Ultrasound elastography (EUS) is a noninvasive imaging technique used to assess tissue elasticity or stiffness. It is primarily employed to characterize soft tissues and has demonstrated considerable potential in diagnosing various conditions, including tumors and musculotendinous diseases. EUS consists of two principal types: strain elastography (SE) and shear wave elastography (SWE). SE utilizes compressive forces to measure tissue deformity or strain, offering qualitative information as color maps and semi-quantitative data through the strain ratio, which compares the elasticity of the tissue of interest with that of surrounding tissues. In contrast, SWE measures the velocity of shear waves (SW) generated by an acoustic radiofrequency pulse and converts this velocity to tissue elasticity (kPa) using Young's modulus [1,2].

Several factors influence the reproducibility of EUS measurements, including lesion size, depth, region of interest (ROI) size, and inherent tissue heterogeneity between the lesion and background tissue [2-10]. These variables can affect the accuracy and reliability of EUS, making it essential to understand their impact for improved clinical application. Previous research has suggested that a smaller ROI can increase inter-observer reproducibility [4]. Through years of clinical practice, the authors have observed that increasing the ROI size relative to the lesion size improves the representativeness of the elasticity value and enhances reproducibility. Additionally, measuring EUS values in in-vivo settings more accurately reflects clinical practice, as it incorporates tissue heterogeneity and lesion dynamics that phantom models cannot replicate.

Epidermal cysts are the most common benign cutaneous tumorlike lesions, typically characterized by a keratinizing squamous epithelium with a granular layer. They are commonly found on the face, scalp, and trunk [11,12]. Although most epidermal cysts remain asymptomatic and slowly enlarge as benign dermal nodules, they can occasionally grow large enough to compress surrounding structures or become inflamed or ruptured; in such cases, surgical excision is usually the treatment of choice [13-15]. Because epidermal cysts are located superficially, they provide a clear and accessible target for EUS measurement. Despite their clinical relevance, however, the reproducibility of EUS measurements in epidermal cysts has not been fully investigated. Prior studies have relied mainly on phantom models, which cannot capture the heterogeneity of real lesions or in-vivo tissue dynamics. Unlike those phantom-based investigations, the present study focuses on the reproducibility of EUS measurements in actual epidermal cysts, offering insights directly applicable to clinical practice. Specifically, the present study evaluated the intra-observer agreement of SE and SWE in the assessment of surgically confirmed epidermal cysts, with particular attention to the impact of ROI size. The findings are hoped to clarify how varying ROI diameter influences EUS measurements and how these findings can be used to improve everyday clinical workflows.

The mean height and length of epidermal cysts were 8.54±4.93 mm and 15.89±8.05 mm, respectively. The mean area of epidermal cysts was 119.41±106.96 mm².

The anatomical locations of epidermal cysts included the face and neck (n=89), back (n=58), chest (n=19), lower extremity (n=21), upper extremity (n=9), and buttocks (n=5). The mean ROI diameter was 2.41±1.81 mm (range, 0.66 to 9.99 mm) for SE images and 2.18±1.66 mm (range, 0.66 to 9.84 mm) for SW images. The mean ROI-to-lesion size ratios were 0.08±0.14 for SE images and 0.07±0.14 for SW images.

In group 1 (ROI diameter <2 mm), the mean ROI diameter was 1.39±0.31 mm for SE images (n=115) and 1.38±0.31 mm for SW images (n=137), with a mean ROI ratio of 0.03±0.04 for SE images and 0.03±0.02 for SW images. In group 2 (ROI diameter ≥2 mm), the mean ROI diameter was 3.77±2.01 mm for SE images (n=86) and 3.89±2.04 mm for SW images (n=64), with a mean ROI ratio of 0.15±0.18 for SE images and 0.18±0.20 for SW images. There was no significant difference in the height or depth of the lesion between the two groups (P>0.05 for both) (Figs. 1, 2).

Radiologist 1 performed the US examinations in 73 cases, while radiologist 2 performed the US examinations in 128 cases. There were no statistically significant differences between cases performed by the two radiologists in terms of lesion length, depth, or frequency of ROI size (P>0.05 for all comparisons).

The intra-observer agreement for strain ratio and SWE values is summarized in Table 1. Intra-observer agreements for strain ratio and SWE values were moderate overall. Subgroup analysis revealed moderate agreement for all parameters except SW velocity in group 2, which showed good agreement (Table 1). The ICC values for strain ratio, SW velocity, and SW elasticity in group 2 were higher than those in group 1, but the differences were not statistically significant.

The relationships between ROI diameter, ROI ratio, and measurement differences for strain ratio and SWE values are presented in Table 2. The mean differences in strain ratio were not significantly associated with ROI diameter, but showed a positive correlation with the ROI ratio. In contrast, mean differences in SW velocity and SW elasticity were significantly negatively correlated with both ROI diameter and ROI ratio (Table 2).

Mean differences in strain ratio did not differ significantly between group 1 and group 2 (1.15±2.19 vs. 1.11±1.56, P=0.089). However, group 2 demonstrated significantly lower mean differences in SW velocity (1.20±2.12 vs. 1.17±1.32, P=0.005) and SW elasticity (79.47±97.89 vs. 49.22±65.53, P=0.030) compared to group 1.

Bland-Altman plots showed small systematic bias for all EUS parameters. The 95% LoAs for strain ratio, SW velocity, and SW elasticity were narrower in group 2 than in group 1 (strain ratio, -4.820 to 4.898 vs. -3.850 to 3.679; SW velocity, -5.440 to 5.963 m/s vs. -3.738 to 3.064 m/s; SW elasticity, -416.604 to 255.348 kPa vs. -290.701 to 143.124 kPa) (Fig. 3). No proportional bias was detected (regression P>0.10 for all).

The most important finding of this study was that an ROI diameter of ≥2 mm resulted in a smaller mean difference for SWE measurements when assessing epidermal cysts. The results showed that the mean difference in strain ratio did not vary significantly with absolute ROI diameter, but it increased as the ROI ratio grew.

A 2-mm cutoff for ROI diameter emerged as optimal because statistical analysis demonstrated that differences in EUS measurements decreased significantly once this threshold was exceeded. Furthermore, previous studies have also referenced 2 mm as the optimal ROI size for EUS measurements, supporting the validity of this threshold in this study. Given these findings, the authors recommend using a minimum ROI diameter of 2 mm to enhance SWE agreement in clinical practice.

In this study, SW velocity in group 2 showed good intra-observer agreement (ICC, 0.770), whereas the other parameters showed moderate agreement. ICC values for SW velocity were slightly higher than those for SW elasticity, likely because the conversion of velocity to Young’s modulus assumes isotropic tissue, such as water, and the added calculation step may therefore lower agreement for SW elasticity [2,3,7].

Prior work on ROI size and reproducibility has yielded inconsistent results. Dudea-Simon et al. [4] found higher inter-observer reproducibility with smaller ROIs (5 mm), whereas Havre et al. [9] reported no significant effect of ROI size on strain ratio in a phantom. Many EUS studies, though not focused on ROI size, have nonetheless adopted diameters of 2-3 mm [3,5,10,17-19]. In this study, epidermal cysts were usually less than 50 mm in diameter, with a mean length of 16.89 mm, which led us to select smaller ROI diameters for assessing intra-observer agreement, compared with the prior study.

The findings of this study indicate that a larger ROI diameter correlated with reduced mean differences in both SW velocity and SW elasticity. These results may account for the higher ICC values for SWE observed in group 2. Although Dudea-Simon et al. [4] reported improved reproducibility with small ROIs, their study defined "small" as 5 mm, a definition that does not contradict these findings.

In this study, the mean difference in strain ratio was unrelated to absolute ROI diameter; however, it increased when the ROI occupied a larger proportion of the lesion. This trend may reflect greater heterogeneity both within the target lesion and in the surrounding reference tissue captured by a larger ROI, thereby amplifying measurement variability. These findings indicate that the ROI ratio is a relevant factor for strain ratio repeatability and should be carefully standardized in future protocols.

This study has several limitations. First, the retrospective design may have introduced selection bias, and ROI sizes were not predetermined but instead evolved with operator experience. Because ROI placement was manual, the criteria for ROI size varied over time, and larger ROIs were selected more frequently in the later period. ROI sizes were adjusted to lesion size during examinations; rather than grouping cases, the present study classified them by ROI diameter and analyzed the ratio of ROI size across the entire dataset. Second, case numbers differed among ROI subgroups, which may have affected the assessment of intra-observer agreement. Prospective studies are needed to validate these findings. Third, manual ROI selection made it impossible to guarantee identical ROI sizes for both the reference tissue and the epidermal cyst. Likewise, SE and SWE images were not always acquired with identical ROIs. Although the researchers aimed for consistency, manual drawing and the time required to switch machine settings between SE and SWE prevented perfect alignment. Finally, the ROI size of the reference tissue may influence the strain ratio relative to the ROI ratio; however, this study did not assess the impact of varying ROI size of reference tissue or the ROI size of the epidermal cyst using fixed relative values.

In conclusion, this study demonstrated good intra-observer agreement for SW velocity measurement of epidermal cysts when the ROI diameter was ≥2 mm. A larger ROI diameter (≥2 mm) was also associated with a smaller mean difference in SWE measurements. Conversely, a large ROI ratio increased the mean difference in strain ratio, underscoring the need to tailor ROI selection to the specific EUS technique. These findings can guide radiologists in optimizing ROI size to achieve better intra-observer agreement during EUS of epidermal cysts.