Optimal tube voltage for abdominal enhanced CT in children: a self-controlled study

Background

The application of computed tomography (CT) examinations is widespread and the number of examinations continues to increase (1). It has been noticed that due to the smaller body sizes of children compared to adults, it is possible to use low-kV CT to improve image contrast and further reduce the possibility of radiation dose; it is also well-known that a lower kV is limited in a larger body weight due to excessive image noise and artifacts (2–5). The use of weight-adapted pediatric CT protocols has been suggested, but a consensus standard has not been established and clinical available studies are not adequate (6–9). The purpose of our study was to evaluate the effect of tube voltage on the image quality of abdominal enhanced CT scans in children with follow-ups and to determine the optimal patient weight-dependent kV selection for abdominal enhanced CT scans in children.

Material and Methods

CT data acquirement

CT data were collected using a HDCT with gemstone detector (GE Discovery 750) with helical pitch set at 1.375:1 and tube rotation speed of 0.4 s. For every patient, the three examinations were performed at three different tube voltages of 120 kV, 100 kV, and 80 kV. The fixed tube current was used and was adjusted to ensure that the radiation doses (calculated as CTDIvol) were identical in each examination. The radiation dose level in terms of the volumetric CT dose index (CTDIvol) was determined according to the experience of our hospital and was set according to children’s weight: in the range of 0.92–2.68 mGy (Table 1). For patients weighing <35.0 kg, small scan field of view (FOV) was used and for patients weighing > 35.1 kg, large scan FOV was used. The raw data obtained were reconstructed into an image of a 512 × 512 matrix, with a section thickness of 5 mm and spacing of 5 mm. Reconstruction was performed using an adaptive iterative reconstruction algorithm at 40% strength (ASIR 40%), with a standard reconstruction kernel. The scan range was from the top of diaphragm to the anterior superior iliac spine, including the liver, spleen, pancreas, and kidney. For those children who were too young to cooperate, sedation with oral chloral hydrate (10%, 0.5 mL/Kg) was applied before the scanning.

OPEN IN VIEWER

Table 1.

The scanning protocol and contrast injection protocol.

Body weight (kg)	Tube current (mA)			Radiation dose (CTDIvol, mGy)	Contrast agent
	80 kV	100 kV	120 kV		Dose (mL)	Injection rate (mL/s)	Injection time (s)
6.1–8.0	150	80	50	0.92	15	0.7	22
8.1–12.0					16	0.7	23
12.1–15.0	195	100	65	1.22	20	0.9	23
15.1–20.0					26	1.1	24
20.1–24.0	240	130	80	1.52	29	1.3	23
24.1–28.0					32	1.4	23
28.1–32.0	300	160	100	1.88	36	1.5	24
32.1–35.0					40	1.7	24
35.1–50.0	360	190	120	2.68	56	2.4	24

Subjective image quality assessment

Images of all scans were transmitted to a GE AW4.5 CT workstation. All children’s related information and scanning parameters were hidden during the scoring process. The image review was randomized in terms of tube voltages. The reviewers could adjust the window width and window level to the position deemed appropriate but were not allowed to access the prior studies (tube voltages). Two pediatric radiologists (with 13 years and four years of diagnostic experience) evaluated the image quality according to the scoring standard at the same time. In cases when different scores were given, a unified result was achieved after discussion. If no agreement could be achieved, the lower score was used.

The image qualities evaluated include: (i) subjective scoring, including the level of image enhancement and overall image quality; and (ii) Objective image noise measurement and calculated CNR.

The subjective image quality was evaluated using a scoring system of 1–4, with 3 representing a satisfying imaging quality. Subjective evaluation included score of image enhancement level and overall image quality. The level of image enhancement referred to the degree of contrast between the various tissues in CT images. It was scored 1–4 according to whether the contrast was obvious between the tissues, whether the boundary was clear, and whether all the structures could be distinguished. The overall image quality was scored mainly based on the image noise, the level of beam-hardening artifacts, and physician’s confidence to make diagnosis using the image (Table 2). The final score was given according to comprehensive analysis of image enhancement level and overall image quality.

OPEN IN VIEWER

Table 2.

Subjective evaluation standard.

	Score point
Score item	4	3	2	1
Image enhancement
Contrast obviously between the tissues	Yes	Yes	Not enough	No
Structures could be distinguished for diagnosis	Yes	Yes	Not enough	No
Boundary clear of tissue	Yes	Not enough	No	No
Overall image quality
Image noise	Less	Acceptable	More	Unacceptable
Beam-hardening artifacts	No affect	No affect	Affect	Affect
Diagnosis confidence	Strong	Acceptable	Less	Unacceptable

Results

Subjective image quality evaluation

The subjective scores are shown in Table 3 and Fig. 1. In the evaluation of subjective scores, the 80-kV images had the best enhancement level with 4 points for all images. Enhancement was reduced in the 100-kV images and was the lowest in the 120-kV images (Fig. 2). The enhancement was not correlated with weight. In terms of overall image quality, the 80-kV images had the best quality with an average score of 3.38 ± 0.63. The quality of 100-kV images decreased to 3.31 ± 0.58 and the quality of 120-kV images were the lowest at 3.16 ± 0.57. Small but statistically significant differences in overall image quality were detected among different voltage settings (F = 3.08, P = 0.047). In different weight subgroups, the 80-kV images of the children with 12.1–28.0 kg had the best quality, with a statistical difference detected compared to the other groups. For children in the 28.1–35.0 kg and 35.1–50.0 kg groups, the 100 kV and 120 kV generated the highest image quality scores of 3.40 and 3.00, respectively, in that group. Due to excessive image noise and beam-hardening artifacts, image quality with <3 points was noticed in the 80-kV images of children weighing 28.1–50.0 kg (Fig. 3).

OPEN IN VIEWER

Table 3.

The average score of subjective evaluation.

	Contrast enhancement					Overall image quality
	80 kV	100 kV	120 kV	F	P	80 kV	100 kV	120 kV	F	P
Total	4.00 ± 0.00	3.18 ± 0.39	3.00 ± 0.00	487.01	<0.01	3.38 ± 0.63	3.31 ± 0.58	3.16 ± 0.57	3.08	<0.05
6.1–12.0 kg	4.00 ± 0.00	3.60 ± 0.50	3.00 ± 0.00	60.17	<0.01	3.20 ± 0.41	3.35 ± 0.49	3.20 ± 1.01	0.32	0.73
12.1–20.0 kg	4.00 ± 0.00	3.00 ± 0.00	3.00 ± 0.00	–	–	3.76 ± 0.44	3.28 ± 0.45	3.28 ± 0.45	11.20	<0.01
20.1–28.0 kg	4.00 ± 0.00	3.00 ± 0.00	3.00 ± 0.00	–	–	3.77 ± 0.44	3.65 ± 0.49	3.06 ± 0.24	14.81	<0.01
28.1–35.0 kg	4.00 ± 0.00	3.40 ± 0.52	3.00 ± 0.00	28.50	<0.01	3.00 ± 0.47	3.40 ± 0.52	3.10 ± 0.32	2.21	0.13
35.1–50.0 kg	4.00 ± 0.00	3.00 ± 0.00	3.00 ± 0.00	–	–	2.46 ± 0.52	2.73 ± 0.79	3.00 ± 0.00	2.76	0.80

OPEN IN VIEWER

Fig. 1.

Subjective scoring of images. Left panel shows the results of all cases. The red points and lines indicate the degree of enhancement. The blue ones indicate the overall image quality. As the scanning voltage increases, the scores of enhancement level and overall image quality decrease, with the decline of enhancement more obvious. In subgroups of different body weights, the 6.1–28.0 kg range showed a similar trend, while the overall image qualities of 80 kV in the 28.1–50 kg group were significantly lower than those of the 100-kV and 120-kV images.

OPEN IN VIEWER

Fig. 2.

A three-year-old boy weighing 11 kg underwent three enhanced CT scans with an interval of 1–2 months between each one. (a) An 80-kV image with a CTDIvol of 1.21 mGy; (b, c) the 100-kV and 120-kV images at the same dose. The contrast enhancement of these three figures were 4.0, 3.0, and 3.0; overall image quality of these three figures was 3.0, 3.0, and 4.0, respectively. The image shows that the difference in image noises among the three voltages is not obvious. All images clearly display the abdominal solid organs, of which the 80-kV image has the best contrast, with the most obvious enhancement of the descending aorta and portal vein, the vessel, and other tissues. Boundaries in (a) are clearer than in (b), and especially clearer than in (c). The enhancement helps to give a high overall image quality score.

OPEN IN VIEWER

Fig. 3.

Images of a 10-year-old girl weighing 36 kg. (a) The 80-kV image with a CTDIvol of 2.62 mGy; (b, c) the 100-kV and 120-kV images at the same dose. The contrast enhancement of these three figures were 3.0, 3.0, and 2.0; overall image quality of these three figures were 2.0, 3.0, and 3.0 respectively. The 80-kV image is with the best degree of enhancement, but also with more observable beam-hardening artifacts (black arrow). As the voltage increases, the image contrast gradually decreases, and the hardening artifacts also gradually decrease.

Objective image quality

Objective measurements are shown in Fig. 4 and the Supplementary Material. In terms of objective measurements, the CT value and the δCT value were the best at 80 kV and decreased with the increase of the voltage. The body weight had no significant effect on the results. These results were similar to the subjective enhancement level. As for the noise measurement, no significant change was noticed in body weight range of 0–28 kg. When body weight was >28 kg, the image noise significantly decreased with increased voltage. The CNR had the opposite change from SD, with the best CNR identified in 80-kV images of children with a body weight of 6.1–28.0 kg and decreased with voltage. Except for the kidney and aorta in the 28.1–50.0 kg group and the liver in the 35.1–50.0 kg group, there was no significant difference in CNR detected in all tissues of children weighing >28.1 kg.

OPEN IN VIEWER

Fig. 4.

Objective measurement results, including CT values (a), SD (b), δCT values (c), CNR values (d) in overall populations and subgroups with different body weights. Yellow point and line instead of aorta, purple is the liver, green is the spleen, red is the pancreas, and blue is the kidney. The results showed that the CT value and the δCT value of 80-kV images were the best and decreased with the increase of the voltage. Body weight had no significant influence on the results. The slope of the noise-voltage curve in the 6.1–28.0 kg curve was without obvious change. Above 28.1 kg, as the voltage increased, the image noise was significantly reduced. The changes in CNR and SD were opposite. The CNR of the 80-kV images in the 6.1–28.0 kg group was the best and it decreased with the drop in voltage drop. No significant change of CNR was noticed in body weight groups ≥28.1 kg.

Discussion

The 80-kV, 100-kV, 120-kV, and 140-kV tube voltages are commonly provided by major CT manufacturers and represent the peak energies for X-rays generated by the X-ray tube. Reduction of voltage will cause exponential attenuation of X-ray flux and an increase of the tube current is needed to ensure the constant X-ray flux. For example, although the index algorithm varies for different manufacturers, the current required at 80 kV is about 120/80 × 10, to match the X-ray flux at 120 kV. On the GE HD 750 scanner, X is close to 3. Our results also confirmed that to ensure equal radiation doses at different voltages, large increase of the current was needed in low-voltage CT to form a “low-kV-high-mAs” CT protocol (Table 1).

A major advantage of low-kV imaging is the ability to increase the absorption of iodide ions to improve the enhancement effect of iodine contrast agents. The CT value of iodine increases with the decrease of photon energy or tube voltage. Lowering the tube voltage makes the average energy of mixed energy X-ray photons closer to the k-edge of iodine absorption (33 keV), thereby increasing the photoelectric effect, resulting in increases of the iodine attenuation (11,12), ultimately increasing the CT value of tissues under enhanced CT (13,14). This feature has been demonstrated in the current study. The CT values of measurement tissues in the 80-kV images increased by 27.53–49.16% compared to those in the 120-kV images, especially for the blood-rich organs such as the major blood vessels and kidneys, for which the CT valued increased by 44.97% and 49.16%, respectively, with significantly increased contrast (Fig. 2). The CNR of the images also increased at low voltage, the CNR of the 80-kV images increased by 42.98–65.11% compared to the 120-kV images, which substantially improved image quality without increasing the radiation dose. For many years, a tube voltage of 120 kV has been used by default in most CT examinations. It was not until recently that the use of low kV has slowly emerged (15–17). For example, low-kV scans were used to detect low-density lesions in solid tissues (18). Nyman et al. used a low concentration contrast agent combined with an 80-kV setting in a group of patients with acute pulmonary embolism and achieved similar degree of enhancement as the combination of conventional concentration of contrast agent with a 120-kV setting (19). Some manufacturers even introduced a CT protocol with a 70-kV setting (20). Low-voltage CT is suitable for smaller children. Since a low voltage would inevitably weaken the penetration of photons and increase image noise, it is generally not recommended for adults or for larger children (21). A study in a group of children indicated that while applying low kV to increase the contrast CT value, attention should be paid to changes in image noise and CNR to maintain image quality (3). These studies have shown that low-kV CT is suitable and should be applied in children; however, it still needs to be clarified with the suitable body size for low-voltage CT application. Although ICRP recommended a low-kV CT program for children based on previous studies (22), it was considered not completely appropriate in our experience. In general, there is still no specific application standard for low-voltage enhanced CT in children and it lacks parallel comparison between images obtained at various voltages. Therefore, the current study was designed to identify the suitable voltages for children with different body weights.

Many children with various solid tumors are admitted in our hospital for a clinical treatment plan that requires multiple enhanced CT scans to evaluate the effect of chemotherapy. These patients generally need 2–5 times more than the regular enhanced CT examination in one treatment period. CT examinations with different voltages, but similar radiation doses, allow the observation of different image characteristics. Since the radiation dose has not changed, it will not cause additional radiation damage to the child. At the same time, we used a fix tube current technique to instead of automatic tube current modulation technique as Singh’s research, to avoid excessive dose reduction in small patients and an increment in dose noted with larger patients (23). This self-controlled study can ensure the consistencies of body weight, body shape, hemodynamics, and organ metabolism to minimize statistical bias between different cases and therefore increase the reliability of the results. With the same radiation dose, the actual protocols used included “low kV-high-mAs” and “high-kV-low mAs” settings. While image contrast will be increased by using low tube voltage, it is likely that the image noise or beam-hardening artifacts would also be increased, which may affect image quality. Applications of these protocols in children undergo re-examination can reduce the impact on the overall diagnostic quality caused by the reduction of image quality. In the current study, although some of the children's images had overall quality < 3 points with the low tube voltage setting, the extent of the lesion and the degree of enhancement could still be clearly shown and images in general could still satisfy the clinical diagnostic requirement. When compared with the previous examination, the change of the tumor can still be judged.

In our study, we noticed that the CT values had obvious differences between different organs, such as between the aorta and the liver (e.g. 204.18 ± 27.07 vs. 128.42 ± 11.83 in 80-kV images). We therefore evaluated the δCT and CNR of different organs independently to avoid the bias of calculation. According to our results, the 80-kV image was the best in the subjective score for the enhancement level which was not affected by body weight. At the same time, image noise and beam-hardening artifacts for patients > 28.1 kg at 80 kV also increased dramatically. Since CNR was calculated using both the contrast enhancement in the numerator and image noise in the denominator, even though the higher contrast enhancement may offset the higher image noise to yield similar CNR values for different kV images of patients >28.1 kg, the higher image noise sometimes caused the images to be unsuitable for diagnosis. Therefore, considering image noise and beam-hardening artifacts, the subjective overall quality scores of the 80-kV images in some of the children >28.1 kg were <3, indicating that excessive noise and beam-hardening artifacts had affected the image quality which could not be ignored just because of the degree of enhancement. Taking the descending aorta as an example, the average noises of 80-kV images increased by 27.74% compared to the 120-kV images in total, while in children with a body weight of 6.1–12.0 kg, it was only increased by 16.24%, but dramatically increased by 46.09% in children with a body weight of 35.1–50.0 kg. Therefore, according to our research, it was not appropriate to use the 80-kV protocol for children > 28.1 kg.

There are several limitations in the current study. First, due to the limited number of children weighing > 28.1 kg, we only observed an increase of image noise and beam-hardening artifacts in children weighing 28.1–50.0 kg and concluded they were not suitable for the 80-kV scan. A higher voltage protocol needs to be selected; however, the optimal voltage for these children needs to be explored in further studies. Second, this study focused on image quality evaluation. Future studies should include the influence of a patient’s underlying medical condition and the diagnostic accuracy. Third, there is no clear guideline of how the children should be divided based on their weight. We referred to the “ICRP publication 121” (22) and combined our prior research to divide patients into five weight groups. Fourth, this study was based on low-dose CT scan protocol for solid tumor follow-up in our hospital. Other higher radiation dose protocols may reduce noise and improve image quality, which may expand the application range of the 80-kV protocol. Furthermore, we only used the ASIR 40% algorithm without consideration of higher weighted ASIR or a more advanced IR method of post-process construction, which has the potential to reduce image noise or reduce the effect of beam-hardening artifacts to improve image quality. These methods, in combination with other IR algorithms, are likely to further reduce image noise and expand the range of applications for 80-kV CT protocols. Finally, the HDCT scanner only provides the tube voltages of 80 kV, 100 kV, 120 kV, and 140 kV; the tube voltage of 70 kV is not available on this type of scanner. The 140-kVp voltage was not selected because the patient weight was limited to < 50 kg and because the contrast enhancement may not be optimal for small patients based on Yu’s study (9). The tube voltages of 80 kV, 100 kV, and 120 kV were selected sequentially in our study rather than randomized, which may introduce some bias. However, this potential bias was reduced with the blind and randomized reading of images of different tube voltages.

In conclusion, for abdominal CT examination in children, an 80-kV setting is suitable for children weighing <28.0 kg and higher tube voltages should be considered for children weighing >28.1 kg.

Supplemental Material