Computer-based quantification of the Achilles tendon thickness in sequential B-mode ultrasound images: a study of feasibility and reliability

• Published in: Archives of orthopaedic and trauma surgery Vol. 134; no. 10; pp. 1443 - 1449
• Main Authors: Syha, R., Grau, S., Nieß, A. M., Hein, T., Werner, M. K., Ketelsen, D., Schick, F., Best, R.
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.10.2014; Springer Nature B.V
• Subjects: 1988; Achilles tendon; Achilles Tendon - anatomy & histology; Achilles Tendon - diagnostic imaging; Adult; Algorithms; Arthroscopy and Sports Medicine; Automation; Contour detection; DISORDERS; Feasibility Studies; Female; Healthy Volunteers; Humans; Image Interpretation, Computer-Assisted - methods; Kirurgi; Male; Medicine; Medicine & Public Health; Middle Aged; Observer Variation; Orthopedics; Ortopedi; P87; RADIOLOGIC CLINICS OF NORTH AMERICA; Reliability; Reproducibility of Results; RNAGE BD; SONOGRAPHY; Surgery; SYSTEM; TENDINOPATHY; Tendons; Ultrasonic imaging; ULTRASONOGRAPHY; Ultrasound; V26; Achilles tendon; Reliability; Ultrasound; Contour detection
• DOI: 10.1007/s00402-014-2043-3
• ISSN: 0936-8051

Introduction Assessment of the Achilles tendon thickness (ATT) using B-mode ultrasound is a common technique for clinical evaluation of chronic mid-part tendinosis. Currently used image-based assessment is limited by relatively high inter- and intra-observer variability. In this study, it was tested whether a new sequence-based automated assessment of ATT provides more reliable and reproducible results than the standard image-based procedure. Materials and methods A total of 118 non-operated tendons of 59 healthy subjects (44, range 28–50 years) were analysed using an automated image based as well as a newly developed automated sequence-based method. Correlation and agreement of both methods were evaluated. The root mean square deviation (RMSD) and a Bland–Altman analysis were performed to highlight observer ( n  = 18 tendons) as well as reader ( n  = 40 tendons) dependent variabilities of both methods. Results A strong correlation was found between image and sequence-based ATT assessment ( p  = 0.92). The Bland–Altman analysis showed a good agreement between both methods (mean difference 0.0018, 95 % CI: −0.046; 0.05). In repetitive examinations, sequence-based analysis showed a significant reduction concerning reader- and observer-dependent variability compared to image-based assessment. The RMSD for repetitive sequence-based measurements was approximately 0.3 mm (compared to 0.6 mm for image-based measurement), respectively. Conclusions The study shows sequence-based automated assessment of ATT being clearly superior to the standard image-based procedure. The new method provides a clear reduction of reader as well as observer-dependent variability. Due to the decreased scattering of measurement data sequence-based measurement seems especially valuable for quantification of small tendon thickness changes such as exercise-induced hypertrophy.