Static and dynamic sonography of facial muscles in healthy subjects – Impact of the process of manual muscle segmentation on inter-observer-reliability

• Published in: Ultraschall in der Medizin
• Main Authors: Schüler, T, Heinzl, A, Volk, GF, Guntinas-Lichius, O
• Format: Conference Proceeding
• Language: English; German
• Published: 18.08.2016
• ISSN: 0172-4614; 1438-8782
• DOI: 10.1055/s-0036-1587892

Purpose: Identifying the impact of manual segmentation on the reliability of static and dynamic ultrasound measurements of facial muscles. The inter-observer-reliability (for scanning and manual segmenting by two independent observers) in recent studies has shown good results in area measurements (intra class coefficient (ICC)/median: 0.960 – 0.997/0.98), but high variability in longitudinal and cross-diameter (0 – 0.954/0.83 Sauer, 2014). The aim of this study was to figure out the impact of the muscle-segmentation on the ICC in contrast to the impact of the ultrasound scans. Material and methods: Bilateral scans of seven facial muscles and two masticatory muscles were performed on 10 volunteers (5 women; age: 21 to 27 years) using an eZono4000, eZono, Germany, with linear probe (L3 – 12). The scans were based on a standardized examination protocol (Sauer, 2014) and were made by two different examiners (T.S. and A.H.) at two separate time points. Both sets of ultrasound pictures were manually segmented by both examiners. Results: There was a high inter-observer-reliability (scanning and segmenting by two different examiners) in area measurements (0.49 – 0.98/0.91) and high variability in diameter measurements (0 – 0.91/0.66). The inter-rater-reliability (segmenting of the same scans by two different examiners) showed similar results with nearly the same ICCs (areas: 0.814 – 0.98/0.92; diameters: 0.12 – 0.95/0.75). Conclusion: The similar ICCs of inter-observer-reliability and inter-rater-reliability prove that most variability is due to the process of manual segmentation, not to the ultrasound scanning. Reversely, this means that the existing ultrasound protocol is already a good instruction to achieve reproducible images of facial muscles. To improve the reproducibility of the segmentation, especially of the longitudinal and cross-diameter measurements, based on the presented results, a new instruction focused on segmentation was developed. Using this segmentation-instruction is believed to further improve the reliability of quantitative ultrasound of facial muscles in future studies.