Introduction

Substantial radiation dose reductions with maintained or improved image quality have been demonstrated with adult contrast-enhanced computed tomography (CT) [1–3] and CT angiography (CTA) [4–9] by lowering CT tube potentials. With smaller cross sectional areas, kVp (kilovoltage peak) reduction has even more potential for dose savings in children as the increased noise from lower kVp is less pronounced in smaller patients and can be compensated for by improved contrast [10–14].

Recent studies have shown a trend towards the use of lower tube potentials in paediatric examinations by manual adjustment of the imaging technique [15]. However, manual adjustments in tube potential can be challenging as they need to account for patient size and examination type at the time of imaging. While reducing tube potential decreases dose and increases iodine contrast, lowering kVp also increases image noise, and hence, an appropriate increase in tube current time product (in milliampere-seconds, mAs) is needed in order to maintain image quality [6, 10, 13].

An automatic tube potential selection tool (ATPS), CARE kV (Siemens Healthcare, Forchheim, Germany), has been recently introduced to facilitate kVp reduction by automatically selecting an optimized combination of tube potential and tube current based on patient body habitus, body part to be imaged, type of examination being performed, and desired image quality [10, 16].

Recent investigations of ATPS addressing adult contrast-enhanced CT and CTA have shown consistent tube potential reduction and radiation dose savings [4, 17–21]. Siegel et al. recently reported a mean 56 % dose reduction in paediatric phantom CTA [22] and a median 27 % dose reduction in paediatric chest and abdominal CT and CTA using ATPS [23]. However, previous studies have been limited with regard to the range of paediatric body sizes investigated and lack of systematic comparison of ATPS results among different examination indication settings. Furthermore, most of these studies have not considered the effects of parameters such as pitch on ATPS, and relied on contrast-to-noise ratio (CNR) as the predominant objective image quality metric.

The purpose of our study was to assess the effects of ATPS on radiation dose, image quality and lesion detectability in paediatric abdominopelvic CT and CTA across a range of paediatric body sizes and examination indications in a phantom. Furthermore, in this study we investigated technique modifications while using ATPS and the resultant impact on dose, image quality and lesion detectability.

Materials and methods

No Institutional Review Board approval was required for this phantom study.

Paediatric phantom

A proprietary modular phantom consisting of five different diameter cylindrical sections was used for this investigation (Fig 1a). The four smallest diameter modules of 12 cm, 18.5 cm, 23 cm, and 30 cm, with approximate water equivalent diameters of 11.2 cm, 17.7 cm, 22 cm, and 29 cm, simulating neonate, young child, adolescent, and young adult abdominal diameters, respectively, were used [24]. Each module contained five radially arranged rod inserts, each with a diameter of 25.4 mm and a thickness of 30 mm; the iodine (8.5 mg I/cc) and polystyrene (the soft tissue insert, providing subtle contrast of roughly 50 HU compared to the polyethylene phantom body), were targeted for analysis in this study (Fig 1b).

Fig. 1
figure 1

Photographs of the phantom. The modular phantom (a) consists of five different diameter cylindrical modules with intervening tapered sections; the four smallest diameter modules of 12 cm, 18.5 cm, 23 cm, and 30 were used in this study. Each module contains five radially arranged rod inserts (b); I = Iodine (8.5 mg I/cc); P = Polystyrene (soft tissue); B = Bone; W = Water; A = Air

Full size image

Automatic tube potential selection (appendix 1)

A commercially available tool for ATPS, CARE kV (Siemens Healthcare, Forchheim, Germany), was used for this study. CARE kV automatically selects a combination of tube potential and tube current according to patient size, prescribed image quality, and examination indication. The examination indication is defined with an incremental slider bar with settings 1 to 12, reflecting progressively increased material (i.e., iodine) attenuation from decreasing kVp. In our study, setting 3 was chosen for non-contrast, setting 6 for contrast-enhanced, and setting 9 for CTA examinations (Fig 2).

Fig. 2
figure 2

Axial CT images of the 18.5 cm module demonstrating the impact on iodine (I) and soft tissue (P) inserts scanned (a) with standard protocol without ATPS (120 kVp), (b) with ATPS in CTA mode (80 kVp chosen by ATPS), and (c) with ATPS in non-contrast mode (100 kVp chosen by ATPS). Note that, while noise increased in the lower kVp scans (3.2 in A, 4.3 in B, 3.4 in C), increased iodine insert contrast (311.4 in A, 483.0 in B, 373.3 in C), and to a lesser degree increased soft tissue insert contrast (90.9 in A, 124.1 in B, and 103.7 in C), was demonstrated at lower kVp

Full size image

CT technique

All CT images were obtained with spiral technique using the same 128-slice dual source multidetector CT system (Somatom Definition Flash, Siemens Healthcare) equipped with AEC to modulate the tube current (CARE Dose 4D, Siemens Healthcare), ATPS (CARE kV), and Sinogram Affirmed Iterative Reconstruction (SAFIRE) technology. Experimental imaging parameters are provided in Table 1.

Table 1 Experimental imaging parameters
Full size table

Experiment 1. Routine pitch

The 12, 18.5, 23, and 30 cm phantom modules were examined independently and identically with a pitch of 1.4, defined as routine pitch, using: 1) routine clinical weight-based paediatric abdominopelvic protocol with a fixed 120 kVp (AEC enabled), and 2) ATPS, with three ATPS examinations performed on each of the four modules (slider bar positions 3, 6, and 9). Each diameter module was examined independently such that the localizer and resultant image adjustments were determined from a single module diameter.

Experiment 2. High pitch

Experiment 1 was repeated using a pitch of 3.0 on all scans.

Experiment 3. Full phantom scan

All four modules were imaged with a single scan at a pitch of 3.0 using the routine non-ATPS protocol, followed by three ATPS examinations (positions 3, 6 and 9). In these full phantom scans, the resultant scan adjustments were determined based on a localizer of the entire phantom including all four modules, rather than from a localizer of the individual modules.

Radiation Dose and Effective Milliampere Seconds

The selected kVp, volume CT dose index (CTDIvol 32cm), and effective mAs were recorded for each scan. CTDIvol 32cm in mGy was used for comparison of delivered radiation dose estimate, and radiation dose subsequently refers to this estimate.

Image Analysis (Appendix 2)

Image analysis was performed with a software program developed specifically for the proprietary phantom [25]. Semiautomated measurements and calculations of iodine and soft tissue insert contrast, noise, CNR, and a previously validated detectability index, d’, were performed [26]. The detectability index assesses the likelihood of detection of a lesion based on lesion contrast and size (in this study set for the detection of a reference 5 mm lesion, 50 Hounsfield Unit feature contrast), lesion edge profile, system resolution, and noise texture, is a more comprehensive and clinically relevant metric of image quality than CNR alone, and has been demonstrated to correlate with human observer performance [26]. A figure of merit, d’2 / CTDIvol, was further calculated for the iodine and soft tissue inserts to assess lesion detectability per radiation dose.

Statistical analysis

The results of the low and high pitch experiments were analyzed independently by fitting a model of the form:

$$ \log \left({Y}_{ij}\right) = \mu + \beta {s}_j + {b}_j + {\epsilon}_{ij} $$

where Y ij is the outcome (CTDIvol, d’ or CNR) for the i-th module diameter (12, 18.5, 23, 30 cm) and j-th exam setting (routine, ATPS 3, 6, or 9). The model explains the variability in response in terms of the baseline mean μ (which represents the expected response for a phantom of size 12 cm using the routine weight-based protocol and 1.4 pitch), the coefficient of the size term βs i , and the effect of exam setting b j . Lastly, ε ij represents measurement error, assumed to have a zero mean Gaussian distribution. The log transform (logY ij ) was used for analysis of d’ and CNR to ensure that the response conformed to model assumptions. A P < 0.05 was considered a significant difference.

Results

Experiment 1. Routine pitch

Tube potential was reduced from 120 kVp to either 100 kVp or 80 kVp in all ATPS examinations, with greater tube potential reductions with the smaller phantom diameters and CTA examinations. As expected, average effective mAs increased with all ATPS examinations compared to the routine protocol (Table 2).

Table 2 Experiment 1. Routine Pitch (1.4): Imaging parameter and image quality results of routine protocol with fixed 120 kVp versus automatic tube potential selection protocol
Full size table

The reduction in kVp was accompanied by a reduction in CTDIvol in all ATPS examinations, with progressive dose reduction in non-contrast, contrast-enhanced, and CTA examinations (Fig. 3). Dose reductions ranged from 27.4–52.4 % in the CTA mode (P < 0.0001), 15.3–32.8 % in the contrast enhanced mode (P = 0.0003), and 8.0–13.6 % in the non-contrast mode (P = 0.03), with the greatest dose reductions in the smaller phantom diameters for each examination type (Table 2; statistical analysis results available in online supplementary material).

Fig. 3
figure 3

Routine Pitch (1.4). Bar graph of radiation exposure, as measured by the volume CT dose index (CTDIvol), relative to examination type for each of the four module diameters using routine (weight-based abdominopelvic pediatric CT protocol with fixed 120 kVp) and ATPS (with variable kVp) protocols. Non-contrast, contrast-enhanced, and angio refer to ATPS exams at slider bar positions 3, 6, and 9, respectively, on the CARE kV interface. Dose reductions were observed with all ATPS examinations (14–52 % reductions with 12 cm module, 13–51 % reductions with 18.5 cm module, 11–46 % reductions with 23 cm module, and 8–27 % reductions with 30 cm module), with the largest dose reductions observed with the smaller modules and angio mode

Full size image

Metrics of image quality were either improved or maintained with ATPS. There was a statistically significant increase in CNRiodine for non-contrast and contrast-enhanced ATPS examinations (P = 0.019 and P = 0.017, respectively), and a statistically significant increase in d’iodine for non-contrast and contrast-enhanced ATPS examinations (P = 0.019 and P = 0.018, respectively). Increases in CNRsoft tissue and d’soft tissue in 83 % and 75 % of ATPS exams, respectively, did not achieve statistical significance (Table 2).

The figures of merit d’2 iodine / CTDIvol and d’2 soft tissue / CTDIvol were increased in all ATPS examinations, consistent with better image quality per dose (Fig. 4a, b).

Fig. 4
figure 4

Figure of Merit. Bar graph of the figure of merit for the (a) iodine insert (d’2 iodine / CTDIvol) and (b) soft tissue insert (d’2 soft tissue / CTDIvol) in the routine pitch experiment. At each module diameter, lesion detectability per unit of radiation dose (mGy) was increased with all ATPS examinations compared to the routine protocol. Non-contrast, contrast-enhanced, and angio refer to ATPS exams at slider bar positions 3, 6, and 9, respectively

Full size image

Our analysis also allows for an assessment of the overall effect of phantom diameter on dose and image quality, with a statistically significant increase in CTDIvol (P < 0.0001) and a statistically significant decrease in image quality (P < 0.0001 for d’iodine and soft tissue and CNRiodine and soft tissue) with increasing phantom size (Table 2).

Experiment 2. High pitch

With the high pitch (3.0) protocol, there was more limited kVp reduction with ATPS, with selection of a reduced kVp (100 kVp) in only 3/12 (25 %) ATPS examinations (Table 3). Dose reductions of 31.1–35.2 % were observed in these three lower kVp examinations (Fig. 5), but an overall statistically significant dose reduction was not achieved given the lack of kVp reduction in the remaining examinations (Table 3; statistical analysis results available in online supplementary material). In the 9/12 ATPS examinations without kVp reduction, both the selected kVp and effective mAs were essentially unchanged from the routine protocol, thus there were no differences in radiation dose (Fig. 5).

Table 3 Experiment 2. High Pitch (3.0): Imaging parameter and image quality results of routine protocol with fixed 120 kVp versus automatic tube potential selection protocol examinations
Full size table
Fig. 5
figure 5

High Pitch (3.0). Bar graph of radiation exposure, as measured by the volume CT dose index (CTDIvol), relative to examination type for each of the four module diameters using routine (weight-based abdominopelvic paediatric CT protocol with fixed 120 kVp) and automatic tube potential selection (with variable kVp) protocols. Non-contrast, contrast-enhanced, and angio refer to ATPS examinations at slider bar positions 3, 6, and 9, respectively, on the CARE kV interface. Dose reductions were observed in only 25 % of the high pitch scans, with dose reductions with the 12 cm module (31 % in contrast-enhanced mode and 35 % in angio mode) and the 18.5 cm module (35 % in angio mode)

Full size image

While the ATPS examinations performed at 100 kVp did demonstrate a trend towards improved CNRiodine, d’iodine, d’2 iodine / CTDIvol and d’2 soft tissue / CTDIvol, overall image quality was maintained but not improved, with no statistically significant difference in d’iodine or soft tissue and CNRiodine or soft tissue between the ATPS and routine protocol examinations (Table 3).

Experiment 3. Full phantom

ATPS did not reduce tube voltage in the full phantom scans. Without a kVp reduction, there was effectively no difference in effective mAs, dose, image noise, insert contrast, or lesion detectability between the ATPS and routine protocol examinations (Table 4).

Table 4 Experiment 3. Full Phantom: Imaging parameter and image quality results of routine protocol with fixed 120 kVp versus automatic tube potential selection protocol examinations
Full size table

Discussion

Our results demonstrate that ATPS technology can be an effective tool to reduce radiation dose while maintaining or improving image quality in paediatric abdominopelvic CT and CTA. However, the way in which this technology is used can impact dose reduction and image quality. The results of our routine pitch experiment are in agreement with previous ATPS phantom studies demonstrating the greatest dose reductions with CTA exams and smaller phantom sizes [4, 22]. The 52.4 %, 51.1 %, and 46.4 % dose reductions with the CTA examinations in the 12 cm, 18.5 cm, and 23 cm modules are similar to recently published results of a mean 56 % dose reduction in paediatric phantom CTA [22]. These findings have important clinical implications regarding potential cancer risks in young patients given the generally recognized increased susceptibility of the youngest, smallest children [27–30]. Moreover, while not tested in prior investigations, we also found substantial dose reductions with the larger phantom diameters and with the contrast-enhanced and non-contrast-enhanced ATPS scans, supporting ATPS applicability to a range of paediatric body sizes and clinical indications.

To our knowledge, this is the first systematic study of ATPS across a range of paediatric sizes and selected examination indications (non-contrast, contrast-enhanced, and angiography). This study provides a more comprehensive and clinically relevant assessment of image quality, as the detectability index, in addition to the standard CNR, was measured and compared for both the iodine and soft tissue inserts. We chose to evaluate both inserts on all scans, thereby assessing the effect of the CTA setting not only on the iodine insert, but also on soft tissue contrast and lesion detectability, and vice versa. We feel this approach affords an expanded and more clinically appropriate image quality analysis. In addition to maintaining or improving CNR for the iodine and soft tissue inserts in all ATPS scans, figures of merit d’2/ CTDIvol were also increased in all ATPS scans, consistent with improved lesion detectability per radiation dose.

In experiment 2 we examined variations on pitch for ATPS. In 75 % of the high pitch ATPS examinations, there was no kVp reduction and, hence, no dose savings, likely a result of limitations in tube power for the high pitch scans. Lowering the kVp typically requires an increase in mAs in order to maintain the desired CNR, and when the existing x-ray tube generator is unable to achieve the necessary mAs, ATPS automatically selects the next highest kVp that can provide the desired image quality. In the high pitch, dual source scan, the maximum effective mAs that the system can reach is limited by a faster rotation time and higher pitch, and thus may not be able to achieve the increased mAs required to examine at a lower kVp. Schindera et al., demonstrated the limitations of maximum tube current on automatic tube potential reduction by using a progressively smaller pitch value in order to perform a lower kVp examination [4]. While increasing pitch is often considered a means for dose reduction, use of a higher pitch may restrict the ability of ATPS to lower kVp and thus may limit ATPS dose savings in certain situations. Application of the results of this investigation can help guide examination performance using ATPS based on patient size, region scanned, and indication. For example, high pitch thoracic CTA in an infant using ATPS may allow for kVp and dose reductions given the small patient size, improved iodine contrast at lower kVp, and the lower effective mAs in thoracic imaging. However, in an older and larger child, the high pitch mode may preclude full realization of ATPS image quality and dose benefits due to tube current limits; using a lower pitch with ATPS may allow for kVp and resultant dose reductions.

The lack of tube voltage adaptation with the full phantom examinations following a full phantom topogram (experiment 3) indicates that the ability of ATPS to reduce kVp is limited with combination CT exams covering multiple body regions of variable diameters where the thickest sections will determine kVp for the entire scan thus limiting kVp reduction. Similar to experiment 2, tube current limitations likely prevented the use of a lower kVp due to insufficient tube power reserve; hence, the reference tube potential (120 kVp) was unchanged and exposure adaptation was due solely to the AEC. Our results stress the importance of limiting the scan area to the region of interest in addition to considering split versus combination examinations in order to maximize kVp reduction. For example, including the neck and chest in a single scan would be a suboptimal use of ATPS technology as kVp adaptation would be biased towards using a higher kVp due to the larger attenuation of the shoulder region.

Our study has limitations that warrant consideration. First, this a phantom study, so actual z-axis and angular variations in patient thickness and organ attenuation that may affect tube potential and tube current modulation would not be accounted for in this study. Second, the experiments were performed on a single CT scanner using a specific manufacturer’s proprietary automatic tube potential selection technology. Third, estimated CTDIvol was used for comparison of radiation dose. While we demonstrate ATPS reductions in CTDIvol, a measure of scanner output that should translate into decreased radiation dose to the patient, we did not evaluate estimated patient effective dose or actual dose to the phantom. The use of CTDIvol 32 cm (as opposed to CTDIvol 16 cm) would underestimate the absolute value of the radiation exposure for the smaller phantom diameters, but percent dose reductions, which were the focus in this study, would remain unchanged. Fourth, we did not perform a qualitative observer evaluation of image quality. However, a strength of our study is the more comprehensive quantitative evaluation of image quality using both CNR and lesion detectability, which would translate to perceptual changes; recent ATPS studies have shown maintenance of diagnostic quality images at reduced kVp [17–21, 23]. Fifth, we selected 120 kVp as the routine protocol kVp and ATPS reference kVp for all examinations. In many institutions, routine weight-based protocols would call for a kVp of less than 120 in smaller paediatric patients [15], but since we were imaging and comparing a range of phantom diameters (including a young adult equivalent) we elected to use 120 kVp. Finally, our selections of positions 3, 6, and 9 to represent non-contrast, contrast-enhanced, and CTA examinations, respectively, differ slightly from the default labels provided on the CARE kV interface (where 3 = non-contrast, 7 = contrast-enhanced liver, and 11 = CTA). The slider bar selection reflects a trade-off between iodine contrast gain and image noise, and our selection of position 9 for CTA reflects a preference for a lower noise CTA examination in our actual practice. Selecting a higher contrast gain (slider bar position >9) would weight the examination towards improved iodine contrast, lower kVp, and potentially more dose savings.

ATPS can reduce radiation dose and maintain image quality across a range of paediatric abdominopelvic examination indications and body sizes, and, where available, this technology can be easily implemented into everyday clinical practice. Using ATPS, dose reductions can be achieved from already low dose protocols and even greater dose reductions are anticipated in the non-paediatric hospital setting.