Shortened Acquisition Duration for Brain Tumor ¹¹C-Methionine Positron Emission Tomography on Silicon Photomultiplier Positron Emission Tomography/Computed Tomography

Abstract

Positron emission tomography/computed tomography (PET/CT) scanners equipped with silicon photomultiplier detectors offer superior sensitivity and count-rate performance. The aim of this study was to evaluate the feasibility and impact of shortening the acquisition duration in brain tumor ¹¹C-methionine PET using a silicon photomultiplier PET/CT system, and to assess how point spread function (PSF) correction influences quantitative values. In the phantom study, a brain tumor phantom was scanned using the Biograph Vision silicon photomultiplier-based PET/CT system. Data were acquired for 10, 5, 3, and 1 min, and the images were reconstructed with and without PSF correction. In the clinical study, 20 patients who underwent ¹¹C-methionine PET were retrospectively analyzed. PET data were acquired over 10 min and subsequently reconstructed for 10, 5, and 3 min. We evaluated quantitative parameters including the maximum standardized uptake value (SUVmax), and their relative errors under shortened acquisition durations were analyzed. In the phantom study, the SUVmax increased with shorter acquisition durations; however, this increase was less pronounced with PSF correction. In the clinical study, relative errors of SUVmax for the 5 and 3 min acquisitions with PSF correction were 2.9 ± 3.8% and 5.2 ± 5.4%, respectively. They were smaller than those without PSF correction (5.5 ± 5.1% and 12.7 ± 8.5%), indicating superior quantitative stability with shortened acquisition duration. The combination of the Biograph Vision system and PSF correction enabled the acquisition of high-quality PET images with shortened scan times.

1. Introduction

¹¹C-methionine is a radiotracer that reflects amino acid metabolism [1] and exhibits lower uptake in normal brain tissue than in ¹⁸F-fluoro-2-deoxy-D-glucose (¹⁸F-FDG). Consequently, ¹¹C-methionine has superior diagnostic performance in brain tumor differentiation [2,3,4,5] compared with that of ¹⁸F-FDG and is recognized as a valuable tracer for discriminating between tumor recurrence and radiation necrosis following radiotherapy [6,7,8,9]. Furthermore, positron emission tomography (PET) studies have reported that the ¹¹C-methionine uptake area corresponds well with the area of tumor extension compared with the contrast-enhanced area on magnetic resonance imaging and is useful for treatment planning [10,11,12].

In clinical PET imaging, determining the appropriate acquisition parameters, particularly the acquisition duration, is critical [13,14,15]. The standard protocol typically involves a 10 min acquisition beginning 20 min after the intravenous administration of ¹¹C-methionine [16]. PET image quality is primarily influenced by the number of detected events, with longer acquisition durations enhancing the signal-to-noise ratio (SNR). However, extended acquisition durations increase the patient burden and the risk of motion artifacts, particularly in individuals who have difficulty maintaining a stable position. Therefore, optimizing the acquisition duration is essential for achieving a balance between image quality and patient comfort. Furthermore, the shortened acquisition duration of ¹¹C-methionine PET leads to cost savings, more efficient staffing, and effective use of medical resources, resulting from an increased number of examinations per day.

Recent advancements in PET technology, such as silicon photomultiplier (SiPM)-based PET/computed tomography (CT), have been significant. A Biograph Vision PET/CT scanner (Siemens Healthineers, Knoxville, TN, USA) equipped with SiPM detectors has a superior SNR and quantitative performance compared with that of a conventional photomultiplier tube (PMT)-based PET/CT scanner [17,18,19]. Reportedly, Biograph Vision can shorten the acquisition duration or reduce the injection dose for whole-body ¹⁸F-FDG PET [20,21,22,23], and it may also enable shorter acquisition durations in clinical ¹¹C-methionine PET. The Japanese Society of Nuclear Medicine recommends a 10 min acquisition as the standard PET protocol for brain tumors [16]; however, this recommendation aims to ensure consistent diagnostic quality across various institutions and scanner types. Therefore, it is important to investigate the minimum acquisition duration required to maintain clinical utility.

Morimoto et al. reported the potential to shorten the acquisition durations in brain tumor ¹¹C-methionine PET [24]. However, their study used PMT-based PET/CT and did not evaluate the effect of point spread function (PSF) correction, which significantly affects PET image quality. PSF reconstruction can enhance image contrast, SNR, and lesion detectability [25,26] but may also influence quantitative measures, such as standardized uptake value (SUV) and metabolic tumor volume (MTV), associated with Gibbs artifacts depending on lesion size [27,28,29,30]. SUV and MTV are important parameters for determining tumor malignancy, prognosis, and treatment strategy [31,32,33,34,35,36]. Therefore, a comparative evaluation of reconstructions with and without PSF correction is necessary to determine whether PSF-related artifacts compromise the utility of a shorter acquisition duration.

The aim of this study is to assess the feasibility and quantitative impact of shortened acquisition duration in ¹¹C-methionine PET using SiPM-based PET/CT. We investigated the relationship between acquisition duration and quantitative metrics with and without PSF correction in image reconstruction using phantom and clinical evaluation.

2. Materials and Methods

2.1. Phantom Study

The phantom study was conducted according to the phantom testing procedure manual for ¹¹C-methionine brain tumor PET imaging established by the Japanese Society of Nuclear Medicine [16]. To simulate a clinical examination, six spheres of varying sizes (φ38, 27, 16, 10, 7.5, and 5 mm) and the background area of a brain tumor phantom were filled with 12.3 and 4.1 kBq/mL ¹⁸F solution at the start of the scan (a radioactivity concentration ratio of 3:1), respectively.

• Data Acquisition and Image Reconstruction

The PET data were acquired using a Biograph Vision 600 PET/CT scanner equipped with SiPM detectors [37,38,39].

To assess the impact of shortened acquisition durations, list-mode data acquired in the ¹⁸F-phantom study were extracted for 523, 280, 173, and 59 s from the beginning of the scan, corresponding to 10, 5, 3, and 1 min of acquisition at ¹¹C-methionine, respectively.

Reconstructions with and without PSF were performed to evaluate the effect of PSF correction using the scanner console. The reconstruction conditions are as follows:

-

Reconstruction with PSF: Ordered subset expectation maximization algorithm with PSF correction and time-of-flight (TOF) technique using five iterations and five subsets.

-

Reconstruction without PSF: Ordered subset expectation maximization algorithm with the TOF technique using three iterations and five subsets.

Reconstruction with PSF was our clinical reconstruction protocol, while the number of iterations for the reconstruction without PSF was determined from the contrast-to-noise curve to ensure image quality comparable to that of the PSF reconstruction. The slice thickness for the PET images was 1.65 mm, and no post-smoothing filter was applied.

CT scans for attenuation correction were performed according to our clinical protocol, and the parameters were as follows: 120 kV, 3 mm slice thickness, 1.0 s rotation, using auto exposure control.

• Data Analysis

Regions of interest with the same size as the spheres were placed on the CT images and copied onto the PET images, as illustrated in Figure 1. In each reconstruction condition, the accuracy of SUVmax for the 10 min acquisition and shortened acquisitions was evaluated using AMIDE [40] software (1.0.6. version).

2.2. Clinical Studies

Twenty consecutive patients who underwent ¹¹C-methionine PET between May 2021 and November 2022 and exhibited abnormal uptake (31 lesions) were retrospectively analyzed. The study population consisted of eleven women and nine men. The patients’ ages and weights ranged from 38 to 87 years and 40.0 to 95.4 kg, respectively. All patients were administered intravenous injections of 313.7–428.0 MBq ¹¹C-methionine.

This study was approved by the ethics committee of our institution (No. 23-39), which determined that additional informed consent was not required, as the analysis fell within the scope of the previously approved protocol (No. 19-10).

• Data Acquisition and Image Reconstruction

PET and CT were performed according to clinical protocols. The CT scan parameters were described in the phantom study section. Twenty minutes after the administration of intravenous ¹¹C-methionine, PET data were acquired in list mode for 10 min with the patient in the supine position. The acquired list-mode data were reconstructed for scan durations of 10, 5, and 3 min, using the same reconstruction parameters as in the phantom study.

• Data Analysis

As shown in Figure 2, a volume of interest (VOI) was placed to encompass the entire lesion, and a spherical VOI (φ7 mm) was placed in the contralateral hemisphere [41]. Voxels with an SUV of at least 1.3 times the SUVmean in normal tissue were classified as tumor regions [24,41,42], and SUVmax, SUVpeak, SUVmean, and MTV were measured for each image using syngo.via (Siemens Healthineers, Erlangen, Germany, version VB80F) software.

The relative error of SUVmax (rΔSUVmax_τ) between the 10 min acquisition and a shortened acquisition duration (τ) was calculated as follows:

$$r\Delta {SUVmax}_{\tau }={\displaystyle \frac{{SUVmax}_{\tau }-{SUVmax}_{10\min }}{{SUVmax}_{10\min }}}\times 100\left[\%\right]$$

where SUVmax_τ represents the SUVmax in the shortened acquisition duration (τ), and SUVmax_10min represents the SUVmax in the 10 min acquisition. Similarly, rΔSUVpeak_τ, rΔSUVmean_τ, and rΔMTV_τ were calculated.

• Visual Assessment

Three radiologists independently evaluated the quality of the PET images. No formal calibration or consensus session was conducted prior to the image interpretation, and each radiologist assessed the images according to their own judgment. The images reconstructed with the clinical conditions (10 min acquisition and reconstruction with PSF) were considered the reference (5 points as a full score). PET images from each reconstruction condition and acquisition duration were assessed relative to the reference images using syngo.via software (Siemens Healthineers). The scoring and evaluation criteria were as follows:

5: Excellent (equivalent image quality to the reference image);

4: Good (slightly inferior to the reference image);

3: Acceptable (no impact on tumor invasion assessment, clinically acceptable);

2: Poor (potential misinterpretation of tumor invasion, not clinically acceptable);

1: Very poor (risk of false negatives or false positives).

Scores of 3, 4 and 5 were considered clinically acceptable; on the contrary, scores of 1 and 2 were deemed suboptimal for diagnosis [43]. For each patient, we calculated the difference in visual assessment score (ΔVAscore) between the reconstructions with and without PSF and evaluated the relationship between ΔVAscore and tumor-normal ratio (TNR) in clinical conditions. ΔVAscore and TNR were calculated as follows:

$$\begin{gathered} \mathsf{\Delta }VAscore={VAscore}_{withPSF}-{VAscore}_{withoutPSF} \\ TNR=SUVmax/{SUVmean}_n \end{gathered}$$

where the VA score represents the average visual assessment score assessed by three radiologists for each patient. SUVmean_n represents the SUVmean in the normal brain tissue from the VOI in the contralateral hemisphere. The lowest TNR was observed in the patients with multiple lesions.

• Statistical Analysis

Correlations between SUVmax, SUVpeak, SUVmean, and MTV for the shortened and 10 min acquisitions were evaluated using Pearson’s correlation coefficient.

To evaluate the agreement between the SUVmax for the 10 min acquisition and those for the shortened acquisitions, Bland–Altman analysis was performed.

Inter-observer agreement for visual assessment was evaluated using Fleiss’ kappa statistics. The kappa values were interpreted according to the criteria of Landis and Koch [44] as follows: kappa values <0 as indicating no agreement and 0–0.20 as none to slight, 0.21–0.40 as fair, 0.41–0.60 as moderate, 0.61–0.80 as substantial, and 0.81–1.00 as almost perfect agreement.

All statistical analyses were performed with JASP software (version 0.95.4, JASP Team, Amsterdam, The Netherlands) [45].

3. Results

3.1. Phantom Study

In the phantom study, an increase in statistical noise (Figure 3) and an overestimation of SUVmax (Figure 4) were observed under both reconstruction conditions even in the standard 10 min acquisition. The SUVmax without PSF was increasingly overestimated as the acquisition duration was shortened, whereas the SUVmax with PSF remained stable and was less influenced by the acquisition duration, irrespective of the sphere size. For example, the SUVmax of the 38 mm sphere in the reconstruction with PSF were 3.5, 3.7, 3.6, and 5.0 for the 10, 5, 3, and 1 min acquisitions, respectively. In contrast, the corresponding values in the reconstruction without PSF were 4.0, 4.8, 5.2, and 7.3.

3.2. Clinical Study

SUVmax, SUVpeak, SUVmean, and MTV showed significant correlations between 10 min acquisition and shortened acquisition durations (Figure 5 and Figure A1). The correlation coefficients were >0.98 for all quantitative parameters, for reconstructions with and without PSF. The regression slopes were close to unity for all conditions, ranging from 0.94 to 1.05, except for the 3 min acquisition for SUVmax without PSF: y = 1.10x + 0.09.

rΔSUVmax was larger for 3 min acquisition durations than for 5 min, and larger in the reconstruction without PSF than with PSF (Figure 6). The values were 2.9 ± 3.8% and 5.2 ± 5.4% for the 5 and 3 min acquisitions with PSF, and 5.5 ± 5.1% and 12.7 ± 8.5% without PSF, respectively. The rΔSUVpeak were smaller compared with rΔSUVmax, and the differences between the reconstruction conditions and the acquisition durations were also smaller. The rΔMTV showed similar trends to rΔSUVmax: in reconstruction with PSF, rΔMTV values for the 5 and 3 s acquisitions were −0.1 ± 9.0% and 1.7 ± 14.2%, and the corresponding values in reconstruction without PSF were 6.9 ± 13.6% and 10.8 ± 22.8%, respectively (Figure A2 in Appendix A).

The Bland–Altman analysis for SUVmax showed a mean bias of −0.23, with 95% limits of agreement ranging from −0.77 to 0.31 between the 10 and 3 min acquisitions with PSF, whereas the corresponding values without PSF were −0.57 and −1.56 to 0.42 (Figure 7). For SUVpeak, the mean bias and 95% limits of agreement were −0.03 (−0.20 to 0.14) with PSF, and −0.02 (−0.17 to 0.13) without PSF (Figure 8).

Figure 9 shows clinical images with a large difference in visual scores between reconstructions with and without PSF. The median scores in the 10, 5, and 3 min acquisitions were 5 (reference), 5 (range 5–5), and 4 (range 4–5) for reconstruction with PSF. The corresponding values for reconstruction without PSF were 2 (range 2–4), 2 (range 1–3), and 2 (range 1–2). Similar to the phantom study, an increase in the statistical noise was observed with shorter acquisition durations.

The visual assessment score was superior in reconstruction with PSF than in reconstruction without PSF (

Table 1. The visual assessment scores for each reconstruction condition and acquisition duration.

Reconstruction Condition	Acquisition Duration (min)
	10	5	3
with PSF	5	5	5
(range)	(reference)	(4–5)	(3–5)
without PSF	4	3	2
(range)	(2–5)	(1–5)	(1–5)

). The median scores with the PSF correction for the 5 and 3 min acquisitions were 5 (range 4–5) and 5 (range 3–5), respectively, with none of the radiologists assigning a score of ≤ 2. In the reconstruction without PSF, the median scores for the 10, 5, and 3 min acquisitions were 4 (range 2–5), 3 (range 1–5), and 2 (range 1–5), respectively. The scores for the 3 min acquisition without PSF ranged from 1 to 5, varying depending on the patient and lesion characteristics. The kappa value was 0.35, indicating fair agreement among readers according to the Landis and Koch criteria.

The ΔVAscores for patients with low TNR lesions tended to be larger. The ΔVAscore for 3 min acquisition and TNR showed moderate correlations, as shown in Figure 10, and the correlation coefficient was 0.53.

4. Discussion

In this study, we evaluated the influence of shortened acquisition duration and PSF correction on quantitative values in ¹¹C-methionine brain tumor PET using the latest generation of SiPM PET/CT scanners. Previous studies have investigated the feasibility of shortened acquisition durations in ¹¹C-methionine PET using conventional PMT-based PET/CT systems [24]; however, to the best of our knowledge, this is the first study to evaluate shortened acquisition duration using the SiPM PET/CT scanner and the effect of PSF correction.

In the phantom and clinical studies, SUVmax and rΔSUVmax increased with shorter acquisition durations; however, these increases were less pronounced in the reconstruction with PSF. SUVmax is highly sensitive to statistical noise, which becomes more prominent with shorter acquisition durations owing to the reduced number of detected events. Akamatsu et al. reported that PET images reconstructed with PSF correction exhibited lower coefficients of variation than those reconstructed without PSF correction [26], and the results of our study are consistent with their observation that reconstruction with PSF suppresses noise-induced variability in SUVmax. In our study, the rΔSUVmax with PSF in the 3 min acquisition was 5.2 ± 5.4%, which is sufficiently smaller than the uptake variability of ¹¹C-methionine in normal brain tissues, which is up to 12.6% [46]. In addition, Bland–Altman plots demonstrated that the SUVmax derived from the shortened acquisitions were in close agreement with those from the 10 min acquisition, with no systematic bias across the measurement range. These findings further support the quantitative reliability of the shortened protocols.

This supports the reliability of SUV measurements in terms of shortened acquisition duration with the SiPM PET/CT device, Biograph Vision system, and PSF reconstruction.

Reportedly, Gibbs artifacts caused by PSF might influence quantitative accuracy [27,28,29]. However, obvious Gibbs artifacts were not observed in the present phantom or clinical studies, and reconstruction with PSF showed a smaller relative error for SUVmax. This indicates that the advantage of the noise-suppressing effect of the PSF outweighs the disadvantage of Gibbs artifacts.

SUVpeak and SUVmean exhibited smaller relative errors than SUVmax. Lodge et al. reported that the SUVpeak is a more robust parameter for assessing the most metabolically active regions of tumors [47]. Similarly, Akamatsu et al. demonstrated that the SUVpeak is a highly reproducible metric with reduced susceptibility to statistical noise [48,49]. Our results are consistent with these findings, indicating that SUVpeak is a more reliable quantitative parameter than SUVmax in clinical ¹¹C-methionine PET studies, particularly with shortened acquisition durations.

In the reconstruction with PSF, the MTV also exhibited smaller relative errors than those without PSF. MTV exhibited larger relative errors than other quantitative indices; however, in the majority of lesions, MTV remained within ±1.0 cm³. Lesions with greater MTV tended to have larger volumes, irregular morphologies, or multiple discrete areas of increased uptake within a single lesion. These characteristics likely result in more extensive boundary regions adjacent to the normal tissue, thereby increasing the potential for cumulative errors in lesion delineation.

The visual assessments further support the advantages of PSF correction for shortened acquisition durations. The PET images with PSF reconstruction were highly scored (> 3), regardless of the acquisition duration, and consistently provided clinically acceptable image quality across a range of lesion uptake intensities. The inter-observer agreement does not represent high reliability, as quantified by Fleiss’ kappa (kappa value = 0.35); however, it suggests a consistent tendency in the readers’ judgments. These results indicate that PSF reconstruction helps evaluate the extent of tumor invasion and distinguishes tumor recurrence from radiation necrosis following radiotherapy, even with shortened acquisition durations. The relationship between ΔVAscore and TNR showed the trend that PSF correction improves the VAscore, particularly for patients with low-contrast lesions to normal brain tissue. For example, images without PSF correction, as shown in Figure 9, did not meet clinical standards and were assigned lower scores (1 or 2) for a shortened acquisition duration; however, the visual scores were significantly improved by the use of PSF correction. These results indicate that PSF correction may be useful for improving the visibility or detectability of low-contrast lesions in a shortened acquisition duration. However, further investigation involving more patients and lesions is needed.

Considering the above, the adverse influences of Gibbs artifacts introduced by PSF correction appear to be minimal with the current reconstruction settings and the SiPM PET/CT device used in the study, the Biograph Vision system. Our findings suggest that the suppression of statistical noise through PSF correction has a more significant impact on the quantitative accuracy than any disadvantage associated with Gibbs artifacts. Furthermore, the combination of PSF correction and SiPM PET/CT allows for clinically acceptable imaging, even with acquisition durations as short as 3 min. However, further studies are needed to validate whether these findings are applicable to various SiPM PET/CT or reconstruction conditions. The shortened acquisition time may reduce patient motion, yield motion-free images, and further improve image quality. The results of this study also showed the potential for reducing the radiation exposure dose for patients or medical staff via injection dose reduction with a conventional acquisition duration [50,51,52]. To fully realize these benefits, continuous education and training are essential [53,54,55], particularly regarding scanner-specific reconstruction settings, PSF correction behavior, and optimized acquisition protocols, which should be addressed in future studies.

This study has some limitations. First, the sample size was relatively small, which may have limited the statistical power and generalizability of the findings, making it difficult to classify and analyze the data according to tumor type, size, or anatomical distribution, as well as patient-related factors such as age, body weight, and administered dose.

Second, the retrospective nature of the study may have introduced a selection bias and unmeasured confounding factors. Third, not all cases were confirmed histopathologically, which may have affected the diagnostic certainty and accuracy of the findings. Despite these limitations, the present study provides preliminary but clinically relevant evidence supporting the feasibility of short acquisition duration for ¹¹C-methionine PET using SiPM PET/CT systems. These findings suggest that scan duration may be reduced without substantially compromising image quality, potentially improving patient comfort and enabling more effective use of medical resources. Future studies with larger cohorts and histopathological confirmation are warranted to validate and generalize these results.

5. Conclusions

The combination of the Biograph Vision system and PSF correction has the potential to produce clinically acceptable ¹¹C-methionine PET images, even with a 3 min acquisition duration, offering advantages over conventional PMT-based PET/CT systems.