Effect of breath control on hepatic shear wave elasticity and dispersion in pediatric patients
Article information
Abstract
Purpose
This study aimed to evaluate the effect of breath control on the reliability of two-dimensional shear wave elastography (2D-SWE) and shear wave dispersion (SWD) measurements in pediatric patients.
Methods
This study included a retrospective cohort of 163 children and a prospective cohort of 27 children (aged 8-17 years). All participants underwent 2D-SWE and SWD under both free-breathing and breath-hold conditions between September 2021 and February 2023. The prospective cohort also underwent magnetic resonance elastography (MRE). Liver stiffness and dispersion values were compared between respiratory conditions. Inter- and intra-rater agreements were assessed, and correlations with MRE were analyzed in the prospective cohort.
Results
Liver stiffness and dispersion values were significantly higher during free-breathing compared to breath-hold (mean differences: 0.22 kPa and 0.39 m/s/kHz, respectively; both P<0.01). Breath-hold improved inter-rater agreement for 2D-SWE (intraclass correlation coefficient [ICC], 0.94 vs. 0.83; P=0.005) and SWD (ICC, 0.85 vs. 0.70; P=0.048). Intra-rater agreement for 2D-SWE (ICC, 0.88 vs. 0.88; P>0.99) and SWD (ICC, 0.70 vs. 0.74; P=0.396) remained moderate to good and did not differ significantly between conditions. The correlation between 2D-SWE and MRE was stronger under breath-hold than free-breathing (r=0.73 vs. r=0.56), although this difference was not statistically significant (P=0.299).
Conclusion
Breath-holding increases the reliability of pediatric 2D-SWE and SWD by improving inter-rater agreement and correlation with MRE. However, free-breathing also demonstrates comparable reproducibility with minimal bias, supporting its clinical feasibility for use in uncooperative pediatric patients.
Introduction
Recent advances in noninvasive ultrasound-based diagnostic techniques have broadened the scope of liver disease evaluation in pediatric patients [1,2]. Among these modalities, shear wave elastography (SWE) and shear wave dispersion (SWD) are effective for assessing liver stiffness and tissue viscosity, respectively [3-5]. SWE plays a crucial role in diagnosing and monitoring pediatric liver diseases, including metabolic dysfunction-associated steatotic liver disease (MASLD), sinusoidal obstruction syndrome, biliary atresia, and various neonatal liver disorders [6-9]. SWD is expected to further enhance the assessment of hepatic viscoelastic properties, enabling more comprehensive evaluation of liver tissue characteristics [10-12].
However, the reliability of ultrasound-based techniques is influenced by multiple patient- and examiner-dependent factors, including the placement of the sampling box and both inter- and intra-observer variability [13-16]. Respiratory motion is a recognized source of measurement variability, particularly in the assessment of liver stiffness and dispersion [17-19]. Breath-holding is generally considered effective for reducing motion artifacts and improving measurement consistency [20,21]. Nonetheless, several studies have reported that two-dimensional shear wave elastography (2D-SWE) performed during free-breathing provides reliability comparable to breath-hold measurements, with additional benefits such as shorter examination time and improved patient compliance [2,22]. Conversely, one study demonstrated that liver stiffness measurements obtained during free-breathing were systematically lower than those acquired during breath-holding, with a mean difference of -11.1% [19]. The reported effects of breath control on liver stiffness measurements remain inconsistent across studies, and intra-rater agreement between free-breathing and breath-hold conditions has not yet been fully investigated.
This study aimed to evaluate the effect of breath control on the reliability of 2D-SWE and SWD in pediatric patients. The findings are expected to contribute valuable evidence supporting the development of pediatric ultrasound guidelines.
Materials and Methods
Compliance with Ethical Standards
This study was approved by the Institutional Review Board of Seoul National University Hospital (IRB No. H-2501-100-1607 for the retrospective cohort; IRB No. H-2111-099-1272 for the prospective cohort). Informed consent was waived for the retrospective cohort. For the prospective cohort, written informed consent was obtained from each participant and their parent or legal guardian.
Study Cohorts
The retrospective cohort comprised 163 children (142 boys; age range, 8 to 17 years; median age, 12 years; mean body mass index, 25.4 kg/m2) who underwent both 2D-SWE and SWD under freebreathing and breath-hold conditions for evaluation of suspected MASLD between September 2021 and February 2023. Patients who were enrolled in the prospective cohort were excluded from the retrospective analysis.
The prospective cohort consisted of 27 children (25 boys; age range, 8 to 17 years; median age, 12 years; mean body mass index, 27.3 kg/m2) who were referred for both ultrasound elastography and magnetic resonance elastography (MRE) between January and September 2022. Each participant underwent two ultrasound elastography examinations, performed 1 week apart by two experienced pediatric radiologists (S.L. and Y.H.C., with 10 and 15 years of experience, respectively). MRE was performed on the same day as the initial ultrasound elastography.
Ultrasound Examination
This study was conducted as part of a quality assurance program to evaluate the adequacy of ultrasound elastography measurements after the introduction of a new ultrasound system (Aplio i800, Canon Medical Systems, Otawara, Japan) equipped with 2D-SWE and SWD capabilities. Ultrasound examinations were performed using a 1-6 MHz convex transducer by a single radiologist (S.L.) for the retrospective cohort and by two radiologists (S.L. and Y.H.C.) for the prospective cohort. For the prospective cohort, follow-up examinations were conducted with the radiologists blinded to the initial results to minimize bias.
All patients were instructed to fast for 2-6 hours before the examination, depending on age. 2D-SWE and SWD images were acquired using a 2×2 cm2 sampling box placed at least 15 mm below the liver capsule in the right hepatic lobe via an intercostal approach, thereby minimizing reverberation artifacts. Quad-view mode was used to simultaneously display the 2D-SWE map, SWD map, propagation map, and corresponding gray-scale image. A circular region of interest (ROI) with a 1 cm diameter was placed over a homogeneous color-coded area within the parallel wave propagation zone to simultaneously measure liver stiffness and dispersion slope. Liver stiffness was reported in kilopascals (kPa) as Young’s modulus, and dispersion was measured in meters per second per kilohertz (m/s/kHz). Measurements were obtained under two respiratory conditions: free-breathing during normal gentle respiration and breath-holding at end-expiration, during which patients were instructed to hold their breath for as long as possible [23]. Both measurements were obtained at the same anatomical location using identical machine parameters.
For each respiratory condition, 10 repeated measurements were acquired, and the median liver stiffness and dispersion slope values were calculated. The interquartile range (IQR) and the IQR-to-median ratio were also recorded. An IQR-to-median ratio <30% was considered indicative of measurement reliability [20]. The coefficient of variation (CV), defined as the ratio of the standard deviation to the mean liver stiffness, was calculated to assess measurement repeatability. A lower CV was interpreted as greater repeatability.
Magnetic Resonance Elastography
Liver magnetic resonance imaging, including MRE, was performed using a 3.0-T scanner (SIGNA Premier, GE Healthcare, Chicago, IL, USA), which served as the reference standard. A motion-sensitized spin-echo sequence was used, with sinusoidal displacement-encoding gradients synchronized to a mechanical driver via a trigger from the magnetic resonance spectrometer [24]. Patients were positioned supine, and a passive driver (18.5 cm in diameter, 3.5 cm thick) was placed on the right anterior chest wall over the liver, centered at the xiphisternum, and connected to an active driver generating 60-Hz vibrations. Imaging was conducted during a single breath-hold at end-expiration using the following parameters: repetition time/echo time, 1,000.3/54 ms; matrix, 64×64; flip angle, 90°; slice thickness, 10 mm; slice interval, 13 mm; and 10 axial slices.
Liver stiffness was quantified by manually placing up to four circular ROIs on a single axial slice covering the largest portion of the right hepatic lobe. ROIs were selected on wave images showing regular propagation with minimal reflection or artifacts, while avoiding large vessels, liver edges, and motion artifacts [25]. Each ROI was transferred to the corresponding stiffness map, and the mean shear modulus (kPa) was calculated by averaging all ROIs.
Statistical Analysis
In the retrospective cohort, the paired t-test was used to compare liver stiffness and dispersion metrics (median, IQR, and CV) between the two breathing conditions. Linear regression and Pearson correlation analyses were performed to evaluate the relationship between free-breathing and breath-hold measurements. Agreement between the two conditions was assessed using Bland-Altman analysis, with agreement defined as the presence of zero within the 95% confidence interval (CI) of the mean difference. Deming regression was also performed, with agreement defined as the presence of 1 within the 95% CI of the slope and zero in the intercept [26]. Pearson correlation coefficients were interpreted as follows: no association (r=0.0-0.2), weak (r=0.2-0.5), moderate (r=0.5-0.8), and strong to perfect (r=0.8-1.0) [27].
In the prospective cohort, linear regression and Pearson correlation analyses were conducted separately for each breathing condition to evaluate the correlation between liver stiffness measured by 2D-SWE and MRE. Differences in correlation coefficients between breathing conditions were assessed using the Fisher r-to-z transformation. Intraclass correlation coefficients (ICCs) were calculated to assess inter- and intra-rater agreement for each breathing condition. Two-way random-effects models were used for inter-rater agreement, and two-way mixed-effects models were used for intra-rater agreement, based on single measurements. ICCs were interpreted as indicating poor (<0.50), moderate (0.50-0.75), good (0.75-0.90), or excellent (>0.90) agreement [28].
All statistical analyses were performed using Python version 3.13.1 (Python Software Foundation, Wilmington, DE, USA) with the following libraries: NumPy (1.22.0), pandas (1.4.0), matplotlib (3.5.1), seaborn (0.11.2), Pingouin (0.4.0), and SciPy (1.8.0).
Results
Results Retrospective Cohort Analysis
For 2D-SWE, the median liver stiffness was significantly lower under breath-hold compared to the free-breathing condition (5.02±1.10 vs. 5.24±1.23 kPa, P=0.005). The IQR and CV were also significantly reduced under breath-hold (0.91±0.63 vs. 1.03±0.69 kPa, P=0.031 and 0.109±0.001 vs. 0.122±0.012, P=0.012, respectively), indicating improved measurement consistency and repeatability. The Pearson correlation coefficient was 0.65 (P<0.001), indicating a moderate positive correlation between the two respiratory conditions. However, Deming regression revealed systematic disagreement (slope, 0.84; 95% CI, 0.73 to 0.95; P=0.006 and intercept, 0.61; 95% CI, 0.01 to 1.22; P=0.046) (Fig. 1A). The mean difference between two conditions was 0.22 kPa (95% CI, 0.07 to 0.37; P=0.005), with limits of agreement ranging from -1.69 kPa (95% CI, -1.95 to -1.44 kPa) to 2.13 kPa (95% CI, 1.87 to 2.39 kPa) (Fig. 1B).
Comparison of ultrasound elastography measurements under free-breathing and breath-hold conditions.
A. Linear regression (solid line) and Deming regression (dashed line) of two-dimensional shear wave elastography (2D-SWE) measurements are shown. B. Bland-Altman analysis of 2D-SWE measurements is shown. C. Linear regression (solid line) and Deming regression (dashed line) of shear wave dispersion (SWD) measurements are shown. D. Bland-Altman analysis of SWD measurements is shown. In panels B and D, the horizontal line at y=0 represents perfect agreement. The bold black dashed line and thin black dashed lines indicate the mean difference and its 95% confidence interval (CI), respectively. The bold gray dashed lines and thin gray dashed lines represent the upper and lower limits of agreement (LoA; mean difference±1.96 standard deviations) and their respective 95% CI.
Similarly, for SWD, the median value was significantly lower under breath-hold than under the free-breathing condition (11.68±1.81 vs. 12.07±1.92 m/s/kHz, P=0.004). Both the IQR and CV were also significantly reduced under breath-hold (1.43±0.99 vs. 1.79±1.16 m/s/kHz, P=0.001 and 0.049±0.001 vs. 0.092±0.045, P=0.001, respectively), suggesting that breath control improved measurement stability. The Pearson correlation coefficient was 0.58 (P<0.001), also indicating a moderate positive correlation. Deming regression showed no evidence of systematic disagreement (slope, 0.90; 95% CI, 0.77 to 1.03; P=0.143 and intercept, 0.79; 95% CI, -0.81 to 2.39; P=0.330) (Fig. 1C). The mean difference was 0.39 m/s/kHz (95% CI, 0.12 to 0.65; P=0.004), with limits of agreement ranging from -2.97 m/s/kHz (95% CI, -3.42 to -2.51) to 3.74 m/s/kHz (95% CI, 3.29 to 4.19) (Fig. 1D).
Prospective Cohort Analysis
The correlation between liver stiffness values measured by 2D-SWE and MRE is presented in Fig. 2. The Pearson correlation coefficient was 0.56 (P=0.002) for free-breathing and 0.73 (P<0.001) for breath-hold, both indicating moderate positive correlations. Although the correlation was stronger during breath-hold, the difference was not statistically significant (P=0.299).
Comparison of two-dimensional shear wave elastography (2D-SWE) and magnetic resonance elastography (MRE).
Linear regression analysis of liver stiffness measured by 2D-SWE and MRE are shown under free-breathing (A) and breath-hold (B) conditions. The regression equations are as follows: (A) μSWE_FB=2.103×μMRE+0.797, and (B) μSWE_BH=2.019×μMRE+0.728. The thick solid line represents the linear regression fit. Dashed lines indicate the 95% confidence intervals, and thinner solid lines indicate the 95% prediction intervals.
Table 1 summarizes the inter- and intra-rater agreement for 2D-SWE and SWD measurements under free-breathing and breath-hold conditions. For 2D-SWE, inter-rater agreement was good under free-breathing conditions (ICC, 0.83; 95% CI, 0.64 to 0.92) and excellent during breath-hold (ICC, 0.94; 95% CI, 0.86 to 0.97), with significantly higher agreement under breath-hold (P=0.005). Intra-rater agreement remained consistently good in both conditions (free-breathing: ICC, 0.88; 95% CI, 0.74 to 0.94 and breath-hold: ICC, 0.88; 95% CI, 0.74 to 0.93), with no significant difference observed (P>0.99).
Inter- and intra-rater agreement of ultrasound elastography measurements under free-breathing and breath-hold conditions
For SWD, inter-rater agreement was moderate under free-breathing (ICC, 0.70; 95% CI, 0.35 to 0.86) and good during breath-hold (ICC, 0.85; 95% CI, 0.68 to 0.93), with significantly higher agreement in the breath-hold condition (P=0.048). Intra-rater agreement was moderate in both conditions (free-breathing: ICC, 0.70; 95% CI, 0.37 to 0.86 and breath-hold: ICC, 0.74; 95% CI, 0.45 to 0.88), again with no significant difference (P=0.396).
Representative color-coded 2D-SWE and SWD maps, as well as the propagation map, obtained under both breathing conditions in an 11-year-old boy with severe hepatic steatosis are shown in Fig. 3.
Representative cases of two-dimensional shear wave elastography (2D-SWE) and shear wave dispersion (SWD) under freebreathing (A) and breath-hold (B) conditions.
Color-coded 2D-SWE map (left), propagation map (middle), and color-coded SWD map (right) obtained from an 11-year-old boy with severe fatty liver. The region of interest was placed within a uniformly color-coded area. The propagation map under free-breathing demonstrates a more pronounced wavy pattern. Based on 10 elastograms, liver stiffness and dispersion values were comparable between free-breathing and breath-hold conditions (2D-SWE: 4.8±0.8 vs. 4.6±0.4 kPa; SWD: 11.33±0.96 vs. 10.71±0.91 m/s/kHz).
Discussion
In this study, both 2D-SWE and SWD measurements obtained during free-breathing and breath-hold conditions demonstrated significant differences, with free-breathing yielding higher liver stiffness and dispersion values (by 0.22 kPa and 0.39 m/s/kHz, respectively). Liver stiffness measured by 2D-SWE showed a higher correlation with MRE under breath-hold conditions, although this difference was not statistically significant. Inter-rater agreement improved under breath-hold for both 2D-SWE and SWD, while intra-rater agreement showed no significant difference in either condition. These findings suggest that breath-holding is preferable when feasible. Nonetheless, given the lack of a significant difference in correlation with MRE, minimal systematic bias, and acceptable inter-rater agreement, free-breathing may serve as a practical alternative when breath-holding is not feasible for measuring liver stiffness and dispersion values.
Current clinical guidelines for SWE recommend acquiring liver stiffness measurements during breath-hold to minimize respiratory motion artifacts [3,20]. However, the influence of breath control appears to differ across elastography modalities, including transient elastography [17], acoustic radiation force impulse imaging [18,29], and 2D-SWE [19,22]. Although recent technical advancements have enabled rapid acquisition of 2D-SWE and reduced motion artifacts, the effect of respiratory phase remains a subject of controversy. Jung et al. reported a strong correlation (r=0.97) between breath-hold and free-breathing measurements; however, their results may have been disproportionately influenced by a few high-stiffness outliers (≥10 kPa), potentially obscuring variability in lower stiffness ranges [22]. In contrast, Yoon et al. [19] reported a systematic bias, with liver stiffness obtained during free-breathing being 11.1% lower than that acquired during breath-hold, which differs from the findings of the present study. This discrepancy may be attributable to the shorter expiration period in children, making it difficult to clearly define the measurement time window, as well as vendor-specific differences in 2D-SWE sampling box stabilization. Although current guidelines suggest that liver stiffness measurements in children can be performed under free-breathing, the evidence remains insufficient to fully support this approach [20].
The measurement disagreement observed between the two breathing conditions in this study may be explained by three principal factors: more consistent hepatic venous return, stable intra-abdominal pressure, and reduced motion artifacts during breath-hold. First, inspiration increases hepatic venous return, thereby decreasing intrahepatic blood volume and lowering liver stiffness [17,30]. Acquiring measurements at the same end-expiratory phase minimizes intrahepatic blood volume fluctuations and maintains a consistent liver position across acquisitions. Second, deep inspiration elevates intra-abdominal pressure, which has been shown to increase liver stiffness relative to resting expiration [23,31]; thus, breath-holding at end-expiration may reduce this effect and help prevent overestimation. Third, it is proposed that respiration-induced motion perpendicular to the direction of shear wave propagation may lead to overestimation of shear wave displacement generated by the acoustic radiation force. Ideally, the isophase lines in a 2D-SWE propagation map should appear parallel, indicating uniform wave propagation. However, respiratory motion can distort these lines, producing apparent diagonal wavefronts [32]. This artifact may artificially increase calculated shear wave velocity, thereby overestimating both liver stiffness and dispersion and contributing to the systematic bias observed in this study.
Although 2D-SWE under breath-hold conditions demonstrated a stronger correlation with MRE compared to free-breathing, the difference was not statistically significant in this study. In the prospective cohort, MRE-derived liver stiffness values were used as the reference standard due to the modality’s higher diagnostic accuracy and lower rate of technical failure [33-35]. Notably, MRE was performed during breath-holding. Despite this, the moderate correlation observed with free-breathing 2D-SWE (r=0.56) suggests that a clinically acceptable degree of agreement can still be achieved without breath control.
In terms of inter- and intra-rater agreement, previous studies in adult populations have reported inter-rater agreement for liver stiffness measurements using 2D-SWE under breath-hold conditions ranging from 0.76 to 0.97, and intra-rater agreement ranging from 0.63 to 0.98 [19,36-39]. These findings are consistent with those of the present study, which demonstrated inter- and intra-rater agreement values of 0.94 and 0.88, respectively. Although breath-holding is generally more challenging for pediatric patients, the high inter-rater agreement observed here may be attributed to the consistent expertise of both examiners, who were experienced pediatric radiologists, in contrast to previous studies that involved more heterogeneous operator backgrounds. Under free-breathing conditions, inter-rater agreement for both 2D-SWE and SWD was lower than under breath-hold but still remained within the moderate to good range. Notably, intra-rater agreement did not differ significantly between the two respiratory conditions for either technique. Given the maintained intra-rater agreement for 2D-SWE under free-breathing (ICC, 0.88), performing follow-up examinations with the same operator may be a practical approach to enhancing measurement reliability in children unable to comply with breath-hold instructions.
This study offers clinical value by evaluating the feasibility of a free-breathing approach to address practical challenges in pediatric ultrasound. In children, liver biopsy is rarely performed, limiting the opportunity to directly validate the accuracy of 2D-SWE for fibrosis staging. Despite this limitation, this study comprehensively assessed the impact of breath control using a relatively large retrospective cohort and a prospective cohort that underwent both MRE and repeated ultrasound examinations. This study population primarily included patients referred for ultrasound elastography due to suspected MASLD, with a high prevalence of obesity, male sex, and pubertal age, which are factors independently associated with MASLD risk [40]. These demographic characteristics strengthen the clinical relevance and real-world applicability of this study’s findings.
This study has several limitations. First, patients under 8 years of age were excluded from analysis. Given the increased respiratory motion and shorter respiratory cycles in younger children, the reliability of free-breathing measurements in this age group remains uncertain. Second, although reduced examination time is often cited as an advantage of free-breathing [22], ultrasound examination time was not measured in this study. Third, histologic staging via liver biopsy was not available, which limits the direct comparison of diagnostic performance between the two respiratory protocols. Fourth, the sample size of the prospective cohort was insufficient to statistically confirm whether breath-holding improves the correlation between 2D-SWE and MRE.
In conclusion, liver stiffness and dispersion values obtained under free-breathing and breath-hold conditions showed statistically significant differences. Breath-holding resulted in improved inter-rater agreement and a higher correlation with MRE compared to free-breathing, although the difference in correlation was not statistically significant. Given the acceptable intra-rater agreement and moderate correlation with MRE under free-breathing conditions, this approach may be considered a feasible alternative when breath-holding is not achievable in pediatric patients.
Notes
Author Contributions
Conceptualization: Rhee C, Lee S, Choi YH, Cheon JE. Data acquisition: Lee S, Choi YH. Data analysis or interpretation: Rhee C, Lee S, Hwang JY, Cheon JE. Drafting of the manuscript: Rhee C. Critical revision of the manuscript: Rhee C, Lee S, Choi YH, Hwang JY, Cheon JE. Approval of the final version of the manuscript: all authors.
Conflict of Interest
Jae-Yeon Hwang and Jung-Eun Cheon serve as Editors for the Ultrasonography, but have no role in the decision to publish this article. All remaining authors have declared no conflicts of interest.
Acknowledgments
This work was partly supported by Canon Medical Systems (Grant No. 0620212080) and by the National Research Foundation of Korea (Grant No. RS-2023-00250062), funded by the Korean government (MSIT).
References
Article information Continued
Notes
Key point
Recent advances in ultrasound elastography have reduced artifacts from respiratory motion through rapid image acquisition; however, the effect of breath control on measurement reliability remains unclear. Two-dimensional shear wave elastography and shear wave dispersion measurements differed significantly between free-breathing and breath-hold conditions, with breath-hold demonstrating stronger correlations with magnetic resonance elastography and higher inter-rater agreement. Given the minimal differences between breathing conditions and the acceptable reliability of free-breathing, this method may be a practical option for pediatric patients who are unable to perform breath-holds.