Tailored CT angiography in follow-up after endovascular aneurysm repair (EVAR): combined dose reduction techniques

Abstract

Background

Endovascular aneurysm repair (EVAR) requires lifelong surveillance by computed tomography angiography (CTA). This is attended by a substantial accumulation of radiation exposure. Iterative reconstruction (IR) has been introduced to approach dose reduction.

Purpose

To evaluate adaptive statistical iterative reconstruction (ASIR) at different levels of tube voltage concerning image quality and dose reduction potential in follow-up post EVAR.

Material and Methods

One hundred CTAs in 67 patients with EVAR were examined using five protocols: protocol A (n = 40) as biphasic standard using filtered back projection (FBP) at 120 kV; protocols B (n = 40), C (n = 10), and D1 (n = 5) biphasic using ASIR at 120, 100, and 80 kV, respectively; and protocol D2 (n = 5) with a monophasic splitbolus ASIR protocol at 80 kV. Image quality was assessed quantitatively and qualitatively. Applied doses were determined.

Results

Applied doses in ASIR protocols were significantly lower than FBP standard (up to 75%). Compared to protocol A, signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) (e.g. arterial CNR intra-/extra-stent lumen: A = 35.4 ± 13.5, B = 34.2 ± 10.0, C = 29.6 ± 6.8, D1 = 32.1 ± 6.3, D2 = 40.8 ± 23.1) in protocol B were equal and in protocols C and D equal to partially inferior, however not decisive for diagnostic quality. Subjective image quality ratings in all protocols were good to excellent without impairments of diagnostic confidence (A–D2: 5), with high inter-rater agreement (60–100%).

Conclusion

ASIR contributes to significant dose reduction without decisive impairments of image quality and diagnostic confidence. We recommend an adapted follow-up introducing ASIR and combined low-kV in the long-term surveillance after EVAR.

Keywords

Computed tomography angiography (CTA)stents interventional aorta aneurysms

Introduction

Abdominal aortic aneurysm (AAA) is a common aortic condition which can cause fatal complications such as acute aortic occlusion or aortic aneurysm rupture (1,2). Endovascular aneurysm repair (EVAR) has become the preferred method of treatment for AAA (3).

Despite the minimally invasive nature of EVAR, various complications have been reported (1,4,5). Therefore, patients who underwent EVAR require lifelong imaging surveillance (1). According to guidelines, patients are often followed up by computed tomography angiography (CTA) (6,7). Long-term follow-up after EVAR is thus attended by a substantial accumulation of radiation exposure and hence a worthwhile aim of dose reduction efforts in clinical routine (3,8,9).

Major approaches to minimize radiation exposure are to lower the tube current and tube voltage. Unfortunately, this is inversely linked to a higher level of image noise, which in turn degrades diagnostic image quality (8,10,11).

To solve this dilemma, iterative reconstruction (IR) techniques have been introduced. These techniques are aimed at compensating for higher levels of image noise by recovering adequate image quality through post-acquisition data processing. However, IR techniques may lead to blurring and loss of critical details (such as stent-graft wires or endoleaks) due to a shift in the noise power spectrum (8,12 –15). Ultimately, the best achievable image quality depends on finding the right balance between image noise and the level of blurring (15).

IR techniques based on statistical algorithms such as adaptive statistical iterative reconstruction (ASIR; GE Healthcare, Milwaukee, WI, USA) are attractive candidates for use in clinical routine. Compared to fully model-based IR techniques, they enable image reconstruction within an acceptable time frame at today's processor speeds (12,13,16). Although a few studies have investigated statistical IR techniques in patients after EVAR, to this point there is no consensus for how to use IR in clinical routine (17). Moreover, to our knowledge, there is no study dealing with the combination of low-kV and ASIR in EVAR follow-up.

The primary purpose of this study is to assess the dose reduction potential of the ASIR technique for post-processing of CTAs performed in the follow-up of patients after EVAR. The secondary purpose is the evaluation of the ASIR technique in combination with kV reduction. Additionally, a recommendation of a tailored long-term follow-up CT scheme for EVAR patients shall be derived.

Materials and Methods

Power analysis and sample size estimation

Before our study, we carried out a power analysis based on empirical values retrospectively collected by CTAs in the clinical routine. We looked at contrast-to-noise ratio (CNR) of intra-stent vs. extra-stent lumina, CNR of intra-stent lumen vs. periaortic adipose tissue, and the signal-to-noise ratio (SNR) of intra-stent lumen and extra-stent lumen, respectively. We considered these four parameters to be the most conclusive for our study purpose. For CNR intra-stent vs. extra-stent lumina and SNR intra-stent lumen, we calculated a sample size of about 35 (anticipated means protocol A of 40 ± 15 and protocol B of 30). For the SNR extra-stent lumen, we calculated a sample size of about 25 (anticipated means protocol A of 5.5 ± 2.5 and protocol B of 3.5). For the CNR intra-stent lumen vs. periaortic adipose tissue, we calculated a sample size of about 35 (anticipated means protocol A of 55 ± 15 and protocol B of 45). Each calculation assumes α of 0.05 and β of 0.2 (meaning a power of about 80%). From this, we assumed a statisitcally realiable sample size of about 40 CTAs per study protocol.

Patients and study design

The study was approved by the local ethics committee. Over a 20-month period from November 2013 to July 2015, the study was conducted prospectively at our university hospital. During this period, a total of 100 CTAs were acquired in the follow-up of 67 patients treated with EVAR. The consecutive series were subsequently allocated to one of five study protocols: the first 40 scans (allocated to protocol A) served as the control protocol. CTA in this protocol was performed using a standard filtered back projection (FBP) protocol at 120-kV tube voltage. The second 40 scans (protocol B) were examined with an ASIR protocol at 120-kV tube voltage (40% use of ASIR to the raw data). Protocols C (n = 10) and D (n = 10) were reduced to smaller numbers in order not to exceed a tolerable inclusion period excessively. They were examined with an ASIR protocol (40% use of ASIR to the raw data) at decreasing levels of tube voltage (B = 120 kV, C = 100 kV, D = 80 kV). Protocol D was subdivided into two subprotocols to further optimize the ASIR protocol: D1 (n = 5) and D2 (n = 5). In protocol D1, CTAs were reconstructed with 40–100% use of reconstruction ASIR (in addition to 40% use of ASIR to the raw data) to evaluate how image noise can be reduced further after data acquisition. CTA in protocol D2 was performed using a monophasic split-bolus contrast-agent administration to further reduce radiation exposure compared with the standard biphasic contrast-agent administration used in protocols A, B, C, and D1.

Group matching

The five protocols were analyzed to identify any statistically significant differences regarding patient's body diameter, scan length, and administered volume of contrast agent (Table 1). Body diameters were used instead of widely spread body mass index (BMI), as the BMI does not necessarily reflect the patient's habitus at scan level and can be a misleading parameter where assessing image noise and radiation exposure. Body diameters were assessed in sagittal and frontal planes at two different levels. The first body diameter was measured at the AAA level. Additionally, the patient's maximum body diameter was obtained at individually different levels in order to account for different body shapes. Technical CT parameters, except kV and ASIR as mentioned above, did not differ between any of the protocols.

Table 1.

Overview of study scans and matching variables.

Protocol	A (FBP, 120 kV)	B (ASIR, 120 kV)	C (ASIR, 100 kV)	D1 (ASIR, 80 kV)	D2 (ASIR, 80 kV, Split)	P value
Scans (n)	40	40	10	5	5	–
Gender	37 M / 3 F	36 M / 4 F	10 M / 0 F	4 M / 1 F	3 M / 2 F
Mean age ± SD (years)	67.5 ± 8.0	71.5 ± 9.7	73.0 ± 8.7	75.0 ± 4.9	75.0 ± 12.5	0.445
Max. body diameter (frontal) (cm)	39.0 ± 3.5	37.9 ± 3.4	39.7 ± 3.3	42.4 ± 1.8	39.0 ± 3.9	0.052
Max. body diameter (sagittal) (cm)	30.7 ± 4.1	28.5 ± 4.2	30.2 ± 3.3	30.6 ± 1.7	28.8 ± 3.9	0.103
Frontal diameter (aneurysm level) (cm)	37.6 ± 4.2	36.0 ± 3.9	37.6 ± 3.3	37.0 ± 1.9	35.2 ± 3.9	0.209
Sagittal diameter (aneurysm level) (cm)	28.1 ± 4.3	25.9 ± 3.9	28.0 ± 3.9	26.6 ± 2.1	25.8 ± 3.9	0.083
Volume of contrast agent (mL)	110.8 ± 18.3	115.3 ± 11.1	118.0 ± 6.3	112.0 ± 11.0	116.0 ± 8.9	0.162
Total scan range (cm)	84.6 ± 30.7	85.6 ± 21.7	83.4 ± 19.4	90.4 ± 8.0	47.1 ± 9,6	0.417
Endoleaks (n)	Ia: 1, Ib: 1, IIa: 11, IIb: 4	Ia: 1 IIa: 8 V: 1	IIa: 3	IIa: 2	IIa: 3

There are no statistically significant differences in any obtained parameter.

Protocols A, B, C, D1 = biphasic CT examination; protocol D2 = monophasic CT examination with split-bolus contrast agent administration.

FBP, filtered back projection; ASIR, adaptive statistical iterative reconstruction; kV, kilovolt.

CT examination

All examinations were performed on a 64-slice multi-detector CT scanner (Lightspeed VCT, GE Healthcare, Milwaukee, WI, USA) using the following protocol: standard administration of 120 mL (the volume could vary up to 40 mL in obese or shortweighted patients) of contrast medium (Xenetix 350, Guerbet, Villepinte, France) at a flow rate of 3.5 mL/s and acquisition of either two spiral series of the entire abdomen from the base of lungs to below the symphysis during the arterial and the venous phase (protocols A, B, C, and D1) or one spiral series during a mixed arterial and venous phase after split-bolus administration (protocol D2). Arterial contrast phase was defined using automated scan-triggering software (SmartPrep, GE Healthcare, Milwaukee, WI, USA). Attenuation threshold in the infrarenal aorta was set to 150 Hounsfield units (HU). Delay times followed our institution's standards for split-bolus and multi-phase protocols. Image acquisition was performed with the tube voltage set according to each study protocol (A = 120 kV, B = 120 kV, C = 100 kV, D = 80 kV), collimation of 64 × 0.625 mm, and pitch of 1.375. Tube current was set to automatic modulation (range of 100–500 mAs) at a standard noise index of 15 (noise index was increased to 25 in protocols C and D to prevent disproportional mA upregulation in low-kV examinations). Raw data were reconstructed at a slice thickness of 0.625 mm and use of 40% reconstruction ASIR for post-processing.

Data analysis

Quantitative image analysis was performed by drawing regions of interest (ROI) in three defined organ regions: intra-stent lumen; extra-stent lumen; and periaortic fat. One ROI was placed in the display field of view (dFOV) outside the patient to measure background noise and its standard deviation (SD).

The attenuation values (in HU) measured in these ROIs were analyzed by calculating SNR and several CNRs. SNR was defined as specific organ attenuation divided by the standard deviation of noise (SNR = ROI/SD). Mean SNR was calculated as mean of all organ values divided by standard deviation of noise (SNR = Σ [(ROI1 + ROI2 + … ROI9)/n]/SD). CNR was defined as the difference between two specific organ attenuations divided by the standard deviation of noise (CNR = [ROI1–ROI2]/SD).

Qualitative image analysis was performed by two experienced radiologists blinded to CTA parameters who assessed image quality in six categories: noise; contrast; artifacts; detectability of small structures (mesh wires of EVAR stent); overall diagnostic quality; and EVAR-related diagnostic confidence. Each category was evaluated separately using a 5-point Likert scale (5 = excellent – 1 = poor). Technical information on the image was obscured to reduce expectation bias.

For radiation dose estimation, the volume CT dose index (CTDIvol) and dose-length product (DLP) values were recorded, and the effective dose was calculated by multiplying DLP values by an age-specific conversion factor for the analyzed body region deriving from the recommendations of the International Commission on Radiological Protection (ICRP), which was 0.0153 for adults at 120-kV tube voltage and 0.0151 at 100-kV and 80-kV tube voltage, respectively (18).

Statistical analysis

Statistical analysis was performed using SPSS Statistics version 22 (IBM, Armonk, NY, USA) and MedCalc Statistics version 15.8 (MedCalc Software bvba, Ostend, Belgium). Ordinal data were tested for significance using the Mann–Whitney U-test between two protocols and the Kruskal–Wallis test for more than two protocols of different sample sizes. Interval data were tested using an students t-test for independent samples with 95% confidence intervals. Descriptive data are given as mean values and standard deviations. Interrater agreement was evaluated using the nonlinear weighted Cohen's K-test. Statistical significance was assumed at a P value of < 0.05.

Results

Group matching

There were no statistically significant differences between protocols A, B, C, and D regarding patients' age, gender distribution, maximal frontal plane body diameter, maximal sagittal body diameter, frontal plane body diameter at aneurysm level, sagittal body diameter at aneurysm level, the administered volume of contrast agent, and total scan range (Table 1). Table 1 furthermore shows details about the prevalence of endoleaks we observed in the different protocols.

Quantitative image grading

The results of quantitative image grading are summarized in Table 2. SNRs and CNRs of ASIR protocol B were the same as in protocol A, which was examined with the standard FBP at tube voltage of 120 kV. In the other ASIR protocols, SNRs and CNRs decreased with tube voltage reduction (protocols C and D). However, standard protocol A did not show statistically significant differences compared to any of the ASIR protocols for CNRs of intra-stent vs. extra-stent lumen. CNRs of intra- and extra-stent lumen vs. periaortic fat in protocols C and D1 were significantly lower (but identical in protocol D2).

Table 2.

Results of quantitative image grading.

Protocol	A (FBP, 120 kV)	B (ASIR, 120 kV)		C (ASIR, 100 kV)		D1 (ASIR, 80 kV)		D2 (ASIR, 80 kV, Split)
Protocol	A (FBP, 120 kV)		P value		P value		P value		P value
SNR (arterial)
Intra-stent lumen	40.8 ± 14.9	39.5 ± 11.0	0.642	32.9 ± 7.3	0.111	35.6 ± 8.4	0.448	45.4 ± 23.1	0.546
Extra-stent lumen	5.4 ± 2.8	5.4 ± 2.1	0.945	3.3 ± 1.1	0.026	3.5 ± 2.3	0.150	4.6 ± 1.8	0.527
Periaortic fat	13.2 ± 4.7	12.1 ± 3.6	0.237	9.4 ± 2.6	0.018	6.8 ± 1.3	0.005	12.4 ± 3.1	0.723
Mean SNR	19.8 ± 6.9	19.1 ± 4.6	0.569	15.2 ± 3.2	0.045	15.3 ± 3.9	0.159	20.8 ± 7.6	0.765
SNR (venous)
Intra-stent lumen	17.9 ± 6.5	19.3 ± 7.4	0.346	15.1 ± 4.1	0.208	14.8 ± 3.3	0.302	–	–
Extra-stent lumen	5.4 ± 2.3	5.5 ± 2.1	0.904	3.3 ± 0.8	0.000	3.7 ± 2.3	0.118	–	–
Periaortic fat	12.4 ± 4.7	11.0 ± 4.6	0.195	9.1 ± 2.1	0.035	6.5 ± 1.5	0.008	–	–
Mean SNR	11.9 ± 4.0	12.0 ± 3.4	0.912	9.2 ± 2.2	0.041	8.3 ± 2.3	0.055	–	–
CNR (arterial)
Intra-stent lumen – extra-stent lumen	35.4 ± 13.5	34.2 ± 10.0	0.636	29.6 ± 6.8	0.194	32.1 ± 6.3	0.590	40.8 ± 23.1	0.456
Intra-stent lumen – periaortic adipose tissue	54.0 ± 18.9	51.2 ± 12.9	0.446	42.3 ± 9.0	0.066	42.4 ± 9.4	0.191	57.8 ± 22.5	0.686
Extra-stent lumen – periaortic adipose tissue	18.6 ± 6.5	17.0 ± 4.7	0.237	12.7 ± 3.0	0.009	10.3 ± 3.5	0.009	17.0 ± 4.7	0.615
CNR (venous)
Intra-stent lumen – extra-stent lumen	12.4 ± 5.3	14.1 ± 6.8	0.240	11.8 ± 3.6	0.322	11.1 ± 1.5	0.337	–	–
Intra-stent lumen – periaortic adipose tissue	30.2 ± 10.4	30.4 ± 9.5	0.959	24.2 ± 5.9	0.055	21.2 ± 4.6	0.041	–	–
Extra-stent lumen – periaortic adipose tissue	17.8 ± 6.1	16.6 ± 4.9	0.334	12.4 ± 2.8	0.012	10.2 ± 3.7	0.007	–	–

All P values are given in relation to FBP. p values <0.05 in italics.

SNR, signal-to-noise ratio; CNR, contrast-to-noise ratio.

Qualitative image grading

The results of qualitative image grading are summarized in Table 3. Means of both reviewers' quality ratings were calculated and showed high levels of overall interrater agreement (62–100%). Cohen's weighted K-test showed fair (0.21–0.40) to moderate (0.41–0.60) interrater agreement (19). Overall diagnostic quality was rated high to excellent (4.2–5.0) and diagnostic confidence was rated excellent (5) for all protocols. Both reviewers described a slightly smoother impression of ASIR images, which is caused by slight softening of sharp contrasts of adjacent tissues. Image noise increases slightly at lower tube voltages (Fig. 1). Critical image findings such as endoleaks were consistently and reliably identified using ASIR images at all levels of tube voltage investigated (Fig. 2). For protocol D1 (80 kV and ASIR), there was agreement between the two reviewers that the image impression was best at an increased level of ASIR reconstruction during postprocessing of approximately 60%, i.e. a 60%ASIR/40%FBP blending (Fig. 3).

Table 3.

Results of qualitative image grading.

Protocol	A (FBP, 120 kV)	B (ASIR, 120 kV)		C (ASIR, 100 kV)		D1 (ASIR, 80 kV)		D2 (ASIR, 80 kV, Split)		TIA (%)	K
Protocol	A (FBP, 120 kV)		P value		P value		P value		P value	TIA (%)	K
Noise	4.4 (3–5)	4.4 (4–5)	0.823	4.0 (3–5)	0.000	3.6 (3–4)	0.000	3.9 (3–4)	0.004	69	0.470
Contrast	4.7 (4–5)	4.8 (4–5)	0.032	4.6 (4–5)	0.602	4.0	0.000	3.8 (3–4)	0.000	62	0.289
Small structures	4.3 (3–5)	4.5 (4–5)	0.030	4.5 (4–5)	0.025	3.6 (3–5)	0.000	4.0 (3–5)	0.000	73	0.308
Artifacts	4.8 (4–5)	4.5 (4–5)	1.000	4.6 (3–5)	0.109	3.7 (3–5)	0.001	3.9 (3–5)	0.134	60	0.435
Overall diagnostic quality	5.0 (4–5)	5.0 (4–5)	0.174	4.9 (4–5)	0.402	4.2 (4–5)	0.000	4.6 (4–5)	0.000	89	0.362
Diagnostic confidence	5.0	5.0	1.000	5.0	1.000	5.0	1.000	5.0	1.000	100	–

All scores are given as mean of both reviewers' ratings. p values <0.05 in italics.

TIA, total interrater agreement; Κ, weighted Cohen's K-value.

Fig. 1.

Intraindividual comparison of FBP vs. ASIR at decreasing levels of tube voltage. Image quality of all scans was rated good to excellent. All structures could be evaluated clearly. Image noise increases slightly at lower levels of tube voltage. Note increasing intra-stent attenuation and higher intra- to extra-stent contrast at decreasing levels, which is attributable to known effects of the iodine K-edge (24).

Fig. 2.

Endoleaks in different patients detected by FBP and ASIR protocols at decreasing levels of tube voltage. All endoleaks could be evaluated reliably (white arrowheads). There are no significant differences in diagnostic quality between arterial and venous phase series. Artifacts increased slightly towards lower levels of tube voltage.

Fig. 3.

Intraindividual comparison of increasing levels of reconstruction ASIR (rASIR) at 80-kV tube voltage (protocol D1). All critical structures could be evaluated (e.g. note the mesh wire and the small aortic endoleak on the right). Increasing levels of reconstruction ASIR seem to lower image noise but images appear increasingly blurred. All series were computed with 40% use of ASIR to the raw data. The use of increasing levels of reconstruction ASIR has no effect on radiation dose (14).

Dose reduction

The results for dose reduction are compiled in Table 4. Compared to the standard FBP protocol (protocol A), the estimated effective dose was 37.4% lower for the protocol with 40% ASIR of raw data at the same tube voltage (protocol B). Dose reduction of 61.4% was achieved at a tube voltage of 80 kV (protocol D1). The highest dose reduction of 75.0% was achieved with the monophasic split-bolus protocol (protocol D2) (Fig. 4)

Table 4.

Radiation exposure expressed as volume CT dose index (CTDIvol), total dose length product (total DLP), and estimated effective dose.

Protocol	A (FBP, 120 kV)	B (ASIR, 120 kV)	C (ASIR, 100 kV)	D1 (ASIR, 80 kV)	D2 (ASIR, 80 kV, Split)
CDTIvol (mGy)	9.9 ± 3.3	5.8 ± 2.1 (41.4%)	5.0 ± 1.0 (49.5%)	3.7 ± 0.9 (62.6%)	4.2 ± 0.7 (57.6%)
Total DLP (mGy x cm)	914.3 ± 256.0	572.3 ± 249.9 (37.4%)	476.4 ± 94.4 (47.9%)	355.9 ± 41.1 (61.1%)	232.0 ± 58.9 (74.6%)
Estimated effective dose (mSv)	14.0 ± 3.9	8.8 ± 3.8 (37.4%)	7.2 ± 1.4 (48.6%)	5.4 ± 0.6 (61.4%)	3.5 ± 0.9 (75.0%)

Each % value gives the reduction in relation to protocol A (standard FBP protocol at 120 kV).

For p values, see Fig. 4.

Fig. 4.

Boxplots displaying (a) the volume CT dose index (CTDIvol), (b) the total DLP, and (c) the estimated effective dose. The non-linear correlation of the total DLP and the effective dose arises from different age-specific conversion factors at 120 and 100 kV (or 80 kV) tube voltage (see “Material and Methods” section) (18). *P < 0.001.

Discussion

Studies investigating ASIR in abdominal CT have shown that IR algorithms significantly reduce dose while maintaining, or in some cases even improving, image quality (14,20,21).

Investigating IR techniques in CT follow-up after EVAR, Schabel et al. reported dose reduction at 120 kV of up to 50% using sinogram-affirmed iterative reconstruction (SAFIRE, Siemens Healthcare, Erlangen, Germany) (17). In comparison, our data show a 37.4% dose reduction for 40% ASIR compared to FBP at 120 kV. However, SAFIRE is not directly comparable to ASIR due to a different statistical IR algorithm. Moreover, Schabel et al. did not assess additional lowered tube voltage (17). Our data show reduced doses by 48.6% at 100 kV and by 61.4% at 80 kV.

Finally, using fully IR techniques such as model-based iterative reconstruction (MBIR; GE Healthcare, Milwaukee, WI, USA) technique, Hansen et al. reported a reduction of radiation dose by up to 73% (compared to ASIR) in CTA after EVAR at 100 kV (8). While this level of dose reduction reflects tremendous progress in radiation safety, the clinical application of fully IR techniques such as MBIR is limited. MBIR has high demands on hardware and is limited by significantly longer reconstruction times, even with today's fastest processor speeds (8,22). Therefore, fully IR techniques have not found their way into clinical routine yet. Nevertheless we could reach similar levels of dose reduction combining 40% ASIR and 80 kV with a single-acquisition CTA protocol using split-bolus contrast medium administration. With this technique, our data show a radiation dose reduction of up to 75%. Reliable visualization of endoleaks without using a second acquisition phase was already reported for this purpose (23).

While IR techniques may lead to blurring the edges and loss of critical details if overdrawn, adequate image quality is essential for preserving diagnostic information (8,12 –15). Hence, the main purpose of this study was to assess a potential negative impact of ASIR technique on image quality in CTA after EVAR. Our data show same quantitative image grading (SNR and CNR) for images using 40% ASIR at 120 kV, compared with the standard protocol, while qualitative image grading was the same or even significantly higher. We explain the higher qualitative image rating by a generally smoother and therefore subjectivley better visual impression of ASIR acquired images compared to the plain FBP algorithm.

The secondary purpose of our study was to evaluate ASIR in combination with kV reduction. By lowering the tube voltage from 100 kV to 80 kV, we found a decrease in image quality, however, without relevantly degrading diagnostic assessment. In particular, the most significant contrast between intra-stent lumen and extra-stent lumen did not significantly decrease at low kV. The intra-stent attenuation and intra-stent to extra-stent contrast even increased at 80 kV. This can be explained by the iodine K-edge, which results in higher X-ray absorption by iodine in the low-kV spectrum (24). For the CT follow-up after EVAR, this may result in better detectability of small endoleaks at low kV compared to 120 kV. Our data show an increase in image noise at low kV. To regain a better visual impression, we evaluated an additional increase in reconstruction ASIR level at 80 kV.

For qualitative image grading, our data show high to total interrater agreement (about 60–100%). In contrast, weighted Cohen's K-test showed only fair to moderate interrater agreement. This paradoxical correlation is attributable to an imbalance in marginal values and results in false low K-values when actual interrater agreement is high (25,26).

As there is still no recommendation for how to combine ASIR and low-kV CT protocols in the follow-up after EVAR, we developed a tailored follow-up based on our results, which optimally exploits the dose reduction potential of ASIR in the long-term surveillance of patients after EVAR. We suggest the use of 40% ASIR in combination with kV reduction depending on BMI in CTA follow-up after EVAR (Fig. 5). However, in the periprocedural setting we argue against a combination with kV reduction, due to the higher rate of complications in the first year after EVAR implantation. Nevertheless, in the long-term follow-up the combination of ASIR with kV reduction is reasonable and can be archieved without decisive impairments of diagnostic confidence based on our results.

Fig. 5.

Flowchart of a tailored CTA protocol in the long-term surveillance of patients after EVAR, based on our results, we suggest the use of 40% ASIR in combination with kV reduction depending on BMI. In the periprocedural setting, we argue against a combination with kV reduction, due to the higher rate of complications in the first year after EVAR implantation. In the long-term follow-up, the combination of ASIR with kV reduction is reasonable and can be archieved without decisive impairments of diagnostic confidence.

Our study has some limitations. No explicit patient group matching was performed; however, patient parameters match well and show no significant differences. The low-kV ASIR protocols include relatively few scans. The necessary size of 35 scans as defined by power analysis was reached for 120-kV protocols A and B, ensuring a statisitcally realiable analysis concerning the main goal of the present study. Due to recruiting problems, the size of 35 scans could not be reached in the low-kV protocols C and D. Hence, our secondary purpose could not be corroborated by solid statistical power and should hence be affirmed in larger numbers of scans. Furthermore we did not focus on the possibility of lowering contrast agent volume due to higher iodine attenuation in low-kV protocols as described, e.g. by Boning et al. in oncologic patients (21). This should be investigated in future studies, as it may reduce the risk of contrast-induced kidney injury (27).

In conclusion, ASIR contributes to significant dose reduction without relevant impairment of image quality or diagnostic confidence in CTA after EVAR. We recommend an adapted follow-up regimen introducing ASIR in combination with low-kV CT acquisition in the long-term surveillance of patients after EVAR.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Maddu

Telleria

Shuaib

et al.

Nontraumatic acute aortic emergencies: Part 2, Pre- and postsurgical complications related to aortic aneurysm in the emergency clinical setting. Am J Roentgenol 2014; 202: 666–674.

Sampson

Norman

Fowkes

et al.

Estimation of global and regional incidence and prevalence of abdominal aortic aneurysms 1990 to 2010. Glob Heart 2014; 9: 159–170.

Prinssen

Wixon

Buskens

et al.

Surveillance after endovascular aneurysm repair: diagnostics, complications, and associated costs. Ann Vasc Surg 2004; 18: 421–427.

Mita

Arita

Matsunaga

et al.

Complications of endovascular repair for thoracic and abdominal aortic aneurysm: an imaging spectrum. Radiographics 2000; 20: 1263–1278.

Liaw

Clark

Gibbs

et al.

Update: Complications and management of infrarenal EVAR. Eur J Radiol 2009; 71: 541–551.

Bauchaortenaneurysma und Beckenarterienaneurysma (S2). Leitlinien zu Diagnostik und Therapie in der Gefäßchirurgie , Berlin, Heidelberg: Springer, 2010.

Moll

Powell

Fraedrich

et al.

Management of abdominal aortic aneurysms clinical practice guidelines of the European society for vascular surgery. Eur J Vasc Endovasc Surg 2011; 41(Suppl 1): S1–S58.

Hansen

Kaza

Maturen

et al.

Evaluation of low-dose CT angiography with model-based iterative reconstruction after endovascular aneurysm repair of a thoracic or abdominal aortic aneurysm. Am J Roentgenol 2014; 202: 648–655.

Nyheim

Staxrud

Jorgensen

et al.

Radiation exposure in patients treated with endovascular aneurysm repair: what is the risk of cancer, and can we justify treating younger patients?. Acta Radiol 2017; 58: 323–330.

10.

Gervaise

Osemont

Lecocq

et al.

CT image quality improvement using adaptive iterative dose reduction with wide-volume acquisition on 320-detector CT. Eur Radiol 2012; 22: 295–301.

11.

Lehti

Nyman

Soderberg

et al.

80-kVp CT angiography for endovascular aneurysm repair follow-up with halved contrast medium dose and preserved diagnostic quality. Acta Radiol 2016; 57: 279–286.

12.

Willemink

de Jong

Leiner

et al.

Iterative reconstruction techniques for computed tomography Part 1: technical principles. Eur Radiol 2013; 23: 1623–1631.

13.

Willemink

Leiner

de Jong

et al.

Iterative reconstruction techniques for computed tomography part 2: initial results in dose reduction and image quality. Eur Radiol 2013; 23: 1632–1642.

14.

Kahn

Grupp

Rotzinger

et al.

CT for evaluation of potential renal donors - how does iterative reconstruction influence image quality and dose?. Eur J Radiol 2014; 83: 1332–1336.

15.

Hung

Sun

et al.

How far can the radiation dose be lowered in head CT with iterative reconstruction? Analysis of imaging quality and diagnostic accuracy. Eur Radiol 2013; 23: 2612–2621.

16.

Beister

Kolditz

Kalender

. Iterative reconstruction methods in X-ray CT. Phys Med 2012; 28: 94–108.

17.

Schabel

Fenchel

Schmidt

et al.

Clinical evaluation and potential radiation dose reduction of the novel sinogram-affirmed iterative reconstruction technique (SAFIRE) in abdominal computed tomography angiography. Acad Radiol 2013; 20: 165–172.

18.

Deak

Smal

Kalender

. Multisection CT protocols: sex- and age-specific conversion factors used to determine effective dose from dose-length product. Radiology 2010; 257: 158–166.

19.

Landis

Koch

. The measurement of observer agreement for categorical data. Biometrics 1977; 33: 159–174.

20.

Boning

Schafer

Grupp

et al.

Comparison of applied dose and image quality in staging CT of neuroendocrine tumor patients using standard filtered back projection and adaptive statistical iterative reconstruction. Eur J Radiol 2015; 84: 1601–1607.

21.

Boning

Kahn

Kaul

et al.

CT follow-up in patients with neuroendocrine tumors (NETs): combined radiation and contrast dose reduction. Acta Radiol 2017. doi: 10.1177/0284185117726101.

22.

Shuman

Green

Busey

et al.

Model-based iterative reconstruction versus adaptive statistical iterative reconstruction and filtered back projection in liver 64-MDCT: focal lesion detection, lesion conspicuity, and image noise. Am J Roentgenol 2013; 200: 1071–1076.

23.

de Bucourt

Muhler

Kroncke

et al.

Endoleak after endovascular aneurysm repair: evaluation of a single-acquisition CTA protocol using a prebolus. J Endovasc Ther 2011; 18: 771–778.

24.

Kalva

Sahani

Hahn

et al.

Using the K-edge to improve contrast conspicuity and to lower radiation dose with a 16-MDCT: a phantom and human study. J Comput Assist Tomogr 2006; 30: 391–397.

25.

Cicchetti

Feinstein

. High agreement but low kappa: II. Resolving the paradoxes. J Clin Epidemiol 1990; 43: 551–558.

26.

Feinstein

Cicchetti

. High agreement but low kappa: I. The problems of two paradoxes. J Clin Epidemiol 1990; 43: 543–549.

27.

Andreucci

Faga

Pisani

et al.

Acute kidney injury by radiographic contrast media: pathogenesis and prevention. Biomed Res Int 2014; 2014: 362725.