Morphological variation in paediatric lower limb bones

Abstract

Available methods for generating paediatric musculoskeletal geometry are to scale generic adult geometry, which is widely accessible but can be inaccurate, or to obtain geometry from medical imaging, which is accurate but time-consuming and costly. A population-based shape model is required to generate accurate and accessible musculoskeletal geometry in a paediatric population. The pelvis, femur, and tibia/fibula were segmented from 333 CT scans of children aged 4–18 years. Bone morphology variation was captured using principal component analysis (PCA). Subsequently, a shape model was developed to predict bone geometry from demographic and linear bone measurements and validated using a leave one out analysis. The shape model was compared to linear scaling of adult and paediatric bone geometry. The PCA captured growth-related changes in bone geometry. The shape model predicted bone geometry with root mean squared error (RMSE) of 2.91 ± 0.99 mm in the pelvis, 2.01 ± 0.62 mm in the femur, and 1.85 ± 0.54 mm in the tibia/fibula. Linear scaling of an adult mesh produced RMSE of 4.79 ± 1.39 mm in the pelvis, 4.38 ± 0.72 mm in the femur, and 4.39 ± 0.86 mm in the tibia/fibula. We have developed a method for capturing and predicting lower limb bone shape variation in a paediatric population more accurately than linear scaling without using medical imaging.

Roadmap for an imaging and modelling paediatric study in rural NZ

“…Pelvis, femurs, tibias and fibulas were manually segmented from the LAVA Flex Autobind WATER sequence using Stradview (University of Cambridge, United Kingdom) (Cignoni et al, 2008). The resulting point cloud was smoothed and remeshed using Meshlab (Carman et al, 2022). Each of the bones meshes in the dataset were fitted to a template mesh (one for the pelvis, one for the femur and one for the tibia/fibula) to achieve nodal correspondence for each bone.…”

Sex differences in linear bone measurements occur following puberty but do not influence femoral or tibial torsion

“…The pelvis, femurs, and tibia/ bulas were segmented from the CT scans and meshed into a 3D surface as outlined in a previous study [20]. A template mesh for each bone was then t to each segmented bone model, to achieve nodal correspondence between meshes.…”

“…A template mesh for each bone was then t to each segmented bone model, to achieve nodal correspondence between meshes. This was achieved by 1) non-rigidly registering the template model and 2) iteratively mesh tting the template model to the segmented data using radial basis functions [20]. Bones were aligned to an International Society of Biomechanics coordinate system convention [23] using automatic detection of bony landmarks calculated from each 3D mesh to ensure all bones were in the same orientation for measurement.…”

“…To use the automated bone measurement methods developed in this study, three-dimensional bone surfaces reconstructed from full lower limb CT or MRI scans are required. This limits the application of our method, however, we have also developed statistical shape models that can be used to predict the bone morphology from sparse anatomical landmarks [20]. Future integration of these methods will enable clinicians to compare patient-speci c data to our population norms.…”

“…However, discrepancies between imaging and measurement modalities [19], make comparisons challenging. Fitting 3D models to medical imaging data can alleviate these challenges [20], [21], providing a consistent and reproducible frame of reference for future comparison [22].…”