Abstract
Background
Most image-analysis software programs were not designed specifically for ultrasound-based measurements, resulting in irrelevant general-purpose features that force users to navigate through multiple menus and making the analysis time-consuming and increasing the likelihood of error. The aim of this study was to present and validate the semi-automated MuscleExpert software against the widely used ImageJ by comparing muscle thickness (MT), cross-sectional area (CSA), and muscle quality (MQ-measured as echo intensity).
Methods
Ten volunteers (five males; age: 44.9 ± 23.5 years) participated in this cross-sectional study. Ultrasound images of the tibialis anterior and medial gastrocnemius muscles were acquired using 2D B-mode ultrasonography and analyzed in both software programs. Equivalence of means was evaluated via the two-one-sided tests (TOST) procedure; the reliability between means was quantified by an intraclass correlation coefficient (ICC); the precision was expressed as the coefficient of variation (CV) and standard error of measurement (SEM); and agreement was assessed using Bland–Altman plots.
Results
The TOST for MT, CSA, and MQ were within their respective equivalence margins. The ICC was excellent for all measures (MT: ICC = 0.999; CSA: ICC = 0.999; MQ: ICC = 1.000). Precision was similarly high, with CVs of 0.49% (MT), 0.54% (CSA), and 0.33% (MQ), demonstrating minimal variability between the two methods. SEM values were low across all outcomes, corresponding to 0.49% (MT), 0.53% (CSA), and 0.23% (MQ). Finally, the Bland–Altman analysis demonstrated minimal systematic differences and no proportional bias between ImageJ and MuscleExpert for all metrics. The present findings indicate that MuscleExpert and ImageJ produce nearly identical results for MT, CSA, and MQ. These findings support the integration of MuscleExpert into clinical and research workflows, offering a more efficient solution for muscle assessment without compromising measurement integrity.
Keywords: Muscle size, Muscle quality, Software, Ultrasound imaging, Muscle thickness, Cross-sectional area, MuscleExpert software
Introduction
Two-dimensional (2D) ultrasound (US) imaging is widely used for clinical and research purposes across the health sciences field (Whittaker & Stokes, 2011). Clinicians often record images of organs to assess structural size and tissue quality (Fang & Sidhu, 2020; Imasawa et al., 2012), while researchers and clinicians use B-mode ultrasound to capture images of specific muscles and quantify both muscle size (Lanza et al., 2025) and muscle quality (e.g., echo intensity) (Stock & Thompson, 2021). The accuracy and consistency of ultrasound-derived measurements can vary depending on the analysis software and user workflow (Cain & Haque, 2008). To ensure reliable and reproducible results, as well as decreased time for data analysis, there is a clear need for software specifically designed for ultrasound image analysis.
Most image-analysis software, such as Horos, OsiriX, and ImageJ, were not specifically designed for ultrasound-based muscle morphology measurements. Consequently, they contain numerous general-purpose features that are irrelevant to this application and force users to navigate through multiple menus and dialogs. In many cases, the overwhelming amount of tools can prolong data analysis and may lead users to select the wrong functions, resulting in measurement errors (Lateef et al., 2024). Thus, this may be the reason for researchers to develop their own custom image-analysis software and/or scripts (Mul et al., 2018; Stein et al., 2023). However, these tools are rarely shared with the wider community, and there is no assurance that the measurements they produce will agree with those generated by established, widely used programs. Consequently, it is possible that measurements derived from such tools differ meaningfully from those obtained using standard, well-documented platforms—introducing potential bias or error into published findings.
Muscle size and quality are types of measurements that are constantly performed by different research groups using ultrasound (Lanza et al., 2025; Franchi et al., 2018; Abe et al., 2000; Chang et al., 2019; Harris-Love et al., 2018; Rock et al., 2021). The image analysis to extract the values from these measurements can be time-consuming. For reference, a typical analysis on ImageJ software to extract both measurements (muscle size and quality) proceeds through the following steps: (1) loading the raw ultrasound image; (2) opening the calibration tool; (3) drawing a reference line over a region of known length (i.e., the probe’s field of view); (4) entering the corresponding distance to set the image scale; (5) opening brightness/contrast settings (not always required); (6) adjusting brightness and/or contrast (not always required); (7) applying image filters to enhance visualization of superficial and deep aponeuroses (not always required); (8) selecting the muscle-thickness measurement tool; (9) tracing a line between the superficial and deep aponeuroses of the muscle; (10) recording the value; (11) selecting the polygon tool to draw the region of interest (ROI) for echo-intensity analysis; (12) tracing the ROI; (13) extracting the mean echo intensity value; and (14) saving or exporting the measurement data. Switching between these tools for each task not only prolongs the process but also increases the risk of inconsistency and user error. Furthermore, such multi-step workflows demand considerable training, and the reliance on user judgment at multiple decision points has been associated with modest inter-rater reliability when different researchers analyze the same images (Carr et al., 2021).
Recently, the MuscleExpert software was released freely on Zenodo (Bahia Lanza, 2025). This semi-automatic Python-based program was designed to reduce analysis time by minimizing manual steps and automatically selecting the appropriate tools based on established muscle-morphology protocols across different research groups (Franchi et al., 2018; Lanza et al., 2022; Larsen et al., 2025; Saito et al., 2024; Nishihara et al., 2024). Unlike ImageJ, MuscleExpert streamlines analysis into three steps: (1) loading the raw US image; (2) cropping the image of the muscle; (3) drawing the ROI. By collapsing more than a dozen manual operations into just these three actions, MuscleExpert might reduce analysis time by approximately 80% (see File S1) and may substantially reduce the risk of tool-selection errors, ensuring faster, and potentially more consistent measurements across users. However, the reproducibility and reliability of MuscleExpert have not been demonstrated, nor has its agreement with other well-known software.
Therefore, the aim of this study is to assess the agreement and reliability of the semi-automated MuscleExpert software with ImageJ, a widely used, freely available image-analysis platform extensively applied in ultrasound-based muscle research, by comparing muscle thickness (MT), cross-sectional area (CSA), and muscle quality (MQ-measured as echo intensity) metrics to assess agreement and reliability between the two approaches. It was hypothesized that the MuscleExpert software would demonstrate excellent agreement and reliability with ImageJ, thereby supporting its adoption as a tool for ultrasound image analysis.
Materials and Methods
Subjects
This cross-sectional study was approved by the University of Maryland Institutional Review Board (HP-00109908). A convenience sample of 10 participants (five males, five females; 18 to 85 years old; age: 44.9 ± 23.5 years; height: 1.65 ± 0.8 m; body mass: 67.3 ± 16.4 kg) provided written informed consent before their participation. All participants, who had no history of neurological or muscular disorders, nor serious injuries or surgery of the lower extremity, volunteered to participate in the study. Portions of this text were previously published as part of a preprint (Lanza, 2025).
Measurements and procedures
Participants remained seated (knees at 90° of flexion) and rested for 10 min before ultrasound measurements. Ultrasound recordings and analysis were performed for tibialis anterior (TA) and medial gastrocnemius (MG) muscles. The anterior and posterior regions of the dominant leg were marked to identify the reference points for the ultrasound image acquisition. The dominant limb was defined by the question: which foot would you use to kick a ball (Lanza et al., 2022). Skin marks were made at 1/3 on the line between the tip of the fibula and the tip of the medial malleolus for the TA muscle, while placed on the most prominent bulge of the muscle for MG. Two ultrasound images per site were obtained using B-mode ultrasonography (Aixplorer, Supersonic Imagine, Aix-en-Provence, France) with a SuperLinear 5–18 MHz frequency probe, gain of 62 dB, a depth of 6.0 cm. An example of US images can be seen in Fig. 1. All images were acquired by the same evaluator (Z.L.), with 10 years of experience in musculoskeletal ultrasound imaging. Probe pressure was kept minimal throughout image acquisition and was continuously monitored by verifying the absence of visible skin or tissue compression within the image.
Figure 1. Example of a cross-sectional area measurement from the medial gastrocnemius from one participant on (A) ImageJ software and (B) MuscleExpert software.
Data analysis
For each muscle, twenty images were generated, yielding a total of forty images for analysis. Image analysis was performed using two software programs: ImageJ software, version 1.53a (National Institutes of Health, Bethesda, MD, USA) and MuscleExpert software, version 1.0. The image analysis comprised two stages. In the first stage, each image was analyzed twice in each program (ImageJ and MuscleExpert), and the mean of those two readings was calculated for each program. In the second stage of analysis, the same procedure was repeated. The grand mean, computed as the average of the two stages, served as the final measurement for all subsequent analyses. Repeating measurements within and between sessions and then averaging them has been shown to reduce random error and to allow formal assessment of intra-rater consistency using intraclass correlation coefficients (Weir, 2005; Atkinson & Nevill, 1998). Moreover, this design enables the detection of any systematic bias (e.g., learning or fatigue effects) between stages, yielding more precise and trustworthy estimates.
Ultrasound images were exported from the imaging device in JPEG format and imported into both software programs. In ImageJ, MT was measured as the straight-line distance between the superficial and deep aponeuroses on each image, while muscle quality (mean echo intensity value) was determined by outlining the ROI in the TA and MG muscles with the polygon tool (Lateef et al., 2024) and recording the average echo intensity value within that area. In MuscleExpert, the polygon tool was used to delineate muscle ROI, but both MT and mean echo intensity were calculated automatically from the drawn region. In both software programs, MT was measured at 50% of the image size as previously (Lanza et al., 2025). In MuscleExpert, the image processing steps followed the three-step sequence described in the Introduction: loading the image, cropping the muscle region, and drawing the ROI, from which all variables were extracted automatically. No additional grayscale normalization was applied, nor was it required: both software programs analyzed the exact same image files, meaning the underlying pixel intensity values were identical across programs. Grayscale normalization is only necessary when comparing measurements derived from images acquired under different conditions; since both programs processed the same image, pixel values were inherently equivalent.
Statistical analysis
Statistical analyses were conducted in Python using the Visual Studio Code IDE (Microsoft Corporation, version 1.90.2), and an alpha level of 0.05 was applied to determine statistical significance. Data from the two muscles were pooled because our goal was to compare software performance rather than muscle-specific differences. All variables were assessed for normality using the Shapiro–Wilk test. Equivalence of means was then evaluated via the two-one-sided tests (TOST) procedure (Lakens, 2017) with a priori equivalence margins (Δ) of ±0.25 cm for thickness, ±0.65 cm2 for area, and ±9 arbitrary units (au.). Two one-sided paired t-tests were performed and equivalence concluded if both p-values were <0.05 and the 90% confidence interval for the mean difference lay entirely within [−Δ, +Δ] (Lakens, 2017). Margins for MT (Lanza et al., 2025; McCreesh & Egan, 2011; Yuguchi et al., 2020; Wang, Hu & Tian, 2018; Arts et al., 2010) and MQ (as echo intensity) (Yuguchi et al., 2020; Arts et al., 2010; Yao et al., 2023; Caresio et al., 2015) in both muscles were calculated from prior studies. For each variable, the range of typical between-measurement differences reported across these studies was identified for each muscle, and the most conservative threshold, that is, the smallest margin, not exceeding 8% of the mean, was selected to ensure the most rigorous criterion for equivalence. Because measuring CSA on longitudinal US images is not standard practice, a conservative 8% margin was applied based on our own ImageJ reference data, consistent with the approach used for the other variables. Absolute agreement was quantified by an intraclass correlation coefficient (ICC(3,1)), interpreted as weak (<0.4), moderate (0.4–0.59), good (0.6–0.74), and/or excellent (0.75–1.0) (Cicchetti, 1994). Precision was expressed as the coefficient of variation (CV = [within-subject SD/overall mean] ×100%). The standard error of measurement (SEM) was computed using the classical test theory formula: SEM = SD ×√(1–ICC), where SD represents the pooled standard deviation of the two measurement methods and ICC refers to the absolute agreement intraclass correlation coefficient (ICC(3,1)). SEM was further expressed as a percentage of the overall mean to provide a scale-free index of absolute error. Agreement across the measurement range was further explored using Bland–Altman plots, reporting mean bias and 95% limits of agreement (mean difference ±1.96 × SD).
Results
The TOST for MT yielded a mean difference of 0.005 cm (90% CI [0.0004–0.010] cm; plower < 0.0001, pupper < 0.0001), which lies entirely within the ±0.25 cm margin, Fig. 2A. CSA differences averaged 0.031 cm2 (90% CI [0.007–0.054] cm2; plower < 0.0001, pupper < 0.0001; Fig. 2B) and MQ differences averaged 1.27 au. (90% CI [1.143–1.397] au.; plower < 0.0001, pupper < 0.0001; Fig. 2C), both comfortably within their respective equivalence margins. Absolute agreement was excellent for all measures (MT: ICC = 0.999; CSA: ICC = 0.999; MQ: ICC = 1.000), Table 1. Precision was similarly high, with CVs of 0.49% for MT, 0.54% for CSA, and 0.33% for MQ, demonstrating minimal variability between the two methods. SEM values were low across all outcomes, corresponding to 0.008 cm (0.49 %) for MT, 0.04 cm2 (0.53 %) for CSA, and 0.36 au. (0.23 %) for MQ, Table 1. Bland–Altman analysis demonstrated minimal systematic differences and no proportional bias between ImageJ and MuscleExpert for all three metrics. For MT bias was 0.0052 cm and 95% limits of −0.0296 to 0.0401 cm (slope = 0.0084, p = 0.2673), Fig. 3A. For CSA the bias was 0.0309 cm2 (95% limits −0.1396 to 0.2014 cm2) and a non-significant slope of 0.0154 (p = 0.0610), Fig. 3B. Lastly, MQ bias was 1.2701 au., with limits of 0.3379 to 2.2023 au. and slope −0.0037 (p = 0.4654), Fig. 3C.
Figure 2. Mean values of each individual measure on ImageJ and MuscleExpert software programs and the mean-difference results for (A) muscle thickness, (B) cross-sectional area, and (C) muscle quality.
Solid black lines denote the pre-specified lower (–Δ ) and upper (+Δ ) equivalence margins used in the two-one-sided tests (TOST). Dashed lines indicate the lower and upper bounds of the 90% confidence interval around the observed mean difference. Equivalence is achieved when the entire dashed interval lies within the solid black margins.
Table 1. Intraclass correlation coefficients (ICC), coefficients of variation (CV), and standard error of measurement (SEM) for MuscleExpert versus ImageJ software.
| Variable | ICC (95% CI) | CV (%) | SEM (95% CI) | SEM (%) |
|---|---|---|---|---|
| Muscle thickness | 0.999 (0.998–1.000) | 0.49 | 0.008 (0.0072–0.0103) cm | 0.49% |
| Cross-sectional area | 0.999 (0.999–1.000) | 0.54 | 0.04 (0.03–0.05) cm2 | 0.53% |
| Muscle quality | 1.000 (0.999–1.000) | 0.33 | 0.23 (0.20–0.26) au. | 0.23% |
Figure 3. Bland-Altman plots for the agreement between MuscleExpert and ImageJ for muscle thickness (A), cross-sectional area (B), and muscle quality (C), Solid lines represent the upper and lower limits of agreement (±1.96 SD).
The dotted line represents the mean bias.
Discussion
Since custom image-analysis pipelines are often neither published nor benchmarked against established standards, there is no guarantee their results align with those produced by validated, widely adopted software. Thus, the first aim of this study was to assess the agreement and reliability of the semi-automated MuscleExpert software against ImageJ by comparing MT, CSA, and MQ measurements for agreement and reliability. The present results support our hypothesis that the MuscleExpert software demonstrates excellent agreement and reliability with ImageJ. It is important to note that this study establishes software-to-software agreement rather than measurement validity; validity would require comparison against a criterion standard independent of both programs, such as direct anatomical measurement. In summary, the MuscleExpert software offers a reliable, efficient alternative to ImageJ for quantitative ultrasound assessment of muscle morphology.
The test of equivalence confirmed that the differences in MT, CSA, and MQ between the two methods fell entirely within the predefined equivalence margins, supporting the practical equivalence of the two methods. Importantly, highly conservative margins were adopted for all variables. This suggests that MuscleExpert can be considered a comparable alternative to ImageJ for these measurements. In addition to that, the exceptionally high intraclass correlation coefficients (ICCs ≥ 0.999) and low coefficients of variation (CVs < 1%) further support the precision and consistency of MuscleExpert, indicating minimal variability and excellent reproducibility across all three muscle morphology measurements. SEMs were also consistently low across all outcomes, with values below 0.05% of the pooled mean, further demonstrating the high accuracy of MuscleExpert compared to ImageJ. These minimal SEM values reinforce the notion that the software introduces minimal measurement error. Bland–Altman analyses reinforced these findings by showing minimal systematic bias and no evidence of proportional bias between the two methods. The narrow limits of agreement and non-significant regression slopes across all metrics suggest that MuscleExpert performs comparably to ImageJ across the full range of values. Notably, the MQ metric, often more susceptible to variability due to echo intensity interpretation (Stock & Thompson, 2021), showed the highest precision (CV = 0.33%) and near-perfect agreement (ICC = 1.000), noting that an ICC of 1.000 is uncommon in biological measurements. This likely reflects both the limited between-subject variability in this convenience sample and the within-method variance reduction introduced by the repeated-measures averaging design. Thus, these results should be interpreted with appropriate caution, highlighting the robustness of MuscleExpert’s echo intensity analysis. These findings collectively support the use of MuscleExpert software as a reliable and efficient tool for muscle assessment, offering potential advantages in clinical and research settings where automation and scalability are increasingly important.
Each program presents both advantages and limitations. MuscleExpert is specifically engineered for ultrasound imaging in health-science disciplines (e.g., exercise physiology, physical therapy), offering a minimalist interface designed to minimize tool-selection errors and reduce configuration time. In contrast, ImageJ is an open-source, extensible environment with comprehensive filters, plugins, scripting support (Java, Python), batch-processing pipelines, and detailed metadata extraction, though its extensive feature set may require a longer learning period. Consequently, MuscleExpert may be better suited for rapid, standardized analyses, whereas ImageJ may better serve users requiring deep customization and automation. From a practical standpoint, the present findings suggest that, under conditions similar to those studied, MuscleExpert may produce measurements comparable to those obtained with ImageJ, potentially allowing data from both platforms to be used interchangeably. For researchers currently using ImageJ, no immediate change in practice may be warranted; however, for groups initiating new projects or seeking to reduce user-dependent variability, MuscleExpert may offer a more accessible and time-efficient alternative. Its streamlined workflow may also support greater standardization across research groups, though broader validation across diverse samples, operators, and imaging conditions is needed before definitive recommendations can be made.
Limitations and strengths of this study should be acknowledged. First, the sample size was small (n = 10) and comprised a convenience sample with a wide age range (18–85 years), which limits the statistical power of the findings and may reduce their generalizability to more homogeneous clinical or research populations. Second, all image analyses were performed by a single operator; therefore, inter-rater reliability was not assessed, and the extent to which these findings generalize to different analysts remains unknown. A notable strength of this study is the inclusion of two anatomically and morphologically distinct muscles, the tibialis anterior and the medial gastrocnemius, analyzed as a pooled dataset. This design reflects an important conceptual consideration: the present study aimed to evaluate whether the same image analysis method, applied across two different software platforms, produces equivalent results, rather than to validate either software for a specific muscle group. The analytical steps used to extract MT, CSA, and echo intensity from an ultrasound image are methodologically consistent regardless of the anatomical structure being examined. Combining measurements from two distinct muscles therefore offers a more comprehensive assessment of equivalence at the methodological level. The consistent agreement observed across both muscles supports the conclusion that the equivalence between MuscleExpert and ImageJ is due to the analytical method, independent of the specific muscle analyzed. In addition, in a partial analysis of image analysis time, MuscleExpert appeared to require considerably less time per image compared to ImageJ, suggesting a potential time advantage that, if confirmed in larger samples, could translate to meaningful time savings in research and clinical workflows (see File S1). Finally, the study was conducted under standardized laboratory conditions using a single imaging protocol and ultrasound device; performance of MuscleExpert under different probe frequencies, image depths, or acquisition settings has not been evaluated and warrants future investigation.
In conclusion, the findings indicate that MuscleExpert and ImageJ produce nearly identical results for MT, CSA, and MQ. The equivalence testing, high ICC, low CVs and SEM values, and Bland–Altman analyses collectively demonstrate that MuscleExpert provides measurements that are statistically equivalent and practically comparable to those obtained with ImageJ. These findings support the integration of MuscleExpert into clinical and research workflows, offering a more efficient solution for muscle assessment without compromising measurement integrity.
Supplemental Information
Compares the time required to measure muscle thickness, cross-sectional area, and echo intensity in 20 ultrasound images using MuscleExpert and ImageJ, including statistical results and estimated time savings for the full study dataset
Acknowledgments
The authors thank the participants for taking part in the study. The authors also thank Dr. Hugo C. Martins-Costa, Dr. Rodrigo R. Diniz, Dr. Gustavo F. Pedrosa, Dr. Lucas T. Lacerda, Dr. Ben Friedman, and Ms. Libyna T. C. Martins for their valuable contributions to testing and providing feedback on the MuscleExpert software.
Funding Statement
The authors received no funding for this work.
Additional Information and Declarations
Competing Interests
The authors declare there are no competing interests.
Author Contributions
Zongpan Li performed the experiments, prepared figures and/or tables, authored or reviewed drafts of the article, and approved the final draft.
Li-Qun Zhang analyzed the data, prepared figures and/or tables, authored or reviewed drafts of the article, and approved the final draft.
Marcel Bahia Lanza conceived and designed the experiments, performed the experiments, analyzed the data, prepared figures and/or tables, authored or reviewed drafts of the article, developed MuscleExpert Software, and approved the final draft.
Human Ethics
The following information was supplied relating to ethical approvals (i.e., approving body and any reference numbers):
UMB Institutional Review Board.
Data Availability
The following information was supplied regarding data availability:
The data is available in the Supplemental Files.
References
- Abe et al. (2000).Abe T, De Hoyos DV, Pollock ML, Garzarella L. Time course for strength and muscle thickness changes following upper and lower body resistance training in men and women. European Journal of Applied Physiology. 2000;81(3):174–180. doi: 10.1007/s004210050027. [DOI] [PubMed] [Google Scholar]
- Arts et al. (2010).Arts IMP, Pillen S, Schelhaas HJ, Overeem S, Zwarts MJ. Normal values for quantitative muscle ultrasonography in adults. Muscle Nerve. 2010;41(1):32–41. doi: 10.1002/mus.21458. [DOI] [PubMed] [Google Scholar]
- Atkinson & Nevill (1998).Atkinson G, Nevill AM. Statistical methods for assessing measurement error (reliability) in variables relevant to sports medicine. Sports Medicine. 1998;26(4):217–238. doi: 10.2165/00007256-199826040-00002. [DOI] [PubMed] [Google Scholar]
- Bahia Lanza (2025).Bahia Lanza M. MuscleExpert Software. https://zenodo.org/records/16919609. [21 August 2025];2025
- Cain & Haque (2008).Cain C, Haque S. Organizational workflow and its impact on work quality. In: Hughes RG, editor. Patient safety and quality: an evidence-based handbook for nurses. Advances in patient safety. Rockville, Maryland: Agency for Healthcare Research and Quality (US); 2008. [4 July 2025]. . http://www.ncbi.nlm.nih.gov/books/NBK2638/ [PubMed] [Google Scholar]
- Caresio et al. (2015).Caresio C, Molinari F, Emanuel G, Minetto MA. Muscle echo intensity: reliability and conditioning factors. Clinical Physiology and Functional Imaging. 2015;35(5):393–403. doi: 10.1111/cpf.12175. [DOI] [PubMed] [Google Scholar]
- Carr et al. (2021).Carr JC, Gerstner GR, Voskuil CC, Harden JE, Dunnick D, Badillo KM, Pagan JI, Harmon KK, Girts RM, Beausejour JP, Stock MS. The influence of sonographer experience on skeletal muscle image acquisition and analysis. Journal of Functional Morphology and Kinesiology. 2021;6(4):91. doi: 10.3390/jfmk6040091. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Chang et al. (2019).Chang KV, Yang KC, Wu WT, Huang KC, Han DS. Association between metabolic syndrome and limb muscle quantity and quality in older adults: a pilot ultrasound study. Diabetes, Metabolic Syndrome and Obesity: Targets and Therapy. 2019;12:1821–1830. doi: 10.2147/DMSO.S219649. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Cicchetti (1994).Cicchetti DV. Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychological Assessment. 1994;6(4):284–290. doi: 10.1037/1040-3590.6.4.284. [DOI] [Google Scholar]
- Fang & Sidhu (2020).Fang C, Sidhu PS. Ultrasound-based liver elastography: current results and future perspectives. Abdominal Radiology. 2020;45(11):3463–3472. doi: 10.1007/s00261-020-02717-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Franchi et al. (2018).Franchi MV, Longo S, Mallinson J, Quinlan JI, Taylor T, Greenhaff PL, Narici MV. Muscle thickness correlates to muscle cross-sectional area in the assessment of strength training-induced hypertrophy. Scandinavian Journal of Medicine & Science in Sports. 2018;28(3):846–853. doi: 10.1111/sms.12961. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Harris-Love et al. (2018).Harris-Love M, Benson K, Leasure E, Adams B, McIntosh V. The influence of upper and lower extremity strength on performance-based sarcopenia assessment tests. Journal of Functional Morphology and Kinesiology. 2018;3(4):53. doi: 10.3390/jfmk3040053. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Imasawa et al. (2012).Imasawa T, Nakazato T, Ikehira H, Fujikawa H, Nakajima R, Ito T, Ando Y, Yoshimura M, Nakayama M, Yahata K, Sasaki O, Yaomura T, Katafuchi R, Yamamura T, Kawaguchi T, Nishimura M, Kitamura H, Kenmochi T, Shimatsu A. Predicting the outcome of chronic kidney disease by the estimated nephron number: the rationale and design of PRONEP, a prospective, multicenter, observational cohort study. BMC Nephrology. 2012;13(1):11. doi: 10.1186/1471-2369-13-11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lakens (2017).Lakens D. Equivalence tests: a practical primer for t tests, correlations, and meta-analyses. Social Psychological and Personality Science. 2017;8(4):355–362. doi: 10.1177/1948550617697177. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lanza (2025).Lanza MB. Muscle ultrasound image analysis with a semi-automated software (MuscleExpert) yields results comparable to ImageJ: muscle thickness, muscle area, and muscle quality. Health Informatics. 2025:1–18. doi: 10.1101/2025.09.01.25334864. Preprint. [DOI] [Google Scholar]
- Lanza et al. (2025).Lanza MB, Baghi R, Frakes N, Lateef S, Martins LTC, Zhang LQ, Gray VL. Differences in muscle morphology assessed by ultrasound at various muscle regions and their impact on voluntary and involuntary force production. Scientific Reports. 2025;15(1):9247. doi: 10.1038/s41598-025-94158-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lanza et al. (2022).Lanza MB, Rock K, Marchese V, Gray VL, Addison O. Ultrasound measures of muscle thickness and subcutaneous tissue from the hip abductors: inter- and intra-rater reliability. Musculoskeletal Science & Practice. 2022;62:102612. doi: 10.1016/j.msksp.2022.102612. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Larsen et al. (2025).Larsen S, Sandvik Kristiansen B, Swinton PA, Wolf M, Bao Fredriksen A, Nygaard Falch H, Van Den Tillaar R, Østerås Sandberg N. The effects of hip flexion angle on quadriceps femoris muscle hypertrophy in the leg extension exercise. Journal of Sports Sciences. 2025;43(2):210–221. doi: 10.1080/02640414.2024.2444713. [DOI] [PubMed] [Google Scholar]
- Lateef et al. (2024).Lateef S, Addison O, Zhang LQ, Gray V, Lanza MB. Exploring discrepancies in muscle analysis with ImageJ: understanding the impact of tool selection on echo intensity and muscle area measurements. Journal of Ultrasound. 2024;27(4):973–977. doi: 10.1007/s40477-024-00934-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- McCreesh & Egan (2011).McCreesh K, Egan S. Ultrasound measurement of the size of the anterior tibial muscle group: the effect of exercise and leg dominance. Sports Medicine, Arthroscopy, Rehabilitation, Therapy & Technology. 2011;3(1):18. doi: 10.1186/1758-2555-3-18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Mul et al. (2018).Mul K, Horlings CGC, Vincenten SCC, Voermans NC, Van Engelen BGM, Van Alfen N. Quantitative muscle MRI and ultrasound for facioscapulohumeral muscular dystrophy: complementary imaging biomarkers. Journal of Neurology. 2018;265(11):2646–2655. doi: 10.1007/s00415-018-9037-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Nishihara et al. (2024).Nishihara K, Kawai H, Ejiri M, Imamura K, Obuchi S. Associations between muscle thickness, motor function, and echo intensity in community-dwelling older japanese men and women. Cureus. 2024;6(12):e76651. doi: 10.7759/cureus.76651. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Rock et al. (2021).Rock K, Nelson C, Addison O, Marchese V. Assessing the reliability of handheld dynamometry and ultrasonography to measure quadriceps strength and muscle thickness in children, adolescents, and young adults. Physical and Occupational Therapy in Pediatrics. 2021;41:540–554. doi: 10.1080/01942638.2021.1881200. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Saito et al. (2024).Saito H, Fujimoto Y, Matsue Y, Yoshioka K, Maekawa E, Kamiya K, Toki M, Iwata K, Saito K, Murata A, Hayashida A, Ako J, Kitai T, Kagiyama N. Ultrasound-measured quadriceps muscle thickness and mortality in older patients with heart failure. Canadian Journal of Cardiology. 2024;40(12):2555–2564. doi: 10.1016/j.cjca.2024.09.007. [DOI] [PubMed] [Google Scholar]
- Stein et al. (2023).Stein T, Taron J, Verloh N, Doppler M, Rau A, Hagar MT, Faby S, Baltas D, Westermann D, Ayx I, Schönberg SO, Nikolaou K, Schlett CL, Bamberg F, Weiss J. Photon-counting computed tomography of coronary and peripheral artery stents: a phantom study. Scientific Reports. 2023;13(1):14806. doi: 10.1038/s41598-023-41854-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Stock & Thompson (2021).Stock MS, Thompson BJ. Echo intensity as an indicator of skeletal muscle quality: applications, methodology, and future directions. European Journal of Applied Physiology. 2021;121(2):369–380. doi: 10.1007/s00421-020-04556-6. [DOI] [PubMed] [Google Scholar]
- Wang, Hu & Tian (2018).Wang J, Hu Y, Tian G. Ultrasound measurements of gastrocnemius muscle thickness in older people with sarcopenia. Clinical Interventions in Aging. 2018;13:2193–2199. doi: 10.2147/CIA.S179445. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Weir (2005).Weir JP. Quantifying test-retest reliability using the intraclass correlation coefficient and the SEM. The Journal of Strength & Conditioning Research. 2005;19(1):231–231. doi: 10.1519/15184.1. [DOI] [PubMed] [Google Scholar]
- Whittaker & Stokes (2011).Whittaker JL, Stokes M. Ultrasound imaging and muscle function. Journal of Orthopaedic & Sports Physical Therapy. 2011;41(8):572–580. doi: 10.2519/jospt.2011.3682. [DOI] [PubMed] [Google Scholar]
- Yao et al. (2023).Yao Q, Zhang Y, Wu J, Shu H, Ye X, Li A. Quantitative evaluation of gastrocnemius medialis mass in patients with chronic heart failure by gray-scale ultrasound and shear wave elastography. Frontiers in Cardiovascular Medicine. 2023;10:1132519. doi: 10.3389/fcvm.2023.1132519. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yuguchi et al. (2020).Yuguchi S, Asahi R, Kamo T, Azami M, Ogihara H. Gastrocnemius thickness by ultrasonography indicates the low skeletal muscle mass in Japanese elderly people. Archives of Gerontology and Geriatrics. 2020;90:104093. doi: 10.1016/j.archger.2020.104093. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Compares the time required to measure muscle thickness, cross-sectional area, and echo intensity in 20 ultrasound images using MuscleExpert and ImageJ, including statistical results and estimated time savings for the full study dataset
Data Availability Statement
The following information was supplied regarding data availability:
The data is available in the Supplemental Files.