Inter-Examiner Disagreement for Assessing Cervical Multifidus Ultrasound Metrics Is Associated with Body Composition Features

Abstract

Ultrasound imaging (US) is a biosensing technique that is widely used in several healthcare disciplines (including physiotherapy) for assessing multiple muscle metrics, such as muscle morphology and quality. Since all biosensors need to be tested in order to demonstrate their reliability, accuracy, sensitivity, and specificity, identifying factors that affect their diagnostic accuracy is essential. Since previous studies analyzed the impact of sociodemographic but not body composition characteristics in US errors, this study aimed to assess whether body composition metrics are associated with ultrasound measurement errors. B-mode images of the lumbar multifidus muscle at the L5 level were acquired and analyzed in 47 healthy volunteers by two examiners (one experienced and one novice). The cross-sectional area, muscle perimeter, and mean echo intensity were calculated bilaterally. A correlation analysis and a multivariate linear regression model were used for assessing the inter-examiner differences with respect to body composition metrics. The results demonstrated good-to excellent reliability estimates for the cross-sectional area, muscle perimeter, aspect ratio, roundness, circularity, and mean brightness metrics (all ICC > 0.85). However, solidity showed unacceptable reliability (ICC < 0.7). Age, height, total lean mass, trunk lean mass, and water volume were associated with inter-examiner disagreement on mean echo intensity. Cross-sectional area, perimeter, and roundness measurement errors were associated with lean mass and water volume.