A novel semi-automatic hip morphology assessment tool is more accurate than manual radiographic evaluations. Issue 1 (2nd January 2020)

• Record Type: Journal Article
• Authors: Khayyeri, Hanifeh; Väänänen, Sami P.; Flivik, Gunnar; Jurvelin, Jukka S.; Dahlberg, Leif; Isaksson, Hanna
• Is Part Of: Computer methods in biomechanics and biomedical engineering. Volume 8:Issue 1(2020)
• Page Start: 76
• Page End: 86
• Publication Date: 2020-01-02
• Subjects: Hip osteoarthritis -- Statistical Appearance Model -- hip morphology -- diagnostics; Imaging systems in biology -- Periodicals; Imaging systems in medicine -- Periodicals; Biomechanics -- Data processing -- Periodicals; Biomedical engineering -- Periodicals; 616.0757
• DOI: 10.1080/21681163.2019.1578266
• Languages: English
• ISSNs: 2168-1163

Abstract

ABSTRACT: Radiological and pathological characteristics of hip osteoarthritis (OA) is joint-space loss due to degradation of articular cartilage. However, patients with early-stage OA do not yet show any radiological signs, which leaves them without diagnosis and treatment. This study evaluates the potential of a novel tool to identify pre-radiographic OA changes based on hip bone morphology. Two statistical appearance models for femur and pelvis were used to estimate the 3Dmorphology of the hip bones based on planar radiographs from patients. Well-known hip geometrical parameters (n = 22) were computed from patient CT scans (truth), 3D reconstructions (new method) and radiographs (calculated manually). The methods were compared by measuring relative error to truth. The new method was significantly more accurate in calculating hip geometrical parameters than the manual 2D calculations. The proposed approach could also capture rotational parameters like cross-over sign and anterior wall sign (100% correct predictions). The method can successfully reconstruct 3D hip shapes and densities for patients that have not yet developed severe osteoarthritis, and provided higher precision than manual estimations. Thus, it may be used to calculate morphological parameters that are predictors of OA and can become a powerful tool in human hip OA research and diagnostics.