Since its inception in the late 1990s [1], hybrid positron emission tomography (PET) and computed tomography (CT), also referred to as dual-modality PET/CT, has become a well-established non-invasive imaging modality for a wide variety of clinical applications. In many oncology indications, PET/CT has been accepted as a standard imaging modality in patient management, providing both metabolic and anatomic information for diagnosis and treatment planning [2–4].

The technological innovations of recent years, culminating in the introduction of total-body (TB) PET/CT systems [5], have brought continuous improvements in system performance and sensitivity [6, 7] and consequently expanded PET/CT imaging to new research areas. For instance, the advent of extended axial field-of-view PET systems allows the simultaneous and quantitative imaging of multiple distant organs, thereby providing the possibility to investigate multi-organ metabolic information and detect potential anomalies from normal metabolic activity patterns [8, 9]. To visualize and quantify metabolic aberrations, it is necessary to establish a normative, organ- or voxel-wise, database based on the images derived from healthy controls. However, to create such a database, a significant amount of data must be collected first in light of the public concerns over ionizing radiation. Radiation exposure from PET/CT imaging, as measured by the effective dose to a subject, scales with the amount of injected PET tracer activity. The new PET/CT systems, with their increased sensitivity, allow for further reduction in injected tracer activity and, subsequently, reduction in radiation exposure [7, 10, 11].

Prior studies of low-dose PET/CT imaging have indicated potential for their adoption in clinical routine. For example, Calderón et al. demonstrated that decreasing levels of injected [18F]fluoro-2-deoxy-D-glucose (FDG) activity ranging from 3.0 MBq/kg to 0.125 MBq/kg affected the mean standardized uptake values (SUVs) by only 8% or less [11]. Kertész et al. investigated the effects of reducing the injected [18F]FDG activity in pediatric oncology patients undergoing whole-body PET/CT examinations and showed that the injected activity levels can be reduced to 75% of the original dose without compromising the PET image quality [12]. Prieto et al. evaluated the impact of a 30% FDG dose reduction on image quality, resulting in steady clinical reading confidence despite a slight reduction in image quality [13]. Taken together, these studies either focused solely on deriving low-count PET images from standard activity images via list-mode resampling rather than using actual low-activity injections or they relied on the improved sensitivity of TB-PET systems, which are not yet widely available in medical facilities. Adding to the above research, Tan et al. compared ultra-low-dose and half-dose [18F]FDG-TB-PET/CT imaging in a test–retest setup within a 72-h time frame [14]. However, this study focused primarily on parametric imaging and assessed SUVs exclusively in the liver, thereby neglecting other organs and the continuous predominance of semi-quantitative SUV evaluations over kinetic modeling in clinical routine [15].

Our study, preceding the installation of a TB-PET/CT system, assesses the impact of reduced PET tracer doses on quantitative organ-based SUV_BW measurements, especially in healthy controls. Focusing on [18F]FDG imaging, we seek to understand the effects of lowering injected tracer doses on healthy organ evaluations and disease-related metabolic changes. Using serial [18F]FDG injections (test/retest), we examined intra-subject variabilities over 90 min and compared lesion uptake variations between low-dose (LD) and standard-dose (STD) PET scans in lung cancer patients. The goal was to create a standard organ-SUV_BW database for analyzing metabolic discrepancies due to diseases [9] and reduce radiation concerns from PET/CT scans.

This study evaluated variations in organ uptake in [18F]FDG-PET/CT images of healthy controls and lung cancer patients undergoing a dual-injection, dual-scan protocol. We demonstrate similar uptake values in key organs for both LD- and STD-PET imaging, with an exception in the heart on a group-based level. Intra-subject variabilities were highest in the brain (7%), skeletal muscles (8%), and heart (20%). All mean SUV_BW were comparable with previously recorded literature values (Figure 7) [22–24]. While our study suggests equivalence of LD- and STD-PET imaging protocols for organ-based quantification in healthy controls, care must be taken when assessing patients since the LD protocol yielded a lower detection rate of actual lesions (Figure 6).

In the present study, test and retest [18F]FDG PET/CT scans were set apart by 90 min (Figure 1), without significant differences in the group mean SUV_BW in key organs (Figures 2, 4) in both healthy controls and lung cancer patients. The intra-subject variability in organ uptakes between LD and STD was also explored. Significant changes in the mean uptake values from test to retest scans of healthy controls were observed in the brain, heart, adrenal glands, and, to a lesser extent, in skeletal muscles (Figure 3; Table 3, Supplementary Table S1). Physiological changes in both the brain and heart can affect the SUVs measured from two PET scans acquired at different time points. This may include changes in blood flow, metabolism, and cardiac function [32–36], which are most likely to occur within the 90 min between the two scans. Notably, in all healthy controls, skeletal muscle uptake was somewhat higher in the test (LD) than in the retest (STD) acquisition (8%, Table 3). This decrease in muscle uptake during the retest protocol could indicate reduced stress and tension levels [37] in participants, who may have relaxed after undergoing the protocol once before. The intra-subject differences in brain, adrenal glands, and skeletal muscle SUVs for LD/STD acquisitions were found only in the healthy cohort (Figure 5; Table. 5), likely because of the increased variability provided by its larger cohort size than that of the patients (Table 1). Inaccurate segmentations may have altered some results as well. For example, adrenal gland uptakes might have been affected by segmentation errors due to their small size and low contrast with surrounding tissues, posing a challenge in distinguishing them from other structures, such as the kidneys and the liver [18], in the LD-CT image.

Imaging of lung cancer patients showed similar mean SUV_BW between LD and STD scans in all segmented organs. However, fewer FDG-avid lesions were identified on LD-PET images than on STD-PET images (Figure 6). Given the increased image noise levels in LD imaging, only 10 lesions were delineated from the LD images (at 10% activity injection), while 17 lesions were subsequently identified on the STD acquisitions. Specifically, delineations of seven smaller lesions (<2 cm³) were not possible from LD images, given the increased noise level (Figure 6). In addition, the volumes of lesions were generally smaller in segmentations obtained from LD images due to reduced image quality. Nonetheless, the SUV_BW values of the corresponding lesions were similar in the LD and STD acquisitions, with a mean %difference equal to 10% (Table 6).

Overall, these findings suggest that low-dose FDG-PET/CT imaging may be a valuable option for reducing radiation exposure in FDG-PET/CT imaging for composing a normative database of healthy control values [9]. Our SUV_BW readouts for both STD and LD acquisitions were similar to published literature values [22–24]. Although variations in the mean SUV_BW values in organs with high metabolic activity and glucose turnover, such as the brain and the heart, were observed, the results were still consistent with the references in terms of both mean values and the minimum and maximum ranges of SUV_BW reported (Figure 7).

Effective doses from STD-PET were also consistent with established references for the standard clinical practice [28, 29]. Administering an activity that is 90% lower than that of the standard dose resulted in a 67% reduction in the total effective dose in healthy controls, thus effectively addressing concerns regarding radiation exposure in FDG-PET/CT imaging, particularly for non-clinical indications.

Exposure from CT plays a significant role in the overall effective dose during a standard examination. Using contrast-enhanced dual-phase CT (STD-CT) in cancer patients contributed 79% to their total effective dose, which, instead, was drastically reduced with the low-dose CT (2 mSv) protocol (Table 7). Mostafapour et al. demonstrated that the radiation dose in CT imaging could be further reduced from an effective dose of 2.6 mSv to less than 0.1 mSv by incorporating a tin filter for noise reduction [31]. However, potential artifacts from noise amplification during CT-based attenuation and scatter correction await further study.

Our study has several limitations. First, it was constrained by its small sample size. A larger cohort of participants could yield more reliable statistical results. Lesion delineation was performed by a single clinician, thus introducing a subjective bias into volume segmentations. The use of average delineations by multiple clinicians would offer a more precise reference. Next, LD PET imaging resulted in lower image quality, compromising its validity for clinical indications. Our study did not explore techniques to reduce noise in LD-PET images, such as AI-based image denoising methods [38, 39] or adjusted image reconstruction parameters for enhanced diagnostic accuracy [12, 40]. In our study, the retest protocol started 90 min after the test injection, followed by an additional hour of dynamic acquisition before static reconstruction. During this period, a portion of LD activity remains undecayed and may exert a minor influence on the subsequent quantification of STD uptakes. Considering our initial injected LD activity of 28 MBq and the behavior of [18F]FDG kinetic signals at long uptake times [41–43], we anticipate an impact smaller than 5% at the time of STD static acquisition on our uptake quantifications. Last, the present study focused solely on the analysis of static images, neglecting the dynamic information of PET images. For instance, Liu et al. demonstrated that whole-body dynamic PET imaging with a 10-fold reduction in injected activity could provide relevant kinetic metrics of [18F]FDG and comparable image contrast to full-activity imaging [44]. The parametric assessment of LD-PET could contribute valuable information to the evaluation of a reference database of normal PET values in healthy controls.

The study demonstrated that a reduction of 90% in the administered [18F]FDG activity is feasible for semi-quantitative whole-body PET/CT imaging without loss of accuracy of organ-based SUV_BW assessment. LD and STD injections provided comparable mean SUV_BW of organs in both healthy controls and lung cancer patients, except in organs with fast a [18F]FDG turnover. However, LD images did not provide sufficient clinical quality for the diagnostic assessment of lung cancer patients. Thus, our study supports the general adoption of LD-PET/CT imaging data for imaging healthy controls for the purpose of building an organ-based normative database.