A statistical shape model of trochlear dysplasia of the knee

• Published in: The knee 2014-03, Vol.21 (2), p.518-523
• Main Authors: Van Haver, A, Mahieu, P, Claessens, T, Li, H, Pattyn, C, Verdonk, P, Audenaert, E.A
• Format: Article
• Language: English
• Subjects: Adolescent; Adult; Age; Automation; Cartilage, Articular - diagnostic imaging; Case-Control Studies; Classification; Discriminant Analysis; Female; Femur - abnormalities; Femur - diagnostic imaging; Geometry; Humans; Image Processing, Computer-Assisted; Imaging, Three-Dimensional; Knee; Male; Methods; Middle Aged; Models, Statistical; Morphology; Multidetector Computed Tomography; Orthopedics; Patellofemoral Joint - abnormalities; Patellofemoral Joint - diagnostic imaging; Principal Component Analysis; Statistical shape analysis; Studies; Trochlear dysplasia; Young Adult
• ISSN: 0968-0160; 1873-5800
• DOI: 10.1016/j.knee.2013.11.016

Abstract Background Trochlear dysplasia is known as the primary predisposing factor for patellar dislocation. Current methods to describe trochlear dysplasia are mainly qualitative or based on a limited number of discrete measurements. The purpose of this study is to apply statistical shape analysis to take the full geometrical complexity of trochlear dysplasia into account. Methods Statistical shape analysis was applied to 20 normal and 20 trochlear dysplastic distal femur models, including the cartilage. Results This study showed that the trochlea was anteriorized, proximalized and lateralized and that the mediolateral width and the notch width were decreased in the trochlear dysplastic femur compared to the normal femur. The first three principal components of the trochlear dysplastic femurs, accounting for 79.7% of the total variation, were size, sulcus angle and notch width. Automated classification of the trochlear dysplastic and normal femora achieved a sensitivity of 85% and a specificity of 95%. Conclusions This study shows that shape analysis is an outstanding method to visualise the location and magnitude of shape abnormalities. Improvement of automated classification and subtyping within the trochlear dysplastic group are expected when larger training sets are used. Clinical relevance Classification of trochlear dysplasia, especially borderline cases may be facilitated by automated classification. Furthermore, the identification of a decreased notch width in association with an increased sulcus angle can also contribute to the diagnosis of trochlear dysplasia.