Sage Journals: Discover world-class research

Abstract

Introduction:

The use of computed tomography (CT) imaging as a diagnostic modality is increasing rapidly and CT is the dominant contributor to diagnostic medical radiation exposure. The aim of this project was to reduce the effective radiation dose to patients undergoing cranial CT examination, while maintaining diagnostic image quality.

Methods:

Data from a total of 1003, 132 and 27 patients were examined for three protocols: CT head, CT angiography (CTA), and CT perfusion (CTP), respectively. Following installation of adaptive iterative dose reduction (AIDR) 3D software, tube current was lowered in consecutive cycles, in a stepwise manner and effective radiation doses measured at each step.

Results:

Baseline effective radiation doses for CT head, CTA and CTP were 1.80, 3.60 and 3.96 mSv, at currents of 300, 280 and 130–150 mA, respectively. Using AIDR 3D and final reduced currents of 160, 190 and 70–100 mA for CT head, CTA and CTP gave effective doses of 1.29, 3.18 and 2.76 mSv, respectively.

Conclusion:

We demonstrated that satisfactory reductions in the effective radiation dose for CT head (28.3%), CTA (11.6%) and CTP (30.1%) can be achieved without sacrificing diagnostic image quality. We have also shown that iterative reconstruction techniques such as AIDR 3D can be effectively used to help reduce effective radiation dose. The dose reductions were performed within a short period and can be easily achievable, even in busy departments.

Keywords

Computed Tomography head CT radiation dosage iterative reconstruction quality improvement

Introduction

While the application of computed tomography (CT) imaging as a diagnostic modality is increasing rapidly, CT is also the dominant contributor to diagnostic medical radiation exposure.¹ In some departments, CT procedures makes up about 15% of the total number of examinations but accounts for some 70% of the exposure.¹ Thus, there is great interest in reducing unnecessary radiation dose from CT examinations.

Although skin burns² and hair loss³ have been reported with CT perfusion (CTP), most diagnostic radiology procedures normally produce relatively low radiation doses; the dose from a typical CT exam (1–14 mSv) is comparable to the annual dose received from natural sources (1–10 mSv).⁴ Singapore advises an annual dose limit for the general public of 1 mSv from medical sources.⁸ While the potential harm from ionising radiation has to be weighed against the useful diagnostic information that radiological procedures provides, this damage can be mitigated by reducing the exposure to ionising radiation to as low as reasonably achievable (the ‘ALARA’ principle), while maintaining sufficient diagnostic imaging quality of the results.

The aim of this study was to reduce effective radiation dose to patients undergoing CT scans of the head, while maintaining diagnostic image quality. This was achieved in a two-pronged approach: (1) by lowering the tube current for three types of head scans: CT head, CT angiography (CTA), and CTP, in consecutive cycles; and (2) by optimising image quality using adaptive iterative dose reduction (AIDR), software provided by the scanner manufacturer Toshiba.

Methods

Investigations were carried out on an Aquilion ONE™, 320-detector row system scanner v4.74 ER004 (Toshiba, Japan) with the X-ray tube voltage set at 120 kV (manufacturer’s recommended parameters). Rotation times varied with current, as needed. The scan data, which included unenhanced CT head scans, CTA and CTP, was collected over a period of 9 months from December 2011 to September 2012, from patients who had to undergo CT head scans at our institution. Exclusion criteria include non-CT head cases, CTA of the neck, and restless or uncooperative patients. Including baseline determination, a total of 1003 CT head, 132 CTA and 27 CTP patients’ data were examined. This project was approved by our institution’s Centralised Institutional Review Board.

It was inevitable that image quality degrades with lowered tube current, with resultant increased graininess of images, but our threshold standard was that diagnostic image quality was maintained throughout. Diagnostic image quality was assessed by a panel of six consultant neuroradiologists every time current was lowered. Preliminary CT scans performed at the new lowered tube current were compared with a recent CT scan performed at the higher tube current, for diagnostic acceptability by assessing lesion conspicuity and grey white matter differentiation.

Quantification of radiation dose

We chose the effective radiation dose as the descriptor of radiation dose as it is commonly accepted, and it takes into account the dose and relative radiosensitivity of all irradiated organs.^6,7 Measured in millisieverts (mSv), the effective radiation dose is not actually a metric of dose but rather a concept that reflects and best quantifies the stochastic risk from an exposure to ionising radiation.^6,8

The radiation dose measurements available on the CT console are volume CT dose index (CTDI_vol) and dose length product (DLP). CTDI_vol is calculated with a pitch value where pitch is the table distance travelled in one 3600 rotation; and DLP is CTDI_vol multiplied by scan length (slice thickness and number of slices). In this study, each sample was calculated from the DLP taken from the scanner after each scanand multiplied by the conversion factor for head (0.0021) to obtain the effective radiation dose in mSV. The effective radiation dose was calculated as the mean of all samples, where samples referred to the number of scans and not patients (e.g. some patients may have one, two, or more scans done).

Image optimisation using AIDR

Adaptive Iterative Dose Reduction 3D, or AIDR 3D, is software provided by the scanner manufacturer Toshiba.⁹ It is an advanced iterative reconstructive algorithm that reduces noise both in the raw data domain and in the three-dimensional reconstruction process, and improves spatial resolution. As the reconstructed image is optimised, scanning can be then performed at lower current settings without losing diagnostic information. Three settings, differing in the amount of noise removal, are available: mild, standard and strong. The standard setting was used throughout, except at PDSA Cycle 2a (elaborated below).

Human-dependent variables

We took steps to ensure that human-dependent variables such as scan range and repeat scans were controlled for in this study. As scan range is directly related to the effective radiation dose, it is important to keep it only as large as needed; this is the simplest way to control dose. A scout view was selected from the CT scanner, and keeping the same scout for each radiographer, the scan range for each radiographer was collected. As the scan range for CTP was set at 16 cm, this measurement step was not carried out for CTP scans.

Another factor that could possibly result in unnecessarily high dose is repeat scans. The number of repeat scans due to patient movement was counted over a period of 7 weeks. There were only 34 repeat scans out of 2582 scans (1%), and we found no factors such as patient age, medication, medical condition or time of day that was associated with the repeat scans. Based on this baseline count, we assumed that repeat scans should not constitute a substantial portion in the effective radiation dose given to patients and that it would be impractical to try to reduce the number of repeat scans further. Repeat scans or data from restless patients were discarded in this study.

Quality improvement methodology

This project was undertaken as an organisational quality improvement initiative within the institution under the Enhancing Performance Improving Care healthcare quality improvement programme. Using the Accelerated Model for Improvement (Ami^TM) framework developed by C. Jane Norman and based on The Model for Improvement by Langley et al.,¹⁰ an AMI charter was drawn up to answer three fundamental questions.

What are we trying to accomplish?

How will we know that a change is an improvement?

What changes can we make that will result in improvement?

These questions inform the Plan-Do-Study-Act (PDSA) iterative cycles that forms an efficient trial-and-learning methodology for stepwise improvement. Built into the PDSA cycle are the ideas of deductive and inductive learning.

The current was lowered according to the aims of the PDSA cycles:

PDSA Cycle 1 aims to establish baselines levels of effective radiation dose;

PDSA Cycle 2 installs AIDR and aims to lower the tube current until the image quality was comparable to pre-AIDR installation; and

PDSA Cycle 3, with AIDR still in place, aims to lower the current further.

We wanted to test (1) the effect of AIDR installation, and (2) the effect of reduced current, on effective radiation dose. Changes are measured by (1) reduction in radiation dose, and (2) radiologist satisfaction.

Statistical Analysis

Statistical analysis was conducted in MiniTab (Version 16) software. Data collected from the PDSA cycles were entered into MiniTab and control and run charts were created. Using standard quality improvement analyses rules used by the Accelerated Model for Improvement (Ami^TM) framework and MiniTab software, different control charts were created depending on the sampling size. “Experiments” with large samples (e.g. CT head) had calculated mean and standard deviation values, and small “experiments” (e.g. CT angiograph and CT perfusion) had calculated mean and range or moving range values. As with any control chart analysis, the key parameter other than the mean (or more accurately, the “Central Line”) is the Upper Control Limit (UCL) and the Lower Control Limit (LCL). We have provided the mean/”Central Line” and UCL and LCL values in this paper.

Results

Human-dependent variables

Scan ranges for CT head for the radiographers involved in the study varied from 125 to 150 mm, with a mean of 135 mm, resulting in a variability range of 25 mm. For CTA, the scan ranges varied from 135 to 150 mm, with a mean of 142 mm leading to a range of only 15 mm. In both cases, we decided that the variability ranges for and between each radiographer were acceptable on a practical process level.

PDSA cycles

At the start of the study, the tube current was 300 mA for CT head, 280 mA for CTA, and 130–150 mA for CTP. These values were taken from the scanner manufacturer’s guidelines.

PDSA Cycle 1 established the baseline values. Calculated from scan data in the scanner database, the mean doses (UCL, LCL) were found to be 1.80 (1.88, 1.73), 3.60 (4.00, 3.19) and 3.95 (4.02, 3.89) mSv for CT head, CTA and CTP, respectively.

In PDSA Cycle 2, current was lowered from 300 to 240 mA for CT head; current was lowered from 280 to 270 mA for CTA; and current was lowered from 130–150 to 130–140 mA for CTP. In the cases of CT head and CTA, rotation time was increased in order to compensate for increased image noise with reduced tube current. Effective radiation dose fell from 1.80 to 1.59 (1.67, 1.50) mSv for CT head, 3.60 to 3.20 (3.56, 2.57) mSv for CTA, and 3.95 to 3.76 (no variation) mSv for CTP, resulting in 11.7%, 11.1% and 4.8% reductions, respectively.

In PDSA Cycle 3, current was further reduced from 240 to 160 mA for CT head; from 270 to 190 mA for CTA and from 130–140 to 70–100 mA for CTP. For CT head, effective radiation dose following current reduction fell from 1.59 to 1.29 (1.32, 1.27) mSv (18.9% reduction), and the cumulative reduction from baseline was 28.3%. For CTA, the final effective radiation dose only fell from 3.2 to 3.18 (3.57, 2.74) mSv (0.6% reduction), with a cumulative reduction of 11.6% from baseline. For CTP, effective radiation dose was reduced from 3.76 to 2.76 (3.03, 2.44) mSv (26.6% reduction), with a cumulative reduction of 30% from baseline.

The results of the PDSA cycles are summarised in Table 1, and a representative control chart showing the effective radiation doses pre- and post-AIDR installation for CT head is shown in Figure 1.

Table 1.

Actions and results of PDSA Cycles 1–3.

	CT head	CTA	CTP
PDSA 1: Baseline establishment
Baseline current (mA)	300	280	130–150
Mean baseline dose (mSv) (UCL, LCL)	1.80 (1.88, 1.73)	3.60 (4.00, 3.19)	3.95 (4.02, 3.89)
Rotation time (s)	0.50	0.50	0.75
No. of scans	314	48	15
PDSA 2: Effect of AIDR and current reduction
Reduction in current (mA)	300 → 240	280 → 270	130–150 → 130–140
Reduction in mean dose (mSv) (UCL, LCL)	1.80 → 1.59 (1.67, 1.50)	3.60 → 3.20 (3.56, 2.87)	3.95 → 3.76 (No variation)
Amount of reduction (mSv)	0.21 (11.7%)	0.40 (11.1%)	0.19 (4.8%)
Rotation time (s)	0.60	0.50	0.75
No. of scans	331	49	7
PDSA 3: Effect of AIDR and further current reduction
Reduction in current (mA)	240 → 160	270 → 190	130–140 → 70–100
Reduction in mean dose (mSv) (UCL, LCL)	1.59 → 1.29 (1.32, 1.27)	3.20 → 3.18 (3.57, 2.74)	3.76 → 2.76 (3.03, 2.44)
Amount of reduction (mSv)	0.30 (18.9%)	0.02 (0.6%)	1 (26.6%)
Rotation time (s)	0.75	0.60	0.75
No. of scans	358	35	5
Cumulative drop in dose (mSv)	0.51 (28.3%)	0.42 (11.6%)	1.19 (30.1%)

AIDR: adaptive iterative dose reduction; CT: computed tomography; CTA: computed tomography angiography; CTP: computed tomography perfusion; PDSA: Plan-Do-Study-Act.

Figure 1.

A representative control chart showing the effective radiation doses pre- and post-AIDR (adaptive iterative dose reduction) installation for CT head. PDSA (Plan-Do-Study-Act) Cycle 3 was done in a two-step manner (3a, from 240 to 180 mA; and 3b, from 180 to 160 mA).

From Table 1, we can see that a 46.7% drop in current (300 to 160 mA) for CT head resulted in a 28.3% (1.80 (1.88, 1.73) to 1.29 (1.32, 1.27) mSv) reduction in effective radiation dose, while a 32.1% drop in current (280 to 190 mA) for CTA resulted in a 11.6% (3.60 (4.00, 3.19) to 3.18 (3.57, 2.74) mSv) reduction in effective radiation dose. The current varies over a range for the CTP procedure, but an approximate reduction by a third of the current resulted in a 30.1% reduction (3.95 (4.02, 3.89) to 2.76 (3.03, 2.44) mSv) in effective radiation dose.

There were situations where scanning time had to be increased to offset for the lowered current; in such cases, the eventual reduction in effective radiation dose was less than initially expected. In particular, although the current for CTA in PDSA cycle 3 was reduced by 29.6% (from 270 to 190 mA), the drop in effective radiation dose was only 0.6% because of the increase in rotation time, from 0.50 to 0.60 s. In contrast, while the current for CTA in cycle 2 was only reduced by 3% (from 280 to 270 mA), the drop in effective radiation dose was 11%; there was no change in rotation time from baseline parameters.

Discussion

We have satisfactorily reduced effective radiation dose to patients undergoing CT scans of the head, while maintaining diagnostic image quality. The reductions were 28.3% for CT head, 11.6% for CTA, and 30.1% for CTP, arriving at final effective radiation doses of 1.29 mSv, 3.18 mSv, and 2.76 mSV, respectively.

Radiation dose is largely determined by the tube current and X-ray voltage, and is a main determinant of CT image quality and therefore diagnostic accuracy.^1,11 In addition to current and voltage, other machine specifications that affect radiation dose include scanning rotation time, slice thickness, scan length and the type of scanner.^11,12 While only current was varied in this study, other studies have shown that dose reduction can be satisfactorily achieved by lowering tube potentials, especially for CT examinations using iodinated contrast media.^13–15

One of the simplest ways to reduce radiation dose is to match the diagnostic aim and not requiring lower noise or higher spatial resolution than necessary; the type of diagnostic examination sets the image quality required. The challenge with head CT is that the tissues scanned are relatively low contrast, compared with high-contrast areas such as the lung or bones, where a dose reduction of more than 50% can be tolerated.^16,17

Another way to reduce radiation dose is to improve radiographers’ skill and experience. Training for new staff as well as refresher courses for existing staff is important, especially whenever a new machine, software or update is installed. Other strategies on the operator’s side include regular calibration of CT scanners, and keeping abreast and installing manufacturers’ updates.

Since the extent to which current can be lowered is limited by diagnostic image quality, optimisation of image quality is an important problem. Image quality can be optimised through hardware and software means. When image quality is improved, the current can be further lowered before diagnostic quality is lost. One of the main challenges to dose reduction is image noise, which results from two main sources: photon noise and electronic noise. In a low-dose scan, when the number of X-ray photons drops to the extent that the electronic noise in the data acquisition system becomes dominant, image quality degrades. The AIDR software used in this study addresses this problem, by reducing both photon and electronic noise, and therefore increases the signal-to-noise ratio, improves spatial resolution and produces a more natural-looking image.⁹ It has a smoothing effect on the images, although higher strength can lead to over smoothing or waxiness of images.¹⁸

Iterative reconstruction, also known as adaptive statistical iterative reconstruction (ASIR), algorithms have been available for some years, and have been shown to significantly improve image quality.^18,19 Though computationally more intensive than the alternative filtered back projection (FBP) algorithms, they are more accurate and perform better, especially when dealing with noisy data.⁵ Vendors will often have their proprietary iterative methods to optimise image quality: while Toshiba has AIDR, Siemens offers Iterative Reconstruction in Image Space (IRIS), and Phillips, iDose.

Our final effective dose of 1.29 mSv for CT head compares well with other image optimisation work on head CT. Rapalino et al.¹⁸ used a blended ASIR and FBP algorithm on cranial CT, and reduced the routine effective dose of 2.66 mSv (at 175 mAs) to 1.95 mSv (at 140 mAs), a reduction of 26.2%. Kilic et al.¹⁹ used ASIR to lower the effective dose from 2.3 to 1.6 mSv, a reduction of 31%. As mentioned earlier, the low-contrast tissue in the head presents a challenge; using ASIR, Flicek et al.²⁰ studied were able to achieve a dose reduction of 50% for CT colonoscopy examination.

Conclusions

Satisfactory reductions in effective radiation dose for CT head (28.3%), CTA (11.6%) and CTP (30.1%) can be obtained without sacrificing diagnostic image quality. The final effective radiation doses were 1.29 mSv for CT head, 3.18 mSv for CTA and 2.76 mSV for CTP. Moreover, image optimisation protocols such as AIDR 3D are an effective aid to current reduction in reducing the effective radiation dose. Finally, using a quality improvement methodology, we demonstrated that dose reductions can be tested in a stepwise manner, measured through assessment of image quality, and implemented within a short timeframe in a busy department.

Footnotes

Declaration of Conflicting Interest

None declared.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

References

Linton

Mettler

Jr . National conference on dose reduction in CT, with an emphasis on pediatric patients. Am J Roentgenol 2003; 181: 321–329.

Baughs

Vrieze

Primak

. CT dosimetry: Comparison of measurement techniques and devices. Radiographics 2008; 28: 235–253.

Bogdanich

. After stroke scans, patients face serious health risks. New York Times, 1 August 2010, p. A1.

McCollough

Primak

Braun

. Strategies for reducing radiation dose in CT. Radiol Clin North Am 2009; 47: 27–40.

Radiation Protection Act, c. 262, s. 28, Second Schedule, http://statutes.agc.gov.sg/aol/download/0/0/pdf/binaryFile/pdfFile.pdf?CompId:ba68c918–0778–4c4d-b180–768ac7abf3a8 (accessed 15 September 2001).

McCollough

Schueler

. Calculation of effective dose. Med Phys 2000; 27: 828–837.

Lockwood

Einstein

Davros

. Diagnostic imaging: Radiation dose and patients’ concerns. Cleveland Clin J Med 2006; 73: 583–586.

International Commission on Radiological Protection (ICRP). Recommendations of the International Commission on Radiological Protection (ICRP #26). Oxford: The International Commission on Radiological Protection, 1977.

Irwan

Nakanishi

Blum

. AIDR 3D reduces dose and simultaneously improves image quality. Visions 2012; 19: 34–39.

10.

Langley

Moen

Nolan

. The improvement guide: A practical approach to enhancing organization performance. 2nd ed. San Francisco: John Wiley & Sons, 2009, pp.23–25.

11.

Huda

Vance

. Patient radiation doses from adult and pediatric CT. Am J Roentgenol 2007; 188: 540–546.

12.

Liu

Leng

. Radiation dose reduction in computed tomography: Techniques and future perspective. Imaging Med 2009; 1: 65–84.

13.

Siegel

Schmidt

Bradley

. Radiation dose and image quality in pediatric CT: Effect of technical factors and phantom size and shape. Radiology 2004; 233: 515–522.

14.

Funama

Awai

Nakayama

. Radiation dose reduction without degradation of low contrast detectability at abdominal multisection CT with a low-tube voltage technique: Phantom study. Radiology 2005; 237: 905–910.

15.

Nakayama

Awai

Funama

. Abdominal CT with low tube voltage: Preliminary observations about radiation dose, contrast enhancement, image quality, and noise. Radiology 2005; 237: 945–951.

16.

Naidich

Marshall

Gribbin

. Low-dose CT of the lungs: Preliminary observations. Radiology 1990; 175: 729–731.

17.

Zwirewich

Mayo

Müller

. Low-dose high-resolution CT of lung parenchyma. Radiology 1991; 180: 413–417.

18.

Rapalino

Kamalian

. Cranial CT with adaptive statistical iterative reconstruction: Improved image quality with concomitant radiation dose reduction, Am J Neuradiol 2012, 33: 609–615.

19.

Kilic

Erbas

Guryildirim

. Lowering the dose in head CT using adaptive statistical iterative reconstruction. Am J Neuroradiol 2011; 32: 1578–1582.

20.

Flicek

Hara

Silva

. Reducing the radiation dose for CT colonography using adaptive statistical iterative reconstruction: A pilot study. Am J Roentgenol 195: 126–131.

Reducing cranial computed tomography effective radiation dose by 30% using adaptive iterative dose reduction

Abstract

Introduction:

Methods:

Results:

Conclusion:

Keywords

Introduction

Methods

Quantification of radiation dose

Image optimisation using AIDR

Human-dependent variables

Quality improvement methodology

Statistical Analysis

Results

Human-dependent variables

PDSA cycles

Discussion

Conclusions

Footnotes

Declaration of Conflicting Interest

Funding

References