- Email: [email protected]
Application of ARFI-SWV in Stiffness Measurement of the Abdominal Wall Musculature: A Pilot Feasibility Study
ARTICLE IN PRESS Ultrasound in Med. & Biol., Vol. 00, No. 00, pp. 18, 2018 Copyright © 2018 World Federation for Ultrasound in Medicine and Biology. All rights reserved. Printed in the USA. All rights reserved. 0301-5629/$ - see front matter
https://doi.org/10.1016/j.ultrasmedbio.2018.05.007
Original Contribution TAGEDH1APPLICATION OF ARFI-SWV IN STIFFNESS MEASUREMENT OF THE ABDOMINAL WALL MUSCULATURE: A PILOT FEASIBILITY STUDYTAGEDN TAGEDPD1X XDAVID A. GABRIELSEND2X X* D3X XMARTIN J. CARNEYD4yX X D5X XJASON M. WEISSLERD6yX X D7X XMICHAEL A. LANNIDyX8 X D9X XJORGE HERNANDEZD10X X* D1X XLAITH R. SULTAND12zX X D13X XFABIOLA ENRIQUEZDyX14 X D15X XCHANDRA M. SEHGALD16zX X D17X XJOHN P. FISCHERD18yX X and D19X XANIL CHAUHANDzX20 X AGEDNTE
TagedP* Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA; y Division of Plastic Surgery, Department of Surgery, University of Pennsylvania, Philadelphia, PA, USA; and z Department of Radiology, University of Pennsylvania, Philadelphia, PA, USA (Received 16 May 2017; revised 9 April 2018; in final from 3 May 2018)
TagedPAbstract—The purpose of this study was to assess the feasibility of acoustic radiation force impulse shear wave velocity and textural features for characterizing abdominal wall musculature and to identify subject-related and technique-related factors that can potentially affect measurements. Median shear wave velocity measurements for the right external abdominal oblique were the same (1.89 § 0.16 m/s) for both the active group (healthy volunteers with active lifestyles) and the control group (age and body mass indexmatched volunteers from an ongoing hernia study). When corrected for thickness, the ratio of right external abdominal oblique shear wave velocity -to-muscle thickness was significantly higher in the control group than in the active volunteers (4.33 s¡1 versus 2.88 s¡1; p value 0.006). From the textural features studied for right external abdominal oblique, 8 features were found to be statistically different between the active and control groups. In conclusion, shear wave velocity is a feasible and reliable technique to evaluate the stiffness of the abdominal wall musculature. Sonographic texture features add additional characterization of abdominal wall musculature. (E-mail: Anil. [email protected]) © 2018 World Federation for Ultrasound in Medicine and Biology. All rights reserved. TagedPKey Words: ARFI-SWV, Abdominal wall musculature, Elastography, Ultrasound.
TagedPrisk factors lack the ability to assess the biomechanical properties of the abdominal wall musculature, an important parameter in predicting successful repair and disease progression (Bittner et al. 2014). Therefore, it is highly desirable to establish an imaging biomarker to fill this important gap in the understanding of the in vivo mechanical properties of the abdominal wall as well as their pathologic potential Gennisson et al. 2010; Tran et al. 2016). TagedPQuantitative imaging, using acoustic radiation force impulse shear wave velocity (ARFI-SWV), promises to be a novel and economical strategy to assess skeletal muscle properties, especially in the abdominal wall (Bercoff et al. 2004; Carpenter et al. 2015). Ultrasound elastography functions by comparing consecutive images before and after an invoked stimulus (McNally 2011). This position change is then quantified into the shear wave velocity (SWV) value, a measure of tissue stiffness (Hug et al. 2015). Elastography has
TAGEDH1INTRODUCTIONTAGEDN TagedPIncisional hernia is among the most common and serious surgical complications of abdominal surgery (Poulose et al. 2012). The incidence can be as high as 70% for high-risk (such as obese) populations (Mudge and Hughes 1985; Nho et al. 2012; Poulose et al. 2012). Despite advances, hernia repair continues to have a high failure rate, resulting in repeat surgeries, bridged repairs and nearly $3 billion lost to the U.S. healthcare system in addition to increasing morbidity for patients (Bower and Roth 2013; Burger et al. 2004; Forbes et al. 2009; Holihan et al. 2015; Poulose et al. 2012). The conventional modalities, computed tomography (CT) and magnetic resonance imaging (MRI), used to assess hernia Address correspondence to: Anil Chauhan, MD, Division of Abdominal Imaging, Department of Radiology, University of Pennsylvania, 3400 Spruce Street, Hospital of the University of Pennsylvania, University of Pennsylvania, Philadelphia, PA 19104, USA. E-mail: [email protected]
1
ARTICLE IN PRESS 2
Ultrasound in Medicine & Biology
TagedPshown success with evaluating neoplastic tissues as well as skeletal muscle biomechanical properties (Fahey et al. 2007; Hug et al. 2015; Palmeri et al. 2008). TagedPTissue texture analysis is another quantitative, yet underused, ultrasound-based imaging method. This technology primarily deals with the arrangement of various components of tissues in relation to one other. Currently, this technique is under investigation for use in the evaluation of liver fibrosis (Cao et al. 2005; Isono et al. 2017). Texture analysis similarly has the potential to quantitatively evaluate the deviation of muscle architecture from normal, such as with fibrofatty infiltration. TagedPGiven the paucity of knowledge regarding the biomechanical properties of the abdominal wall and the promise that SWV and texture analysis hold as a bridge to that gap, it is important to establish a set of standardized methods of evaluation (Botanlioglu et al. 2013; Kuo et al. 2013; Kwon et al. 2012). At least two studies have found SWV as a reliable method for the evaluation of the biomechanical properties of the abdominal wall (MacDonald et al. 2016; Tran et al. 2016). Further studies however are needed to specifically evaluate the various technical and subject-related factors that can impact the SWV measurements of abdominal wall musculature. Additionally, the texture analysis protocol while evaluating the abdominal wall musculature has not been established. Therefore, we tested the feasibility of SWV and texture analysis protocols of the abdominal wall musculature and to identify subject-related and technique-related factors that can potentially affect these modalities. TAGEDH1MATERIALS AND METHODSTAGEDN Patients TagedPA prospective cohort study was performed at the University of Pennsylvania with approval granted by the hospital’s institutional review board. A written consent was obtained from every subject. Patients were recruited from two sources: (i) from the surgery and radiology departments, 10 healthy active volunteers demonstrating high activity and a body mass index (BMI) <30, and (ii) from a prospective control group database of the study “ARFI Evaluation of Hernia Patients,” 10 patients. These patients acted as controls in our hernia study, having been age and BMI matched to hernia patients of the respective study, thereby creating a distinct group. Because demographic characteristics of these two patients differed significantly, we present the ultrasound findings in these two groups separately. TagedPPatients with a history of abdominal wall hernia or any abdominal surgeries were excluded from the study. Patients who exercised at least 48 h before the exam were also excluded. All individuals were screened using
Volume 00, Number 00, 2018
TagedPthe International Physical Activity Questionnaire (IPAQ) long-protocol guidelines to quantify their amount of activity (Craig et al. 2003). High-activity was defined according to the IPAQ guidelines that require patients to engage in “vigorous-intensity activity on at least 3 days and accumulating at least 1500 metabolic equivalent (MET)-minutes/week or 7 or more days of any combination of walking, moderate-intensity or vigorous intensity activities achieving a minimum of at least 3000 MET-minutes/week.” Image acquisition TagedPUltrasound images and SWV measurements were obtained using Siemens Acuson S3000 (Siemens Medical Inc., Mountain View, CA, USA). The linear 9 L4 probe was the primary probe used to obtain gray-scale images and SWV measurements in all the muscle groups. The right external oblique (REO) muscle was primarily evaluated, because this is the most mobilized musculature in hernia repair surgeries, and the right side is traditionally approached by a sonographer for abdominal exams. The curved 6 C1 and linear 12 L4 were additionally used for obtaining SWV measurements of the REO musculature in a similar fashion as that used for the 9 L4 probe, to also evaluate whether SWV velocities of musculature differ among various probes. All measurements were performed by one of the three sonographers, each with at least 1 y of experience in elastography and trained by a board-certified radiologist with 4 y of ultrasound experience. Gray-scale Images TagedPTwo gray-scale images in orthogonal planes of the external oblique were taken, using a 9 L4 probe with the focal zone in the longitudinally centermost portion of the muscular layer. The entire abdominal layer could be seen down to the peritoneum in each image. These images were acquired in the right flank at midpoint of the ribcage and anterior superior iliac spine, while patients were in the supine position. This position was also utilized for SWV measurement for REO and internal oblique musculature. SWV measurement TagedPMultiple SWV data sets of the REO were obtained: (i) baseline with minimal possible probe pressure (Fig. 1), (ii) after applying manual pressure with the probe (just before the point of discomfort for patients), (iii) after Valsalva maneuver, and (iv) using 12 L4 and 6 C1 probes in the same manner as that described using the 9 L4 probe. Additional data sets for the right internal oblique (RIO), the right rectus abdominis (RRM) and the left external oblique musculature (LEO) were obtained, using only the 9 L4 probe.
ARTICLE IN PRESS Elastography of Abdominal Wall Musculature D. A. GABRIELSEN et al.
3
Fig. 1. Technique of acoustic radiation force impulse shear wave velocity (ARFI-SWV) measurement. Region of interest (green square) is placed in the external oblique musculature,
TagedPTen successive SWV measurements were obtained for each data set, using point-shear wave technology. Each measurement represented a 10 £ 10 mm region of interest (ROI), which was placed at the center of the superior-most part of each muscle of interest. SWV was measured with parallel orientation of the external oblique muscle fibers to the ultrasound probe, as seen on B-mode ultrasonography, while patients were holding their breath. Parallel orientation was confirmed upon recognition of a few muscle-facial interfaces, appearing as a thin, continuous bright line. The median of 10 measurements was considered the final SWV measurement. The ratio of interquartile range (IQR) to the median value (IQR/M) was measured, and data sets with IQR/M >0.3 were considered unreliable (Bota et al. 2011). The data sets with >40% of invalid measurement were considered unreliable (i.e., when the machine records X.XX m/s as the SWV, alternatively meaning a void measurement). Texture analysis TagedPOne gray-scale image demonstrating all abdominal wall layers, with REO muscle fibers oriented parallel to the ultrasound probe, was identified for texture analysis using subcutaneous fat as an internal control (Fig. 2). Images were analyzed for a total of 21 ultrasound-based texture features using Mazda software (v. 4.6, Lodz, Poland). These features were extracted from ROIs placed in the external and internal oblique muscles as well as in the subcutaneous fat (Fig. 2). ROIs were placed by a research fellow, under direct supervision of a board-certified radiologist with 4 y of experience in quantitative ultrasound. Each ROI was as large as possible within the tissue of interest while staying away from artifacts.
Fig. 2. (a) shows a gray-scale image of abdominal wall musculature. The abdominal wall layers are labeled accordingly. SF, EO, IO and TA represents subcutaneous fat, external oblique, internal oblique and transversus abdominis, respectively. (b) Example of ROI placement on SF (blue), EO (red) and IO (green) for texture analysis.
TagedPThe texture features extracted from each image are statistical computations that indirectly represent the real texture by the nondeterministic properties that govern the distributions and relationships between the gray levels of an image. These features included parameters derived from the gray-level histogram, gray-level cooccurrence matrix, run-length matrix and absolute gradient analysis for each muscle (Szczypinski et al. 2009). In this study, we averaged distances for each parameter, calculating a single average parameter measurement used for analysis. TagedPThe first-order histogram is computed from the intensity of pixels. Features included mean brightness, variance, skewness, kurtosis and percentiles. The graylevel co-occurrence matrix is a second-order histogram, computed from intensities of pairs of pixels, where the spatial relationship of the two pixels in a pair is defined. The co-occurrence, matrix-based features are derived from a matrix to demonstrate statistics, such as angular second moment, contrast, correlation, sum of squares and various averages, variances, inverse moments and entropies. The run-length matrix holds counts of pixel runs with the specified gray-scale level and length. Features of this matrix include short-run emphasis inverse moment, long-run emphasis moment, gray-level nonuniformity, run-length nonuniformity and fraction of image
ARTICLE IN PRESS 4
Ultrasound in Medicine & Biology
Volume 00, Number 00, 2018
TagedPin runs. The absolute gradient is calculated using mean absolute gradient, variance of absolute gradient, skewness of absolute gradient and kurtosis of absolute gradient.
TagedPand the whole subject population was 19.234.2 (26.07 § 4.63). There were expected significant differences in age and BMI between the two groups (p value <0.0001).
Statistical analysis TagedPFor analytical purposes, the two subject groups (active and control) were separately analyzed in most scenarios, owing to the unique age and BMI distribution. Statistical analysis was performed to assess for significant differences between the two groups (in specific scenarios) as well as between various muscle groups. Analysis was performed using Microsoft Excel (Microsoft Corporation, Redmond, WA, USA) and the R programming language. The primary tool of comparison was a Student’s two-sample t test, assuming unequal variance. All reported p values were based on a two-tailed t test. The independent samples t test is used to test the hypothesis that the difference between the means of two groups is equal to 0 (the null hypothesis). When the p value is less than the conventional 0.05, the null hypothesis is rejected and the conclusion is that the two means do indeed differ significantly.
Validity and reliability of SWV measurements TagedPOut of a total of 160 attempted data sets, 2 data sets were invalid (high number of X.XX m/s values). These were present in the control group (post-Valsalva REO and 6 C1 probe REO measurements). Four data sets of REO using six C1 probes could not be obtained in the active volunteer group, because the system was unable to scan muscle located less than 1 cm beneath skin, such was the case in patients with minimal subcutaneous fat. Only three data sets had IQR/M more than 0.3 (two REO data sets using Valsalva maneuver and one REO data set using hard probe pressure).
TAGEDH1RESULTSTAGEDN Patients TagedPThe demographic characteristics of each group are detailed in Table 1. The age range for the control group was 5569 y (mean 61.6 § 4.86), the active volunteer group was 2333 y (mean 26.9 § 3.28) and the whole population was 2369 y (44.25 § 18.25). The BMI of the control group was 27.834.2 (mean 29.9 § 2.03), the volunteer group was 19.228 (mean 22.24 § 2.92)
Standard setup TagedPThe range of baseline resting median SWV measurements in m/s for REO was 1.502.96 (mean 1.89 § 0.16) and 1.402.51 (mean 1.89 § 0.13) for the active and control groups, respectively, with a range of 1.402.96 (mean 1.89 § 0.37) in all the patients. ARFI-SWV results TagedPThe SWV measurements were divided into intrinsic factors and extrinsic factors. Intrinsic factors included any passive, subject-related variables, such as the muscle to be analyzed and activity level. In the control group, the median SWV measurements in m/s for REO, RIO, RRM and LEO were 1.89, 1.40, 1.89 and 1.85, respectively. Similar values in m/s in the active volunteer group were 1.89, 1.80, 2.07 and 2.21, respectively.
Table 1. Subject demographic characteristics Subject group Control group
Active volunteers
Subject ID 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Sex Female Female Female Male Male Male Male Female Female Male Female Male Female Male Male Male Female Female Female Female
Age (y)
Height (in)
Weight (lb)
BMI (kg/m2)
Subcutaneous fat thickness (cm)
63 64 57 55 61 56 64 69 59 68 30 33 28 29 28 23 23 24 25 26
62.0 63.0 64.9 76.0 70.0 65.0 71.0 64.0 66.0 72.0 66.0 70.0 63.0 70.0 72.0 75.0 63.0 72.0 68.0 69.0
173 193 167 239 194 174 212 186 181 212 130 176 112 195 160 200 109 161 126 144
31.6 34.2 27.9 29.1 27.8 29.0 29.6 31.9 29.2 28.7 21.0 25.3 19.8 28.0 21.7 25.0 19.3 21.8 19.2 21.3
2.07 2.53 1.19 2.17 1.19 0.72 1.73 1.13 1.58 0.77 0.54 0.47 0.64 0.54 0.52 0.35 0.53 0.49 0.68
ARTICLE IN PRESS Elastography of Abdominal Wall Musculature D. A. GABRIELSEN et al.
5
Table 2. SWV and thickness measurements for ROI and activity levels Subject group Control group
Active volunteers
Region of interest
Range of SWVs (m/s)
Mean SWV § Variance (m/s)
Mean thickness § variance (cm)
p SWV against REO
p SWV against control
REO RIO RRM LEO REO RIO RRM LEO
1.492.96 0.901.71 1.143.65 1.242.71 1.402.51 1.432.39 1.132.29 1.502.80
1.89 § 0.16 1.40 § 0.05 1.89 § 0.52 1.85 § 0.18 1.89 § 0.13 1.80 § 0.10 2.07 § 0.23 2.21 § 0.19
0.43 § 0.22 0.81 § 0.22 0.65 § 0.98 0.98 § 0.31 -
0.004 0.97 0.81 0.57 0.35 0.09
0.99 0.005 0.5 0.07
SWV = shear-wave velocity; ROI = region of interest; REO = right external oblique; RIO = right internal oblique; RRM = right rectus abdominis; LEO = left external oblique.
TagedPInterestingly, the right IO SWV was significantly higher in the active volunteer group compared with the controls (p = 0.005), even though no statistically significant difference was noted for the REO, LEO or RRM SWV measurements (Table 2). TagedPExtrinsic factors were imposed while scanning by either the sonographer or the subject. These factors included high manual pressure by the sonographer, Valsalva maneuver by the subject and probe frequency. Compared with the standard data set of REO, significant differences were shown with Valsalva in control group (average 115.3% SWV increase, p < 0.001) as well as in the active volunteers (average 84.1% SWV increase, p < 0.001). Significant differences were also shown with high probe pressure in the control group (an average of 60.2% SWV increase, p < 0.001) and the active volunteers (an average of 95.8% SWV increase, p < 0.001). The 12 L4 and 6 C1 probes did not yield significant differences in SWV measurement compared with the standard 9 L4 probe in the REO in either group (Table 3). Muscle thickness correction of SWV measurements TagedPLiterature has suggested that muscles with higher thickness (or bulk) can lead to higher resting SWV measurements (Akagi et al. 2012; Kuo et al. 2013). To analyze this, we corrected the SWV measurements of REO and RIO muscles to the muscle thickness. The ratio of
TagedPREO SWV to muscle thickness was significantly higher in the control group than in the active volunteers (4.33 versus 2.88; p value 0.006), and there was no significant difference for these values in the RIO muscle (1.69 versus 1.76; p value 0.81). Texture analysis results TagedPFrom the features studied for external oblique muscles, 8 features were found to be statistically different between the 2 groups (Table 4). Mean brightness for the control group was higher (2.14) in compared with active patients (1.24) (p < 0.001). In addition, 1st, 10th, 50th and 90th percentiles (first histogram); contrast; sum average; difference variance (gray-level co-occurrence matrix); and gray-level variance (absolute gradient) were all significantly higher in the matched group. For features extracted from the internal oblique muscle, five texture features representing the first order histogram and gray level co-occurrence matrix were significantly higher in the control group compared with the active patients (Table 5). TAGEDH1DISCUSSIONTAGEDN TagedPThe primary aim of this study was to establish a baseline protocol to assess the SWV and texture of the abdominal wall musculature, as well as to identify
Table 3. Probe frequency, Valsalva and high manual probe pressure versus no pressure and 9 L4 probe in REO Subject group Control group
Active volunteers
Operator pressure / Valsalva
Probe frequency
Mean SWV § svariance (m/s)
p compared with 9 L4, no pressure
p SWV compared with active patients
No pressure No pressure No pressure Valsalva High pressure No pressure No pressure No pressure Valsalva High pressure
9 L4 6 C1 12 L4 9 L4 9 L4 9 L4 6 C1 12 L4 9 L4 9 L4
1.89 § 0.16 1.78 § 0.10 1.88 § 0.18 4.01 § 3.54 3.01 § 0.56 1.89 § 0.13 1.61 § 1.56 1.90 § 0.12 3.61 § 4.41 3.59 § 0.85
0.52 0.93 0.005 0.001 0.53 0.98 0.02 0.0001
1.0 0.7 0.9 0.6 0.1 -
SWV = shear-wave velocity; REO = right external oblique.
ARTICLE IN PRESS 6
Ultrasound in Medicine & Biology
Volume 00, Number 00, 2018
Table 4. Ultrasound texture features of active and control groups in the external oblique muscles Brightness Perc 1% Perc 10% Perc 50% Perc 90% Contra Correl SumAv DifVarnc RLNonUni GLevNonU GrVar Control group Active volunteers p value
2.14 1.24 0.00
7.99 1.49 0.01
3.87 1.30 0.01
2.27 1.27 0.00
1.72 1.21 0.00
1.20 0.83 0.02
0.96 1.12 0.03
1.01 1.00 0.01
1.15 0.83 0.01
0.78 1.40 0.01
0.68 1.15 0.01
1.14 0.84 0.02
Brightness, Perc 1%(first percentile), Perc 10%, Perc 50 % and Perc 90% represent first-order histogram. Contra (contrast), Correl (correlation), SumAv (sum of average) and, DifVarnc (difference of variance) represent grey level co-occurrence matrix. RLNonUni (run length nonuniformity) represents run length matrix. GLevNonU (gray level nonuniformity) and GrVar (gray variance) represent absolute gradient.
TagedPvarious subject- or technique-related factors that could affect the SWV measurements. In doing so, two distinct subject groups (active volunteers and control patients) were enrolled. TagedPVarious elastography studies in the extremities have shown that physiologic factors, such as age and BMI, do not correlate with muscle stiffness, although there are notable exceptions regarding BMI and the trapezius, biceps brachii and biceps femoris muscles (Aubry et al. 2013; Berko et al. 2014; Debernard et al. 2011; Dubois et al. 2015; Gennisson et al. 2005; Kuo et al. 2013). The same has not been confirmed in the case of abdominal wall musculature. Our study confirmed no significant difference between the two groups for the SWV measurements of any muscle except the RIO muscle. TagedPOne hypothesis that accounts for this difference was the varied muscle thickness between active and control groups. Some groups have shown that muscle thickness positively correlates with the measurements of the muscle stiffness (Akagi et al. 2012; Kuo et al. 2013). When muscle thickness is corrected for in our study, the REO had a higher stiffness in the control patients (p = 0.006). Additionally, the RIO was no longer significantly different (p = 0.8) between the two groups. Therefore, it is possible that the reason the SWV of RIO is different between the two groups is primarily attributable to the differences in thickness. Our texture data show that the mean brightness of the REO in the control group was significantly higher than in the active group, which we believe is reflective of higher fibrofatty infiltration (likely reflective of higher age and BMI of the control group) and could explain higher muscle-thicknesscorrected SWV measurements of the REO muscle of the control group. TagedPExtrinsic factors were those that could be affected at the time of the scan including high manual-probe
TagedPpressure by the sonographer, Valsalva maneuver by the subject and probe frequency. The Valsalva maneuver and high sonographer manual-probe pressure both led to a significant increase in SWV measurements from baseline, as expected. This is consistent with previous findings of elastography studies of the extremities and is accounted for by changes in stiffness because of muscle contraction (Gennisson et al. 2005; Kot et al. 2012). The choice of probe did not appear to have any difference in SWV measurement, which was consistent with in vitro findings in a study by Mulabecirovic et al. (2016). TagedPRecently, other groups have similarly investigated the use of muscle elastography in the evaluation of abdominal muscle stiffness (MacDonald et al. 2016; Tran et al. 2016). McDonald et al. (2016) found fair to excellent intrasession and interrater reliability in the use of supersonic shear imaging to evaluate the shear modulus at different depths in various muscle groups of the abdominal wall of healthy volunteers. Although the primary technology used by this group was different than the one used in this study, their findings show that elastography can be used with adequate reliability and repeatability in the abdominal wall. Tran et al. (2016) also investigated the use of shear wave elastography in the abdominal wall to assess the in vivo biomechanical properties of the abdominal wall. Our study seeks to corroborate the results of Tran et al. (2016) and expand on them by including key subject-related factors and probe frequency. In doing so, we hope to establish an imaging protocol that will further the understanding of the biomechanical properties of the abdominal wall, especially as it pertains to hernia-related risk factors. TagedPOur study also shows SWV measurements to be very reliable. Only 2% (3 of 152) of the valid data sets demonstrate poor IQR/M values. All three cases occurred while assessing either Valsalva maneuver or
Table 5. Ultrasound texture features between control groups and active volunteers in the internal oblique muscles
Control group Active volunteers p value
Brightness
Perc 50%
Perc 90%
Perc 99%
SumAverg
1.39 0.87 0.01
1.45 0.87 0.02
1.26 0.88 0.01
1.20 0.88 0.01
1.00 0.99 0.04
Brightness, Perc 50%, Perc 90% and Perc 99 represent first-order histogram. SumAver (sum of average) represents gray level co-occurrence matrix.
ARTICLE IN PRESS Elastography of Abdominal Wall Musculature D. A. GABRIELSEN et al.
TagedPhard manual-probe pressure. The variability in Valsalva and manual pressure can be expected, given these maneuvers are heavily subjective and dependent on either the sonographer or patient. Although reproducibility of the SWV was not assessed in these patients, the reliability of the majority of these measurements is reassuring. TagedPOur study also attempts to evaluate textural features of the abdominal wall musculature, which was performed to establish a basic protocol in healthy volunteers, as well as to determine whether there are any differences between the two groups, owing to their distinct demographic characteristics. The brightness and first-order histogram features were found to be higher in the control group. These features are related to the echointensity of the tissue components on the ultrasound image and do not take into consideration the distribution and relation of these component to each other. Hence, the higher brightness and first-order histogram observed could be explained by higher fatty infiltration in this group, which can be further explained by higher age and BMI of this group compared with active volunteers (Fig. 3). On the other hand, the observed difference in the remaining texture features between the two groups is related to the changes in the architecture of the muscles and relation of different components to each other, again likely reflective of the extent of fatty infiltration.
7
Limitations TagedPThe most pertinent limitation of this study is the small sample size. On similar lines of this small sample size, statistical significance alone will have its own limitations when it comes to reproducibility and practicality of the techniques used. Future studies with a larger sample size may be able to assess the subtle effects of technical factors on SWV measurements that might not have been apparent in our study. This will also further help to accurately identify whether age and BMI truly affect the textural differences and, if so, to what extent. TagedPAdditionally, we did not establish reproducibility of SWV measurements in this study, as outlined by Hopkins (2000). However, this was not practical in our study design, where the exams lasted nearly 45 min and patients became fatigued toward the end of exam, especially with multiple breath holding and maneuvers applied, thereby inducing a bias against the second observer. TagedPAnother possible limitation is the lack of a gold standard for comparison; however, these patients do not routinely undergo tissue biopsy. Our results are consistent insofar as the results themselves were consistent, but we could not justify performing a tissue biopsy on healthy patients for a gold-standard comparison. To establish this as a routine imaging procedure, future studies should compare this technique with other imaging modalities such as MRI or tissue samples from patients taken during surgery. TagedPFinally, none of the patients was morbidly obese, a prevalent group in patients with incisional hernia. However, BMI was certainly significantly higher in the control group compared with the active group. This could potentially lead to a higher thickness of subcutaneous fat and, in theory, more variability in high BMI patients. Our SWV data sets, however, had only a 2% failure rate. Nonetheless, further studies are needed to identify the potential impact of large amounts of subcutaneous fat on the SWV measurements of abdominal wall musculature, a known limiting factor in liver SWV measurements.
TAGEDH1CONCLUSIONSTAGEDN
Fig. 3. (a) Ultrasound and (b) CT images of a subject with no fatty infiltration of external oblique (EO) musculature (hollow arrow in a, solid arrow in b). (c) Ultrasound and (d) CT images of a different subject patient with fatty infiltration of EO musculature (hollow arrow in [a] and solid arrow in b). Note the brightness of EO in (c), one of the components of texture analysis, which at least in part reflects fatty infiltration.
TagedPSWV is a feasible and reliable technique to measure stiffness of the abdominal wall musculature and provides consistent results based on IQR/M values. Correcting for muscle thickness elucidates the significant differences in the SWV of the active and control groups in the REO. Textural features can potentially yield information regarding a change in muscle architecture, likely reflective of fatty infiltration in patients with a higher age and BMI.
ARTICLE IN PRESS 8
Ultrasound in Medicine & Biology
TAGEDH1FUNDINGTAGEDN TagedPThe work on this study was in part sponsored by Research Seed Grant, Radiological Society of North America Research & Education Foundation. TagedPThe project described was supported by the National Center for Research Resources, Grant UL1RR024134, and is now at the National Center for Advancing Translational Sciences, Grant UL1TR000003. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. TagedPSupported in part by the Institute for Translational Medicine and Therapeutics’ (ITMAT) Transdisciplinary Program in Translational Medicine and Therapeutics. TAGEDH1REFERENCESTAGEDN TagedPAkagi R, Chino K, Dohi M, Takahashi H. Relationships between muscle size and hardness of the medial gastrocnemius at different ankle joint angles in young men. Acta Radiol 2012;53:307–311. TagedPAubry S, Risson J, Kastler A, Barbier-Brion B, Siliman G, Runge M, Kastler B. Biomechanical properties of the calcaneal tendon in vivo assessed by transient shear wave elastography. Skeletal Radiol 2013;42:1143–1150. TagedPBercoff J, Tanter M, Fink M. Supersonic shear imaging: A new technique for soft tissue elasticity mapping. IEEE Trans Ultrason Ferroelectr Freq Control 2004;51:396–409. TagedPBerko NS, Fitzgerald EF, Amaral TD, Payares M, Levin TL. Ultrasound elastography in children: Establishing the normal range of muscle elasticity. Pediatr Radiol 2014;44:158–163. TagedPBittner R, Bingener-Casey J, Dietz U, Fabian M, Ferzli GS, Fortelny RH, K€ ockerling F, Kukleta J, LeBlanc K, Lomanto D, Misra MC, Morales-Conde S, Ramshaw B, Reinpold W, Rim S, Rohr M, Schrittwieser R, Simon T, Smietanski M, Stechemesser B, Timoney M, Chowbey P. Guidelines for laparoscopic treatment of ventral and incisional abdominal wall hernias (International Endohernia Society [IEHS])—Part 2. Surg Endosc 2014;28:353–379. TagedPBota S, Sporea I, Sirli R, Popescu A, Danila M, Sendroiu M. Factors that influence the correlation of acoustic radiation force impulse (ARFI) elastography with liver fibrosis. Med Ultrason 2011;13: 135–140. TagedPBotanlioglu H, Kantarci F, Kaynak G, Unal Y, Ertan S, Aydingoz O, Erginer R, Unlu MC, Mihmanli I, Babacan M. Shear wave elastography properties of vastus lateralis and vastus medialis obliquus muscles in normal subjects and female patients with patellofemoral pain syndrome. Skeletal Radiol 2013;42:659–666. TagedPBower C, Roth JS. Economics of abdominal wall reconstruction. Surg Clin North Am 2013;93:1241–1253. TagedPBurger JW, Luijendijk RW, Hop WC, Halm JA, Verdaasdonk EG, Jeekel J. Long-term follow-up of a randomized controlled trial of suture versus mesh repair of incisional hernia. Ann Surg 2004; 240:578–585. TagedPCao G, Shi P, Hu B. Liver fibrosis identification based on ultrasound images. Conf Proc IEEE Eng Med Biol Soc 2005;6:6317–6320. TagedPCarpenter EL, Lau HA, Kolodny EH, Adler RS. Skeletal muscle in healthy subjects versus those with GNE-related myopathy: Evaluation with shear-wave US—A pilot study. Radiology 2015;277: 546–554. TagedPCraig CL, Marshall AL, Sj€ostr€om M, Bauman AE, Booth ML, Ainsworth BE, Pratt M, Ekelund U, Yngve A, Sallis JF, Oja P. International physical activity questionnaire: 12-country reliability and validity. Med Sci Sports Exerc 2003;35:1381–1395. TagedPDebernard L, Robert L, Charleux F, Bensamoun SF. Characterization of muscle architecture in children and adults using magnetic
Volume 00, Number 00, 2018 rTagedP esonance elastography and ultrasound techniques. J Biomech 2011; 44:397–401. TagedPDubois G, Kheireddine W, Vergari C, Bonneau D, Thoreux P, Rouch P, Tanter M, Gennisson JL, Skalli W. Reliable protocol for shear wave elastography of lower limb muscles at rest and during passive stretching. Ultrasound Med Biol 2015;41:2284–2291. TagedPFahey B, Nelson R, Bradway D, Hsu S, Dumont D, Trahey G. In vivo visualization of abdominal malignancies with acoustic radiation force elastography. Phys Med Biol 2007;53:279. TagedPForbes S, Eskicioglu C, McLeod R, Okrainec A. Metaanalysis of randomized controlled trials comparing open and laparoscopic ventral and incisional hernia repair with mesh. Br J Surg 2009;96:851–858. TagedPGennisson J, Deffieux T, Mace E, Montaldo G, Fink M, Tanter M. Viscoelastic and anisotropic mechanical properties of in vivo muscle tissue assessed by supersonic shear imaging. Ultrasound Med Biol 2010;36:789–801. TagedPGennisson JL, Cornu C, Catheline S, Fink M, Portero P. Human muscle hardness assessment during incremental isometric contraction using transient elastography. J Biomech 2005;38:1543–1550. TagedPHolihan JL, Alawadi Z, Martindale RG, Roth JS, Wray CJ, Ko TC, Kao LS, Liang MK. Adverse events after ventral hernia repair: The vicious cycle of complications. J Am Coll Surg 2015;221:478–485. TagedPHopkins WG. Measures of reliability in sports medicine and science. Sports med 2000;30:1–15. TagedPHug F, Tucker K, Gennisson JL, Tanter M, Nordez A. Elastography for muscle biomechanics: Toward the estimation of individual muscle force. Exerc Sport Sci Rev 2015;43:125–133. TagedPIsono H, Hirata S, Yamaguchi T, Hachiya H. Analysis of fluctuation for pixel-pair distance in co-occurrence matrix applied to ultrasonic images for diagnosis of liver fibrosis. J Med Ultrason (2001) 2017;44:23–35. TagedPKot BCW, Zhang ZJ, Lee AWC, Leung VYF, Fu SN. Elastic modulus of muscle and tendon with shear wave ultrasound elastography: Variations with different technical settings. PloS one 2012;7:e44348. TagedPKuo W, Jian D, Wang T, Wang Y. Neck muscle stiffness quantified by sonoelastography is correlated with body mass index and chronic neck pain symptoms. Ultrasound Med Biol 2013;39:1356–1361. TagedPKwon DR, Park GY, Lee SU, Chung I. Spastic cerebral palsy in children: Dynamic sonoelastographic findings of medial gastrocnemius. Radiology 2012;263:794–801. TagedPMacDonald D, Wan A, McPhee M, Tucker K, Hug F. Reliability of abdominal muscle stiffness measured using elastography during trunk rehabilitation exercises. Ultrasound Med Biol 2016;42:1018–1025. TagedPMcNally EG. The development and clinical applications of musculoskeletal ultrasound. Skeletal Radiol 2011;40:1223–1231. TagedPMudge M, Hughes L. Incisional hernia: A 10 year prospective study of incidence and attitudes. Br J Surg 1985;72:70–71. TagedPMulabecirovic A, Vesterhus M, Gilja OH, Havre RF. In vitro comparison of five different elastography systems for clinical applications, using strain and shear wave technology. Ultrasound Med Biol 2016;42:2572–2588. TagedPNho RLH, Mege D, Oua€ıssi M, Sielezneff I, Sastre B. Incidence and prevention of ventral incisional hernia. J Visc Surg 2012;149:e3–e14. TagedPPalmeri ML, Wang MH, Dahl JJ, Frinkley KD, Nightingale KR. Quantifying hepatic shear modulus in vivo using acoustic radiation force. Ultrasound Med Biol 2008;34:546–558. TagedPPoulose BK, Shelton J, Phillips S, Moore D, Nealon W, Penson D, Beck W, Holzman MD. Epidemiology and cost of ventral hernia repair: Making the case for hernia research. Hernia 2012;16:179– 183. TagedPSzczypinski PM, Strzelecki M, Materka A, Klepaczko A. MaZda—A software package for image texture analysis. Comput Methods Programs Biomed 2009;94:66–76. TagedPTran D, Podwojewski F, Beillas P, Ottenio M, Voirin D, Turquier F, Mitton D. Abdominal wall muscle elasticity and abdomen local stiffness on healthy volunteers during various physiological activities. J Mech Behav Biomed Mater 2016;60:451–459.
https://doi.org/10.1016/j.ultrasmedbio.2018.05.007
Original Contribution TAGEDH1APPLICATION OF ARFI-SWV IN STIFFNESS MEASUREMENT OF THE ABDOMINAL WALL MUSCULATURE: A PILOT FEASIBILITY STUDYTAGEDN TAGEDPD1X XDAVID A. GABRIELSEND2X X* D3X XMARTIN J. CARNEYD4yX X D5X XJASON M. WEISSLERD6yX X D7X XMICHAEL A. LANNIDyX8 X D9X XJORGE HERNANDEZD10X X* D1X XLAITH R. SULTAND12zX X D13X XFABIOLA ENRIQUEZDyX14 X D15X XCHANDRA M. SEHGALD16zX X D17X XJOHN P. FISCHERD18yX X and D19X XANIL CHAUHANDzX20 X AGEDNTE
TagedP* Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA; y Division of Plastic Surgery, Department of Surgery, University of Pennsylvania, Philadelphia, PA, USA; and z Department of Radiology, University of Pennsylvania, Philadelphia, PA, USA (Received 16 May 2017; revised 9 April 2018; in final from 3 May 2018)
TagedPAbstract—The purpose of this study was to assess the feasibility of acoustic radiation force impulse shear wave velocity and textural features for characterizing abdominal wall musculature and to identify subject-related and technique-related factors that can potentially affect measurements. Median shear wave velocity measurements for the right external abdominal oblique were the same (1.89 § 0.16 m/s) for both the active group (healthy volunteers with active lifestyles) and the control group (age and body mass indexmatched volunteers from an ongoing hernia study). When corrected for thickness, the ratio of right external abdominal oblique shear wave velocity -to-muscle thickness was significantly higher in the control group than in the active volunteers (4.33 s¡1 versus 2.88 s¡1; p value 0.006). From the textural features studied for right external abdominal oblique, 8 features were found to be statistically different between the active and control groups. In conclusion, shear wave velocity is a feasible and reliable technique to evaluate the stiffness of the abdominal wall musculature. Sonographic texture features add additional characterization of abdominal wall musculature. (E-mail: Anil. [email protected]) © 2018 World Federation for Ultrasound in Medicine and Biology. All rights reserved. TagedPKey Words: ARFI-SWV, Abdominal wall musculature, Elastography, Ultrasound.
TagedPrisk factors lack the ability to assess the biomechanical properties of the abdominal wall musculature, an important parameter in predicting successful repair and disease progression (Bittner et al. 2014). Therefore, it is highly desirable to establish an imaging biomarker to fill this important gap in the understanding of the in vivo mechanical properties of the abdominal wall as well as their pathologic potential Gennisson et al. 2010; Tran et al. 2016). TagedPQuantitative imaging, using acoustic radiation force impulse shear wave velocity (ARFI-SWV), promises to be a novel and economical strategy to assess skeletal muscle properties, especially in the abdominal wall (Bercoff et al. 2004; Carpenter et al. 2015). Ultrasound elastography functions by comparing consecutive images before and after an invoked stimulus (McNally 2011). This position change is then quantified into the shear wave velocity (SWV) value, a measure of tissue stiffness (Hug et al. 2015). Elastography has
TAGEDH1INTRODUCTIONTAGEDN TagedPIncisional hernia is among the most common and serious surgical complications of abdominal surgery (Poulose et al. 2012). The incidence can be as high as 70% for high-risk (such as obese) populations (Mudge and Hughes 1985; Nho et al. 2012; Poulose et al. 2012). Despite advances, hernia repair continues to have a high failure rate, resulting in repeat surgeries, bridged repairs and nearly $3 billion lost to the U.S. healthcare system in addition to increasing morbidity for patients (Bower and Roth 2013; Burger et al. 2004; Forbes et al. 2009; Holihan et al. 2015; Poulose et al. 2012). The conventional modalities, computed tomography (CT) and magnetic resonance imaging (MRI), used to assess hernia Address correspondence to: Anil Chauhan, MD, Division of Abdominal Imaging, Department of Radiology, University of Pennsylvania, 3400 Spruce Street, Hospital of the University of Pennsylvania, University of Pennsylvania, Philadelphia, PA 19104, USA. E-mail: [email protected]
1
ARTICLE IN PRESS 2
Ultrasound in Medicine & Biology
TagedPshown success with evaluating neoplastic tissues as well as skeletal muscle biomechanical properties (Fahey et al. 2007; Hug et al. 2015; Palmeri et al. 2008). TagedPTissue texture analysis is another quantitative, yet underused, ultrasound-based imaging method. This technology primarily deals with the arrangement of various components of tissues in relation to one other. Currently, this technique is under investigation for use in the evaluation of liver fibrosis (Cao et al. 2005; Isono et al. 2017). Texture analysis similarly has the potential to quantitatively evaluate the deviation of muscle architecture from normal, such as with fibrofatty infiltration. TagedPGiven the paucity of knowledge regarding the biomechanical properties of the abdominal wall and the promise that SWV and texture analysis hold as a bridge to that gap, it is important to establish a set of standardized methods of evaluation (Botanlioglu et al. 2013; Kuo et al. 2013; Kwon et al. 2012). At least two studies have found SWV as a reliable method for the evaluation of the biomechanical properties of the abdominal wall (MacDonald et al. 2016; Tran et al. 2016). Further studies however are needed to specifically evaluate the various technical and subject-related factors that can impact the SWV measurements of abdominal wall musculature. Additionally, the texture analysis protocol while evaluating the abdominal wall musculature has not been established. Therefore, we tested the feasibility of SWV and texture analysis protocols of the abdominal wall musculature and to identify subject-related and technique-related factors that can potentially affect these modalities. TAGEDH1MATERIALS AND METHODSTAGEDN Patients TagedPA prospective cohort study was performed at the University of Pennsylvania with approval granted by the hospital’s institutional review board. A written consent was obtained from every subject. Patients were recruited from two sources: (i) from the surgery and radiology departments, 10 healthy active volunteers demonstrating high activity and a body mass index (BMI) <30, and (ii) from a prospective control group database of the study “ARFI Evaluation of Hernia Patients,” 10 patients. These patients acted as controls in our hernia study, having been age and BMI matched to hernia patients of the respective study, thereby creating a distinct group. Because demographic characteristics of these two patients differed significantly, we present the ultrasound findings in these two groups separately. TagedPPatients with a history of abdominal wall hernia or any abdominal surgeries were excluded from the study. Patients who exercised at least 48 h before the exam were also excluded. All individuals were screened using
Volume 00, Number 00, 2018
TagedPthe International Physical Activity Questionnaire (IPAQ) long-protocol guidelines to quantify their amount of activity (Craig et al. 2003). High-activity was defined according to the IPAQ guidelines that require patients to engage in “vigorous-intensity activity on at least 3 days and accumulating at least 1500 metabolic equivalent (MET)-minutes/week or 7 or more days of any combination of walking, moderate-intensity or vigorous intensity activities achieving a minimum of at least 3000 MET-minutes/week.” Image acquisition TagedPUltrasound images and SWV measurements were obtained using Siemens Acuson S3000 (Siemens Medical Inc., Mountain View, CA, USA). The linear 9 L4 probe was the primary probe used to obtain gray-scale images and SWV measurements in all the muscle groups. The right external oblique (REO) muscle was primarily evaluated, because this is the most mobilized musculature in hernia repair surgeries, and the right side is traditionally approached by a sonographer for abdominal exams. The curved 6 C1 and linear 12 L4 were additionally used for obtaining SWV measurements of the REO musculature in a similar fashion as that used for the 9 L4 probe, to also evaluate whether SWV velocities of musculature differ among various probes. All measurements were performed by one of the three sonographers, each with at least 1 y of experience in elastography and trained by a board-certified radiologist with 4 y of ultrasound experience. Gray-scale Images TagedPTwo gray-scale images in orthogonal planes of the external oblique were taken, using a 9 L4 probe with the focal zone in the longitudinally centermost portion of the muscular layer. The entire abdominal layer could be seen down to the peritoneum in each image. These images were acquired in the right flank at midpoint of the ribcage and anterior superior iliac spine, while patients were in the supine position. This position was also utilized for SWV measurement for REO and internal oblique musculature. SWV measurement TagedPMultiple SWV data sets of the REO were obtained: (i) baseline with minimal possible probe pressure (Fig. 1), (ii) after applying manual pressure with the probe (just before the point of discomfort for patients), (iii) after Valsalva maneuver, and (iv) using 12 L4 and 6 C1 probes in the same manner as that described using the 9 L4 probe. Additional data sets for the right internal oblique (RIO), the right rectus abdominis (RRM) and the left external oblique musculature (LEO) were obtained, using only the 9 L4 probe.
ARTICLE IN PRESS Elastography of Abdominal Wall Musculature D. A. GABRIELSEN et al.
3
Fig. 1. Technique of acoustic radiation force impulse shear wave velocity (ARFI-SWV) measurement. Region of interest (green square) is placed in the external oblique musculature,
TagedPTen successive SWV measurements were obtained for each data set, using point-shear wave technology. Each measurement represented a 10 £ 10 mm region of interest (ROI), which was placed at the center of the superior-most part of each muscle of interest. SWV was measured with parallel orientation of the external oblique muscle fibers to the ultrasound probe, as seen on B-mode ultrasonography, while patients were holding their breath. Parallel orientation was confirmed upon recognition of a few muscle-facial interfaces, appearing as a thin, continuous bright line. The median of 10 measurements was considered the final SWV measurement. The ratio of interquartile range (IQR) to the median value (IQR/M) was measured, and data sets with IQR/M >0.3 were considered unreliable (Bota et al. 2011). The data sets with >40% of invalid measurement were considered unreliable (i.e., when the machine records X.XX m/s as the SWV, alternatively meaning a void measurement). Texture analysis TagedPOne gray-scale image demonstrating all abdominal wall layers, with REO muscle fibers oriented parallel to the ultrasound probe, was identified for texture analysis using subcutaneous fat as an internal control (Fig. 2). Images were analyzed for a total of 21 ultrasound-based texture features using Mazda software (v. 4.6, Lodz, Poland). These features were extracted from ROIs placed in the external and internal oblique muscles as well as in the subcutaneous fat (Fig. 2). ROIs were placed by a research fellow, under direct supervision of a board-certified radiologist with 4 y of experience in quantitative ultrasound. Each ROI was as large as possible within the tissue of interest while staying away from artifacts.
Fig. 2. (a) shows a gray-scale image of abdominal wall musculature. The abdominal wall layers are labeled accordingly. SF, EO, IO and TA represents subcutaneous fat, external oblique, internal oblique and transversus abdominis, respectively. (b) Example of ROI placement on SF (blue), EO (red) and IO (green) for texture analysis.
TagedPThe texture features extracted from each image are statistical computations that indirectly represent the real texture by the nondeterministic properties that govern the distributions and relationships between the gray levels of an image. These features included parameters derived from the gray-level histogram, gray-level cooccurrence matrix, run-length matrix and absolute gradient analysis for each muscle (Szczypinski et al. 2009). In this study, we averaged distances for each parameter, calculating a single average parameter measurement used for analysis. TagedPThe first-order histogram is computed from the intensity of pixels. Features included mean brightness, variance, skewness, kurtosis and percentiles. The graylevel co-occurrence matrix is a second-order histogram, computed from intensities of pairs of pixels, where the spatial relationship of the two pixels in a pair is defined. The co-occurrence, matrix-based features are derived from a matrix to demonstrate statistics, such as angular second moment, contrast, correlation, sum of squares and various averages, variances, inverse moments and entropies. The run-length matrix holds counts of pixel runs with the specified gray-scale level and length. Features of this matrix include short-run emphasis inverse moment, long-run emphasis moment, gray-level nonuniformity, run-length nonuniformity and fraction of image
ARTICLE IN PRESS 4
Ultrasound in Medicine & Biology
Volume 00, Number 00, 2018
TagedPin runs. The absolute gradient is calculated using mean absolute gradient, variance of absolute gradient, skewness of absolute gradient and kurtosis of absolute gradient.
TagedPand the whole subject population was 19.234.2 (26.07 § 4.63). There were expected significant differences in age and BMI between the two groups (p value <0.0001).
Statistical analysis TagedPFor analytical purposes, the two subject groups (active and control) were separately analyzed in most scenarios, owing to the unique age and BMI distribution. Statistical analysis was performed to assess for significant differences between the two groups (in specific scenarios) as well as between various muscle groups. Analysis was performed using Microsoft Excel (Microsoft Corporation, Redmond, WA, USA) and the R programming language. The primary tool of comparison was a Student’s two-sample t test, assuming unequal variance. All reported p values were based on a two-tailed t test. The independent samples t test is used to test the hypothesis that the difference between the means of two groups is equal to 0 (the null hypothesis). When the p value is less than the conventional 0.05, the null hypothesis is rejected and the conclusion is that the two means do indeed differ significantly.
Validity and reliability of SWV measurements TagedPOut of a total of 160 attempted data sets, 2 data sets were invalid (high number of X.XX m/s values). These were present in the control group (post-Valsalva REO and 6 C1 probe REO measurements). Four data sets of REO using six C1 probes could not be obtained in the active volunteer group, because the system was unable to scan muscle located less than 1 cm beneath skin, such was the case in patients with minimal subcutaneous fat. Only three data sets had IQR/M more than 0.3 (two REO data sets using Valsalva maneuver and one REO data set using hard probe pressure).
TAGEDH1RESULTSTAGEDN Patients TagedPThe demographic characteristics of each group are detailed in Table 1. The age range for the control group was 5569 y (mean 61.6 § 4.86), the active volunteer group was 2333 y (mean 26.9 § 3.28) and the whole population was 2369 y (44.25 § 18.25). The BMI of the control group was 27.834.2 (mean 29.9 § 2.03), the volunteer group was 19.228 (mean 22.24 § 2.92)
Standard setup TagedPThe range of baseline resting median SWV measurements in m/s for REO was 1.502.96 (mean 1.89 § 0.16) and 1.402.51 (mean 1.89 § 0.13) for the active and control groups, respectively, with a range of 1.402.96 (mean 1.89 § 0.37) in all the patients. ARFI-SWV results TagedPThe SWV measurements were divided into intrinsic factors and extrinsic factors. Intrinsic factors included any passive, subject-related variables, such as the muscle to be analyzed and activity level. In the control group, the median SWV measurements in m/s for REO, RIO, RRM and LEO were 1.89, 1.40, 1.89 and 1.85, respectively. Similar values in m/s in the active volunteer group were 1.89, 1.80, 2.07 and 2.21, respectively.
Table 1. Subject demographic characteristics Subject group Control group
Active volunteers
Subject ID 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Sex Female Female Female Male Male Male Male Female Female Male Female Male Female Male Male Male Female Female Female Female
Age (y)
Height (in)
Weight (lb)
BMI (kg/m2)
Subcutaneous fat thickness (cm)
63 64 57 55 61 56 64 69 59 68 30 33 28 29 28 23 23 24 25 26
62.0 63.0 64.9 76.0 70.0 65.0 71.0 64.0 66.0 72.0 66.0 70.0 63.0 70.0 72.0 75.0 63.0 72.0 68.0 69.0
173 193 167 239 194 174 212 186 181 212 130 176 112 195 160 200 109 161 126 144
31.6 34.2 27.9 29.1 27.8 29.0 29.6 31.9 29.2 28.7 21.0 25.3 19.8 28.0 21.7 25.0 19.3 21.8 19.2 21.3
2.07 2.53 1.19 2.17 1.19 0.72 1.73 1.13 1.58 0.77 0.54 0.47 0.64 0.54 0.52 0.35 0.53 0.49 0.68
ARTICLE IN PRESS Elastography of Abdominal Wall Musculature D. A. GABRIELSEN et al.
5
Table 2. SWV and thickness measurements for ROI and activity levels Subject group Control group
Active volunteers
Region of interest
Range of SWVs (m/s)
Mean SWV § Variance (m/s)
Mean thickness § variance (cm)
p SWV against REO
p SWV against control
REO RIO RRM LEO REO RIO RRM LEO
1.492.96 0.901.71 1.143.65 1.242.71 1.402.51 1.432.39 1.132.29 1.502.80
1.89 § 0.16 1.40 § 0.05 1.89 § 0.52 1.85 § 0.18 1.89 § 0.13 1.80 § 0.10 2.07 § 0.23 2.21 § 0.19
0.43 § 0.22 0.81 § 0.22 0.65 § 0.98 0.98 § 0.31 -
0.004 0.97 0.81 0.57 0.35 0.09
0.99 0.005 0.5 0.07
SWV = shear-wave velocity; ROI = region of interest; REO = right external oblique; RIO = right internal oblique; RRM = right rectus abdominis; LEO = left external oblique.
TagedPInterestingly, the right IO SWV was significantly higher in the active volunteer group compared with the controls (p = 0.005), even though no statistically significant difference was noted for the REO, LEO or RRM SWV measurements (Table 2). TagedPExtrinsic factors were imposed while scanning by either the sonographer or the subject. These factors included high manual pressure by the sonographer, Valsalva maneuver by the subject and probe frequency. Compared with the standard data set of REO, significant differences were shown with Valsalva in control group (average 115.3% SWV increase, p < 0.001) as well as in the active volunteers (average 84.1% SWV increase, p < 0.001). Significant differences were also shown with high probe pressure in the control group (an average of 60.2% SWV increase, p < 0.001) and the active volunteers (an average of 95.8% SWV increase, p < 0.001). The 12 L4 and 6 C1 probes did not yield significant differences in SWV measurement compared with the standard 9 L4 probe in the REO in either group (Table 3). Muscle thickness correction of SWV measurements TagedPLiterature has suggested that muscles with higher thickness (or bulk) can lead to higher resting SWV measurements (Akagi et al. 2012; Kuo et al. 2013). To analyze this, we corrected the SWV measurements of REO and RIO muscles to the muscle thickness. The ratio of
TagedPREO SWV to muscle thickness was significantly higher in the control group than in the active volunteers (4.33 versus 2.88; p value 0.006), and there was no significant difference for these values in the RIO muscle (1.69 versus 1.76; p value 0.81). Texture analysis results TagedPFrom the features studied for external oblique muscles, 8 features were found to be statistically different between the 2 groups (Table 4). Mean brightness for the control group was higher (2.14) in compared with active patients (1.24) (p < 0.001). In addition, 1st, 10th, 50th and 90th percentiles (first histogram); contrast; sum average; difference variance (gray-level co-occurrence matrix); and gray-level variance (absolute gradient) were all significantly higher in the matched group. For features extracted from the internal oblique muscle, five texture features representing the first order histogram and gray level co-occurrence matrix were significantly higher in the control group compared with the active patients (Table 5). TAGEDH1DISCUSSIONTAGEDN TagedPThe primary aim of this study was to establish a baseline protocol to assess the SWV and texture of the abdominal wall musculature, as well as to identify
Table 3. Probe frequency, Valsalva and high manual probe pressure versus no pressure and 9 L4 probe in REO Subject group Control group
Active volunteers
Operator pressure / Valsalva
Probe frequency
Mean SWV § svariance (m/s)
p compared with 9 L4, no pressure
p SWV compared with active patients
No pressure No pressure No pressure Valsalva High pressure No pressure No pressure No pressure Valsalva High pressure
9 L4 6 C1 12 L4 9 L4 9 L4 9 L4 6 C1 12 L4 9 L4 9 L4
1.89 § 0.16 1.78 § 0.10 1.88 § 0.18 4.01 § 3.54 3.01 § 0.56 1.89 § 0.13 1.61 § 1.56 1.90 § 0.12 3.61 § 4.41 3.59 § 0.85
0.52 0.93 0.005 0.001 0.53 0.98 0.02 0.0001
1.0 0.7 0.9 0.6 0.1 -
SWV = shear-wave velocity; REO = right external oblique.
ARTICLE IN PRESS 6
Ultrasound in Medicine & Biology
Volume 00, Number 00, 2018
Table 4. Ultrasound texture features of active and control groups in the external oblique muscles Brightness Perc 1% Perc 10% Perc 50% Perc 90% Contra Correl SumAv DifVarnc RLNonUni GLevNonU GrVar Control group Active volunteers p value
2.14 1.24 0.00
7.99 1.49 0.01
3.87 1.30 0.01
2.27 1.27 0.00
1.72 1.21 0.00
1.20 0.83 0.02
0.96 1.12 0.03
1.01 1.00 0.01
1.15 0.83 0.01
0.78 1.40 0.01
0.68 1.15 0.01
1.14 0.84 0.02
Brightness, Perc 1%(first percentile), Perc 10%, Perc 50 % and Perc 90% represent first-order histogram. Contra (contrast), Correl (correlation), SumAv (sum of average) and, DifVarnc (difference of variance) represent grey level co-occurrence matrix. RLNonUni (run length nonuniformity) represents run length matrix. GLevNonU (gray level nonuniformity) and GrVar (gray variance) represent absolute gradient.
TagedPvarious subject- or technique-related factors that could affect the SWV measurements. In doing so, two distinct subject groups (active volunteers and control patients) were enrolled. TagedPVarious elastography studies in the extremities have shown that physiologic factors, such as age and BMI, do not correlate with muscle stiffness, although there are notable exceptions regarding BMI and the trapezius, biceps brachii and biceps femoris muscles (Aubry et al. 2013; Berko et al. 2014; Debernard et al. 2011; Dubois et al. 2015; Gennisson et al. 2005; Kuo et al. 2013). The same has not been confirmed in the case of abdominal wall musculature. Our study confirmed no significant difference between the two groups for the SWV measurements of any muscle except the RIO muscle. TagedPOne hypothesis that accounts for this difference was the varied muscle thickness between active and control groups. Some groups have shown that muscle thickness positively correlates with the measurements of the muscle stiffness (Akagi et al. 2012; Kuo et al. 2013). When muscle thickness is corrected for in our study, the REO had a higher stiffness in the control patients (p = 0.006). Additionally, the RIO was no longer significantly different (p = 0.8) between the two groups. Therefore, it is possible that the reason the SWV of RIO is different between the two groups is primarily attributable to the differences in thickness. Our texture data show that the mean brightness of the REO in the control group was significantly higher than in the active group, which we believe is reflective of higher fibrofatty infiltration (likely reflective of higher age and BMI of the control group) and could explain higher muscle-thicknesscorrected SWV measurements of the REO muscle of the control group. TagedPExtrinsic factors were those that could be affected at the time of the scan including high manual-probe
TagedPpressure by the sonographer, Valsalva maneuver by the subject and probe frequency. The Valsalva maneuver and high sonographer manual-probe pressure both led to a significant increase in SWV measurements from baseline, as expected. This is consistent with previous findings of elastography studies of the extremities and is accounted for by changes in stiffness because of muscle contraction (Gennisson et al. 2005; Kot et al. 2012). The choice of probe did not appear to have any difference in SWV measurement, which was consistent with in vitro findings in a study by Mulabecirovic et al. (2016). TagedPRecently, other groups have similarly investigated the use of muscle elastography in the evaluation of abdominal muscle stiffness (MacDonald et al. 2016; Tran et al. 2016). McDonald et al. (2016) found fair to excellent intrasession and interrater reliability in the use of supersonic shear imaging to evaluate the shear modulus at different depths in various muscle groups of the abdominal wall of healthy volunteers. Although the primary technology used by this group was different than the one used in this study, their findings show that elastography can be used with adequate reliability and repeatability in the abdominal wall. Tran et al. (2016) also investigated the use of shear wave elastography in the abdominal wall to assess the in vivo biomechanical properties of the abdominal wall. Our study seeks to corroborate the results of Tran et al. (2016) and expand on them by including key subject-related factors and probe frequency. In doing so, we hope to establish an imaging protocol that will further the understanding of the biomechanical properties of the abdominal wall, especially as it pertains to hernia-related risk factors. TagedPOur study also shows SWV measurements to be very reliable. Only 2% (3 of 152) of the valid data sets demonstrate poor IQR/M values. All three cases occurred while assessing either Valsalva maneuver or
Table 5. Ultrasound texture features between control groups and active volunteers in the internal oblique muscles
Control group Active volunteers p value
Brightness
Perc 50%
Perc 90%
Perc 99%
SumAverg
1.39 0.87 0.01
1.45 0.87 0.02
1.26 0.88 0.01
1.20 0.88 0.01
1.00 0.99 0.04
Brightness, Perc 50%, Perc 90% and Perc 99 represent first-order histogram. SumAver (sum of average) represents gray level co-occurrence matrix.
ARTICLE IN PRESS Elastography of Abdominal Wall Musculature D. A. GABRIELSEN et al.
TagedPhard manual-probe pressure. The variability in Valsalva and manual pressure can be expected, given these maneuvers are heavily subjective and dependent on either the sonographer or patient. Although reproducibility of the SWV was not assessed in these patients, the reliability of the majority of these measurements is reassuring. TagedPOur study also attempts to evaluate textural features of the abdominal wall musculature, which was performed to establish a basic protocol in healthy volunteers, as well as to determine whether there are any differences between the two groups, owing to their distinct demographic characteristics. The brightness and first-order histogram features were found to be higher in the control group. These features are related to the echointensity of the tissue components on the ultrasound image and do not take into consideration the distribution and relation of these component to each other. Hence, the higher brightness and first-order histogram observed could be explained by higher fatty infiltration in this group, which can be further explained by higher age and BMI of this group compared with active volunteers (Fig. 3). On the other hand, the observed difference in the remaining texture features between the two groups is related to the changes in the architecture of the muscles and relation of different components to each other, again likely reflective of the extent of fatty infiltration.
7
Limitations TagedPThe most pertinent limitation of this study is the small sample size. On similar lines of this small sample size, statistical significance alone will have its own limitations when it comes to reproducibility and practicality of the techniques used. Future studies with a larger sample size may be able to assess the subtle effects of technical factors on SWV measurements that might not have been apparent in our study. This will also further help to accurately identify whether age and BMI truly affect the textural differences and, if so, to what extent. TagedPAdditionally, we did not establish reproducibility of SWV measurements in this study, as outlined by Hopkins (2000). However, this was not practical in our study design, where the exams lasted nearly 45 min and patients became fatigued toward the end of exam, especially with multiple breath holding and maneuvers applied, thereby inducing a bias against the second observer. TagedPAnother possible limitation is the lack of a gold standard for comparison; however, these patients do not routinely undergo tissue biopsy. Our results are consistent insofar as the results themselves were consistent, but we could not justify performing a tissue biopsy on healthy patients for a gold-standard comparison. To establish this as a routine imaging procedure, future studies should compare this technique with other imaging modalities such as MRI or tissue samples from patients taken during surgery. TagedPFinally, none of the patients was morbidly obese, a prevalent group in patients with incisional hernia. However, BMI was certainly significantly higher in the control group compared with the active group. This could potentially lead to a higher thickness of subcutaneous fat and, in theory, more variability in high BMI patients. Our SWV data sets, however, had only a 2% failure rate. Nonetheless, further studies are needed to identify the potential impact of large amounts of subcutaneous fat on the SWV measurements of abdominal wall musculature, a known limiting factor in liver SWV measurements.
TAGEDH1CONCLUSIONSTAGEDN
Fig. 3. (a) Ultrasound and (b) CT images of a subject with no fatty infiltration of external oblique (EO) musculature (hollow arrow in a, solid arrow in b). (c) Ultrasound and (d) CT images of a different subject patient with fatty infiltration of EO musculature (hollow arrow in [a] and solid arrow in b). Note the brightness of EO in (c), one of the components of texture analysis, which at least in part reflects fatty infiltration.
TagedPSWV is a feasible and reliable technique to measure stiffness of the abdominal wall musculature and provides consistent results based on IQR/M values. Correcting for muscle thickness elucidates the significant differences in the SWV of the active and control groups in the REO. Textural features can potentially yield information regarding a change in muscle architecture, likely reflective of fatty infiltration in patients with a higher age and BMI.
ARTICLE IN PRESS 8
Ultrasound in Medicine & Biology
TAGEDH1FUNDINGTAGEDN TagedPThe work on this study was in part sponsored by Research Seed Grant, Radiological Society of North America Research & Education Foundation. TagedPThe project described was supported by the National Center for Research Resources, Grant UL1RR024134, and is now at the National Center for Advancing Translational Sciences, Grant UL1TR000003. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. TagedPSupported in part by the Institute for Translational Medicine and Therapeutics’ (ITMAT) Transdisciplinary Program in Translational Medicine and Therapeutics. TAGEDH1REFERENCESTAGEDN TagedPAkagi R, Chino K, Dohi M, Takahashi H. Relationships between muscle size and hardness of the medial gastrocnemius at different ankle joint angles in young men. Acta Radiol 2012;53:307–311. TagedPAubry S, Risson J, Kastler A, Barbier-Brion B, Siliman G, Runge M, Kastler B. Biomechanical properties of the calcaneal tendon in vivo assessed by transient shear wave elastography. Skeletal Radiol 2013;42:1143–1150. TagedPBercoff J, Tanter M, Fink M. Supersonic shear imaging: A new technique for soft tissue elasticity mapping. IEEE Trans Ultrason Ferroelectr Freq Control 2004;51:396–409. TagedPBerko NS, Fitzgerald EF, Amaral TD, Payares M, Levin TL. Ultrasound elastography in children: Establishing the normal range of muscle elasticity. Pediatr Radiol 2014;44:158–163. TagedPBittner R, Bingener-Casey J, Dietz U, Fabian M, Ferzli GS, Fortelny RH, K€ ockerling F, Kukleta J, LeBlanc K, Lomanto D, Misra MC, Morales-Conde S, Ramshaw B, Reinpold W, Rim S, Rohr M, Schrittwieser R, Simon T, Smietanski M, Stechemesser B, Timoney M, Chowbey P. Guidelines for laparoscopic treatment of ventral and incisional abdominal wall hernias (International Endohernia Society [IEHS])—Part 2. Surg Endosc 2014;28:353–379. TagedPBota S, Sporea I, Sirli R, Popescu A, Danila M, Sendroiu M. Factors that influence the correlation of acoustic radiation force impulse (ARFI) elastography with liver fibrosis. Med Ultrason 2011;13: 135–140. TagedPBotanlioglu H, Kantarci F, Kaynak G, Unal Y, Ertan S, Aydingoz O, Erginer R, Unlu MC, Mihmanli I, Babacan M. Shear wave elastography properties of vastus lateralis and vastus medialis obliquus muscles in normal subjects and female patients with patellofemoral pain syndrome. Skeletal Radiol 2013;42:659–666. TagedPBower C, Roth JS. Economics of abdominal wall reconstruction. Surg Clin North Am 2013;93:1241–1253. TagedPBurger JW, Luijendijk RW, Hop WC, Halm JA, Verdaasdonk EG, Jeekel J. Long-term follow-up of a randomized controlled trial of suture versus mesh repair of incisional hernia. Ann Surg 2004; 240:578–585. TagedPCao G, Shi P, Hu B. Liver fibrosis identification based on ultrasound images. Conf Proc IEEE Eng Med Biol Soc 2005;6:6317–6320. TagedPCarpenter EL, Lau HA, Kolodny EH, Adler RS. Skeletal muscle in healthy subjects versus those with GNE-related myopathy: Evaluation with shear-wave US—A pilot study. Radiology 2015;277: 546–554. TagedPCraig CL, Marshall AL, Sj€ostr€om M, Bauman AE, Booth ML, Ainsworth BE, Pratt M, Ekelund U, Yngve A, Sallis JF, Oja P. International physical activity questionnaire: 12-country reliability and validity. Med Sci Sports Exerc 2003;35:1381–1395. TagedPDebernard L, Robert L, Charleux F, Bensamoun SF. Characterization of muscle architecture in children and adults using magnetic
Volume 00, Number 00, 2018 rTagedP esonance elastography and ultrasound techniques. J Biomech 2011; 44:397–401. TagedPDubois G, Kheireddine W, Vergari C, Bonneau D, Thoreux P, Rouch P, Tanter M, Gennisson JL, Skalli W. Reliable protocol for shear wave elastography of lower limb muscles at rest and during passive stretching. Ultrasound Med Biol 2015;41:2284–2291. TagedPFahey B, Nelson R, Bradway D, Hsu S, Dumont D, Trahey G. In vivo visualization of abdominal malignancies with acoustic radiation force elastography. Phys Med Biol 2007;53:279. TagedPForbes S, Eskicioglu C, McLeod R, Okrainec A. Metaanalysis of randomized controlled trials comparing open and laparoscopic ventral and incisional hernia repair with mesh. Br J Surg 2009;96:851–858. TagedPGennisson J, Deffieux T, Mace E, Montaldo G, Fink M, Tanter M. Viscoelastic and anisotropic mechanical properties of in vivo muscle tissue assessed by supersonic shear imaging. Ultrasound Med Biol 2010;36:789–801. TagedPGennisson JL, Cornu C, Catheline S, Fink M, Portero P. Human muscle hardness assessment during incremental isometric contraction using transient elastography. J Biomech 2005;38:1543–1550. TagedPHolihan JL, Alawadi Z, Martindale RG, Roth JS, Wray CJ, Ko TC, Kao LS, Liang MK. Adverse events after ventral hernia repair: The vicious cycle of complications. J Am Coll Surg 2015;221:478–485. TagedPHopkins WG. Measures of reliability in sports medicine and science. Sports med 2000;30:1–15. TagedPHug F, Tucker K, Gennisson JL, Tanter M, Nordez A. Elastography for muscle biomechanics: Toward the estimation of individual muscle force. Exerc Sport Sci Rev 2015;43:125–133. TagedPIsono H, Hirata S, Yamaguchi T, Hachiya H. Analysis of fluctuation for pixel-pair distance in co-occurrence matrix applied to ultrasonic images for diagnosis of liver fibrosis. J Med Ultrason (2001) 2017;44:23–35. TagedPKot BCW, Zhang ZJ, Lee AWC, Leung VYF, Fu SN. Elastic modulus of muscle and tendon with shear wave ultrasound elastography: Variations with different technical settings. PloS one 2012;7:e44348. TagedPKuo W, Jian D, Wang T, Wang Y. Neck muscle stiffness quantified by sonoelastography is correlated with body mass index and chronic neck pain symptoms. Ultrasound Med Biol 2013;39:1356–1361. TagedPKwon DR, Park GY, Lee SU, Chung I. Spastic cerebral palsy in children: Dynamic sonoelastographic findings of medial gastrocnemius. Radiology 2012;263:794–801. TagedPMacDonald D, Wan A, McPhee M, Tucker K, Hug F. Reliability of abdominal muscle stiffness measured using elastography during trunk rehabilitation exercises. Ultrasound Med Biol 2016;42:1018–1025. TagedPMcNally EG. The development and clinical applications of musculoskeletal ultrasound. Skeletal Radiol 2011;40:1223–1231. TagedPMudge M, Hughes L. Incisional hernia: A 10 year prospective study of incidence and attitudes. Br J Surg 1985;72:70–71. TagedPMulabecirovic A, Vesterhus M, Gilja OH, Havre RF. In vitro comparison of five different elastography systems for clinical applications, using strain and shear wave technology. Ultrasound Med Biol 2016;42:2572–2588. TagedPNho RLH, Mege D, Oua€ıssi M, Sielezneff I, Sastre B. Incidence and prevention of ventral incisional hernia. J Visc Surg 2012;149:e3–e14. TagedPPalmeri ML, Wang MH, Dahl JJ, Frinkley KD, Nightingale KR. Quantifying hepatic shear modulus in vivo using acoustic radiation force. Ultrasound Med Biol 2008;34:546–558. TagedPPoulose BK, Shelton J, Phillips S, Moore D, Nealon W, Penson D, Beck W, Holzman MD. Epidemiology and cost of ventral hernia repair: Making the case for hernia research. Hernia 2012;16:179– 183. TagedPSzczypinski PM, Strzelecki M, Materka A, Klepaczko A. MaZda—A software package for image texture analysis. Comput Methods Programs Biomed 2009;94:66–76. TagedPTran D, Podwojewski F, Beillas P, Ottenio M, Voirin D, Turquier F, Mitton D. Abdominal wall muscle elasticity and abdomen local stiffness on healthy volunteers during various physiological activities. J Mech Behav Biomed Mater 2016;60:451–459.