1. Introduction

With ongoing technological developments in radiation protection, CT has become integral to pediatric radiology, and has established itself as an important part of the diagnostic algorithm [1–6]. Nevertheless, the awareness of the possible effects of ionizing radiation in the young and growing bodies of children requires that radiation dose and scan protocols be adapted to size, age, and clinical needs [6–9], according to the ALARA (as low as reasonably achievable) principles. In addition to stringent justification requirements, continuous optimization is imperative. Well-established radiation dose reduction methods are available for every modern CT scanner, such as automatic tube current modulation [10–12] and tube potential optimization [12,13]. In addition, correct patient positioning in the scanner isocenter [12–14], individually adjusted scan boundaries, the choice of an anterior–posterior or posterior–anterior supine projection, and appropriate reduction of the Scan Projection Radiograph (SPR) dose [12,15], must be considered. Another very important issue is avoiding multiple scan series in pediatric CT whenever possible. If necessary, clinically unstable and non-cooperative children should be sedated to reduce movement artifacts and prevent repeated scans [9]. One of the most powerful optimization tools is to compare the doses delivered to Diagnostic Reference Levels (DRLs), which are also now becoming available for pediatric CT [16–20]. In most cases, DRLs represent the 3rd quartile of doses from dose surveys, indicating a level of dose below which 75% of all institutions operate. However, especially in pediatric CT, these dose levels have been lowered quite a bit recently, indicating that there is still some more potential for optimization. Nevertheless, these values cannot be thought of as representing the optimum values. Rather, they are values which, when continuously exceeded, should trigger a process to determine the reasons these unusually high doses were used and an attempt should be made to lower them. To estimate doses to children from CT Dose Index (CTDI) readings, the concept of the Size Specific Dose Estimate (SSDE) [21] helps to visualize the relation of pediatric doses to doses delivered to adults, since, due to their reduced body diameter, the same CTDI values result in considerably higher organ and tissue doses in children.

Last but not least, radiation staff education is one of the most efficient ways to enforce scientific “good practice” in radiological institutes and to reduce the radiation dose to patients [22–25].

The purpose of our study was to evaluate the efficacy of staff training and continuous education on the radiation doses applied in pediatric CT scans.

2. Materials and methods

Pediatric CT scans were performed in the emergency department (Somatom Sensation Cardiac 64; Siemens Medical, Erlangen, Germany), the divisions of musculoskelatal radiology (Brilliance 64, Philips, The Netherlands), traumatology (Somatom Sensation Open, Siemens, Germany), neuroradiology (Somatom Sensation 4 until September 2010, Somatom Sensation 64 thereafter, both Siemens, Germany), and surgery (Somatom Definition Flash, Siemens, Germany). On the emergency and the surgical scanner, pediatric radiologists were in charge, and radiologists with pediatric radiology experience were present in the divisions of musculoskeletal radiology, neuroradiology, and traumatology. The technical staff consisted of licensed radiographers. On all scanners, quality control programs were performed, including regular (semi-annual) CTDI calibration and monthly image quality tests.

To avoid image quality loss below diagnostic requirements, the radiologists were made aware of the on-going optimisation and asked to report image quality issues immediately.

3. Results

Examination numbers were 1799 in 2010, 1582 in 2011, and 1525 in 2012. Dose reports were not included into the PACS for 79 series in 2010, 53 in 2011, and 31 in 2012. Another 477 series were excluded in 2010 because of a non-standard scan range; 407 in 2011; and 449 in 2012. Thus, in total, the rejection rate was 27.3% (1496 scan series). Table 1 shows how cases were distributed between the departments’ scanners.

No image quality issues were reported by staff (radiographers and radiologists) during optimisation.

DLP values exhibited the best correlation with age using an exponential model (R² from 0.52 for CCT, 0.61 and 0.63 for abdomen–pelvis and thoracic scans, respectively, and 0.71 for trunk scans) and the least correlation when a linear model was applied (R² from 0.34 to 4.49), with a quadratic model only fitting minusculely better than the linear model. Therefore, a linear correlation of log(DLP) with age was utilized to determine whether statistically significant dose reductions were achieved.

Table 3 summarizes the ANCOVA results that assessed whether the effects of dose reduction were statistically significant. Patient doses were significantly reduced for CCT both from 2010 to 2011, and 2012 with respect to 2011. The same was true for thoracic scans. For abdomen–pelvis scans, a reduction in the DLPs was significant for 2011 with respect to 2010. Average DLP values were also reduced slightly in the following year, but significance could not be demonstrated. For trunk scans, a slight reduction in average DLPs occurred; however, it was not significant (p > 0.05). This can also be seen in Fig. 1a–d, showing the box plots for the data. Case numbers (see also Table 2) for abdomen–pelvis, and, especially trunk scans, were quite low. Fig. 1a–d also clearly demonstrates a reduction in the range of variation within the age bands, seen from the widths of the boxes (1st to 3rd quartile) and the whiskers (10th to 90th percentile).

1. Download high-res image (816KB)
2. Download full-size image

Fig. 1. (a–d): Box plots of dose length products (DLPs) distributions in age bands compared for 2010, 2011, and 2012.

Table 2. Case numbers per department and year.

Department	Cranial CT			Thorax CT			Abdomen–Pelvis CT			Trunk CT
	2010	2011	2012	2010	2011	2012	2010	2011	2012	2010	2011	2012
Neuroradiology (old scanner)	589
Neuroradiology (new scanner)	202	684	707	–	1	–	–	2	2	–	2	–
Traumatology	119	109	63	2	4	2	8	15	5	40	42	15
Surgery	4	5	1	115	155	159	34	43	30	10	22	18
Musculoskeletal radiology	2	1	1	15	17	23	3	16	6	7	–	–
Emergency	174	132	125	54	39	30	38	32	33	10	20	5

Table 3. ANCOVA results for logarithms of dose reference level (DLP) values.

	Average log (DLP)			P
	2010	2011	2012
CCT	6.43	6.29*	6.23*	<0.01
Thorax CT	4.45	4.12*	3.97*	<0.01
Abdomen–Pelvis CT	5.14	4.87*	4.76	<0.01
Trunk CT	5.15	5.10	4.67	>0.05

*

Indicates statistically significant reduction compared to preceding year.

4. Discussion

The application of Dose Reference Levels, in combination with staff education and awareness, has been demonstrated to be effective in dose optimization in pediatric computed tomography. The presentation of NDRL for pediatric cranial and body computed tomography to our already well-trained radiology technologists and radiologists enabled a radiation dose reduction for most pediatric CT examinations. This effect was also shown to be sustainable. This corresponds nicely to a rather recent publication indicating that one reason for unusually high doses in pediatric radiology was that radiation dose levels were never determined, and thus, not known [17].

Paolicchi et al. demonstrated the role of radiological staff training for adult chest, abdomen–pelvis, and whole body CT scans [25], indicating that training of radiologists and technologists is a key issue in optimizing CT protocols, and thus, can significantly reduce radiation dose [25]. They published also a radiation dose reduction while preserving diagnostic image quality in pediatric head CT examinations, after a radiologic staff training [27]. Schindera et al. reported the same effect for adult paranasal sinuses, brain, chest, pulmonary arteries, and abdomen CTs [24]. The authors reported that the largest decrease in radiation dose was reached for paranasal sinus and pulmonary arteries scans. No significant dose reduction for abdominal CTs was found; suboptimal image quality for low-contrast lesions in abdominal CTs was most likely identified as the reason. Education and training programs for radiological institutes was concluded to be effective in achieving substantial reduction in CT dose [24]. Sheyn et al. reported that staff radiation safety education can improve radiation safety practices in pediatric interventional radiology, and thus, decrease exposure to radiation for both staff and patients [22]. Georges et al. demonstrated in 2009 that training in radiation protection for interventional cardiologists, and the use of simple and cost-free dose-reduction techniques, were associated with a 50% reduction in radiation exposure to patients undergoing invasive cardiac procedures, without any loss of diagnostic information [23].

This work extends the scope of these studies, demonstrating the effects of continuing education and training on pediatric CT, and endorses the efficacy of the concept of DRLs. These two concepts complemented each other very well. We observed a significant decrease in DLP for all of our pediatric CT scans, including cranial, thoracic, abdomen–pelvis, and trunk CTs, during the observation interval from 2011 and 2012, compared to 2010, with the only exception being trauma CT scans. However, the DRLs are not directly applicable to trauma CT scans because of their specific requirements. This is most obvious in head scans, where the DRLs correspond to brain scans rather than the bony skull. Nevertheless, future work should be invested in imaging criteria including appropriate dose levels for trauma scans with special emphasis on pediatric examinations.

The relative number of scans performed at a dose exceeding the appropriate reference level was assessed. However, it must be noted that DRLs can only be applied on average doses and cannot be used on an individual basis. Therefore, it is inappropriate and misleading to argue that individual doses (DLPs in this case) that exceeded the DRLs indicated that too high a dose was used. If the values constantly exceeded that for standard patients and examinations, the dose level applied would be graded as “unusually high”, triggering the need to research the reasons for this higher level. Therefore, this type of assessment does not comply with the basic idea of reference Levels. However, if an unusual or rather high number of scans exceed the DRLs, and the rate decreases during optimization without any image quality issues arising, it seems reasonable to declare the education effort a success. Normally, doses quite below the appropriate DRLs can be applied since the actual values are not optimal values [28,29], but merely represent values below which 75% of institutions operate. However, care should be taken to ensure that patients and diagnostic requirements are comparable. In this work, DRLs from other countries were used, since a European consensus on values has not yet been reached, and Austrian levels do not cover all the indications we assessed in this work.

5. Conclusion

Modern CT scanners provide numerous possibilities for optimizing radiation dose and applying it more efficiently, like automatic tube current modulation, individual kVp optimization, and iterative reconstruction, to name a few [30]. However, continuous education of personnel to ensure that they are familiar with these possibilities, in order to utilize the potential for dose reduction, must be emphasized.