Reconstruction of the lower limb bones from digitised anatomical landmarks using statistical shape modelling

Statistical-Shape Prediction of Lower Limb Kinematics During Cycling, Squatting, Lunging, and Stepping—Are Bone Geometry Predictors Helpful?

“…The gold standard to obtain individualized bone geometry is the segmentation of shapes from high resolution 3D medical imaging ( Nolte et al, 2020 ). Therefore, the pelvis and lower limb bones of the study cohort were scanned using a 3-Tesla MAGNETOM Trio-Tim System MRI device (Siemens AG, Erlangen, Germany).…”

“…Recent advances in computational methodology allow for improved characterization of bone morphometry as well as motion at a population wide level. Statistical shape modeling enables to describe individualized bone geometry more precisely than consensus bone geometry or linearly scaled generic bone models ( Audenaert et al, 2019a ; Cerveri et al, 2020 ; Nolte et al, 2020 ). Similarly, statistical modeling of kinematics by non-linear methods as well as improvements in curve alignment methods during the pre-processing phase, might provide more reliable and stronger correlations between human anatomy and motion as opposed to previous reports ( Freedman and Sheehan, 2013 ; Moissenet et al, 2019 ; De Roeck et al, 2020 ).…”

Automatic Generation of Personalised Skeletal Models of the Lower Limb from Three-Dimensional Bone Geometries

“…This issue can be solved by combining the STAPLE toolbox with a statistical shape workflow to reconstruct entire bone geometries from sparse datasets (Nolte et al, 2016;Suwarganda et al, 2019). It is worth mentioning that, although statistical shape modelling workflows present the advantage of reconstructing bone geometries from sparse segmentations or even skin landmarks, bone models from medical image segmentation still provide the most accurate estimations of joint parameters; for example, median root-mean-squared errors up to 11.09 mm and larger than 13.8 mm (Nolte et al, 2020) have been reported in the identification of the centre of the femoral head using statistical shape models. Considering that radiological scans are routinely collected to plan musculoskeletal surgical interventions and that the time required to segment bones (Noguchi et al, 2020) has decreased by orders of magnitudes thanks to recent deep learning techniques, the generation of personalised lower limb models in a number of clinical applications appears technically feasible.…”

“…Statistical shape modelling workflows have recently demonstrated high potential for reconstructing bone geometries from sparse anatomical datasets (Davico et al, 2019;Nolte et al, 2016;Suwarganda et al, 2019) and landmarks digitized in the gait lab (Nolte et al, 2020;Zhang et al, 2016), but to the best of the authors' knowledge they do not yet offer methods to generate articulated skeletal models of the complete lower limb. The bone reconstructions are limited to the long bones (Nolte et al, 2020;Nolte et al, 2016) or omit the talus and foot bones (Davico et al, 2019;Suwarganda et al, 2019;Zhang et al, 2016), and in musculoskeletal modelling contexts they have been employed to perform non-linear scaling of pre-existing muscle attachments (Nolte et al, 2016) with scarce focus towards joint modelling. Hence, a comprehensive approach to generate entire lower limb models from personalised bone geometries is still missing.…”

Machine learning methods to support personalized neuromusculoskeletal modelling

“…The free and open-source framework the Musculoskeletal Atlas Project Client (MAP) (Zhang et al 2014) employs principal component analysis scaling as a method to synthesis 3-D bone geometries from sparse data. A principal component analysis scaling is more sophisticated than simple linear scaling (available in most musculoskeletal modelling software) and can accurately reconstruct bone shapes (Bahl et al 2019;Nolte et al 2016a;Nolte et al 2020;Suwarganda et al 2019;Zhang and Besier 2017;Zhang et al 2015), but is limited by the variation contained within the training data.…”