition, a custom-built cable is used. When activated, the system issues an end-expiratory trigger signal to start the gated data acquisition in list mode and to mark the beginning and end of the respiratory cycle, which was divided into six equal periods for 5 min. The system works by distributing PET data into each of these bins according to the corresponding temporal phase; for a patient exhibiting rhythmic breathing, the temporal phase corresponds to the position of the thorax in the respiratory cycle [Figure 1a]. Software, i.e. Q.Freeze and Q.Static (by GE Healthcare), were used for image reconstruction of the gated data.

Close

Figure 1 (a) Q.Freeze – Each respiratory cycle data divided into 6 bins. The image formed by coregistration of corresponding bins of all respiratory cycles with quiescent phase as reference. (b) Tumor volumes – nongated (left) versus gated (right). (c) Q.Static – Extract data from the quiescent part of respiratory cycles

Export to PPT

RG (by Q.Freeze and Q.Static software) and nonrespiratory-gated datasets of the desired area were generated from the same time list mode data and compared, instead of the initial whole-body image, which is taken at a different time frame.

Results

A total of 18 patients with 55 lesions were analyzed. The mean age of the participants was 56.28 ± 14.05 years. These patients had mostly lung or breast as primary malignancy. Among 55 lesions studied, 65% were in the lungs, 33% were in the liver, and 2% were in the spleen. Ten patients had only lung lesions, four patients had lung and liver lesions, three patients had only liver lesions, and one patient had spleen and lung lesions [Table 1].

There was an overall significant difference in SUV_max between different gating methods as compared to nongated method [P < 0.001, Tables 2-6]. On gating methods, there was a higher SUV_max than that of nongating method. The percentage increase in SUV_max was 7.74% (with GFCR0), 18.13% (with GFCR1), and 13.22% (with GS) in gated images.

There was a significant difference in MTV between different gating methods as compared to nongated method (P < 0.001). MTV was less on gating methods than that of nongating method. The percentage decrease of MTV was 18.3% (with GFCR0), 32.9% (with GFCR1), and 15.4% (with GS) in gated images. There was a significant difference in TLG in GFCR1 and GS as compared with nongated method [P = 0.02, Tables 2-6].

In this study, among the three gating methods, Q.Freeze data with cycle rejection (GFCR1) showed the highest increase in SUV_max (18.13%) and reduction in MTV (32.9%), but the problem with few cases was more rejection of cycles causing count loss and excess noise. This caused to exclude these few cases in overall statistical calculation for comparison with other methods affecting the overall parameters of GFCR1 [Table 1].

Discussion

Overall, in comparison with nongating (G0) method, there was a significant increase in SUV_max (P < 0.001) in all gating methods in this study. A review article has quoted the maximum increase in SUV_max to be 83.3% with a maximum increase for small lung lesions (<10 mm) and those located near the diaphragm.[15] This review also summarized that RG improved the diagnostic performance of PET/CT for equivocal lung nodules (i.e. improved Sn, Sp, and accuracy). SUV is lower when a tracer-avid feature is moving during acquisition because the number of events is distributed over a larger volume of interest. Hence, RG should result in higher SUV_max which is demonstrated in our study results.

FDG PET/CT has found its way into oncologic treatment planning as another method for the determination of gross MTV of primary tumors and metastases.[16] However, respiratory motion of intrapulmonary lesions leads to an erroneous larger target irradiation area causing more damage to healthy surrounding lung tissue.[17] Breath-hold techniques are compromised if patients cannot comply with breathing instructions, which can be a problem in lung cancer patients. Robin et al., in their study, found a significant reduction of MTV (18%) and TLG with gating which is also found in our study.[18] This improvement is important in treatment planning before radiation therapy to define the irradiation target volume more accurately.[19] The visual blur in the static image resulting from respiratory motion generally appears stretched along the direction of the motion and can be quantified by tumor volume change [Figure 2].

Close

Figure 2 (a and b) Nongated coronal and axial images of the left lung nodule appear blurred (arrow). (c and d) Gated images – the lesion is better delineated with increased SUV (arrow). (e) Nonavid left lung nodule near diaphragm on nongated image shows avidity on the gated image (arrow) (f)

Export to PPT

Both phantom and retrospective clinical studies have demonstrated superior image quality and increased accurate quantitative parameters in gated images compared to conventional static images.[2021]

Majority of patients tend to spend more breathing time dwelling at the end-expiration quiescent period and breathe out to the same end-expiratory location. These results indicate that imaging at end-expiration could involve less motion while still retaining a large fraction of the PET data.

Q.Static (GS), known as quiescent phase gating, extracts a fraction of PET data from end-expiratory quiescent portions of patient breathing cycles and forms a single 3D PET image volume. Thus, there is an increased signal-to-noise ratio. It is an automated motion correction method that can be integrated into the clinical whole-body PET protocol and provides motion-corrected images of the desired area after reconstruction simultaneously with the conventional PET acquisition and reconstruction. Thus is more convenient to use in regular clinical practice.

RG improves the accuracy in assessment of upper abdominal lymph node involvement. In one patient in this study, new tiny avid lymph nodes in the abdomen were identified on gated image which was not detected in ungated image [Figure 3a and b]. New lung nodules were also found after gating in two patients which were not appreciated in nongated study [Figure 3c and d]. Even non-FDG-avid lung nodule on nongated image showed FDG uptake on gated study [Figure 3e and f]. Hence, it suggests that gating could facilitate the detection of very small lesions which has a major impact on diagnosis, management, and staging.[22] The sharpness of the margins of the lung nodules is also better than nongated images, particularly those at interface of the lung and diaphragm, confusing as liver lesions [Figure 4a and b].[23]

Close

Figure 3 (a, c and e) Nongated axial and sagittal images and (b, d and f) its corresponding gated images. Few new FDG-avid abdominal lymph nodes and lung nodules were detected/better appreciated after respiratory gating (arrows in b, d and f)

Export to PPT

Close

Figure 4 (a) Nongated sagittal image, (b) Gated image. The lung nodule which was smeared with the liver surface is seen separately after gating (arrows). (c) Nongated coronal image, (d) Gated image. A liver lesion better detected after respiratory gating (arrow)

Export to PPT

Liver lesions are also affected by respiratory movement which on fused PET-CT images appear blurred and smeared. Physiological uptake of the surrounding liver parenchyma under voluntary respiration also affects the assessment of the lesions which are better appreciated with gating [Figure 4c and d]. Therefore, this feasibility study shows RG PET/CT images having higher SUV_max and lower MTV in liver lesions [Table 1]. A similar diagnostic performance was reported by Cheung et al.[24]

The results of our study suggest that RG improves the semi-quantitative assessment of lesions adjacent to the diaphragm on PET/CT. Although some additional time and training are needed for setup and acquisition, RG can be incorporated practically and efficiently into clinical routines. New PET/CT scanners have fully automatic processes for RG data sorting and image reconstruction (like the Q.Static software), making the RG technique feasible in routine clinical setting. 4D phase-based RG using Q.Freeze should be reserved for truly equivocal cases when the detection and true FDG activity of a lesion in the lower thorax or upper abdomen would likely influence the clinical management of the patient, like for staging or planning targeted therapy. RG also has the potential for providing accurate measurements of metabolic parameters for targeted therapy, monitoring of the therapeutic response, assessment of tumor viability, textural analysis, and prognostication. It is likely to be an important component of further future advancements in molecular imaging, where the highest spatial resolution and sensitivity are desired. Further extensive studies are needed to define the impact of these improvements on diagnosis, radiation therapy planning, therapy monitoring, and textural analysis.[25] The major limitations to widespread clinical adoption of respiratory motion correction to date are the need for extra acquisition, need for training, and expertise for the external device-driven RG. This can be eliminated if RG can be incorporated directly in initial image acquisition, which seems possible with Q.Static software. Messerli et al., studied feasibility of using data-driven RG [DDG] with Q.Static, showing promising results impacting clinical management.[26] DDG PET/CT also showed more accurate quantification of lung and upper abdomen lesions in another study.[27]

The limitation of this study is that the number of patients included is relatively small. This was due to the heavy workload of the department to acquire another study for the patient and difficulty in training the patient for regular respiration. Furthermore, since the patients analyzed in this study had multiple metastatic lesions, the data did not significantly impact the management but would definitely be useful for follow-up/response evaluation. A larger-scale and prospective investigation to assess the clinical impact of RG PET/CT is required.

Conclusion

From this study, it is concluded that RG improves the quality of PET imaging in lesions adjacent to the diaphragm as well as its quantification and feasible in routine clinical practice for selected patients who would be benefited. We have shown that image smearing can be reduced by phase-based motion-corrected RG FDG PET/CT. This technique allows a more accurate definition of the metabolic parameters with the identification of new lesions, useful for diagnosis, staging, treatment, prognostication, and response evaluation. It is also useful to differentiate between lung and liver lesions near the diaphragm further impacting clinical decision and management.