Use this URL to cite or link to this record in EThOS:
Title: Using statistical models of shape in musculoskeletal biomechanics and orthopaedic reconstruction
Author: Nolte, Daniel
ISNI:       0000 0004 9356 9086
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Access from Institution:
Bone shapes significantly influence function, biomechanics and kinematics of the musculoskeletal system through the shape of articulating surfaces, muscle attachments and lever arms of the muscles defined by wrapping around boney prominences. Orthopaedic conditions such as trauma and osteoarthritis affecting bone shapes, therefore, have a significant impact on the individual’s mobility and independence. In the diagnosis and treatment of such conditions 3D bone models are widely used for patient-specific musculoskeletal simulations and surgical planning. Three-dimensional medical imaging is commonly used to evaluate bone shapes and the musculoskeletal anatomy, but this is not always available, is costly, and frequently requires significant trained expertise. Statistical shape modelling provides a compact parameterised representation of shapes of a larger population and a method that enables analysis and shape reconstruction. In this thesis methods and models supporting the diagnosis and surgical planning for orthopaedic reconstruction using statistical shape models were developed and evaluated. In the first part methods were developed to customise patient-specific musculoskeletal models by reconstructing bone shapes using landmark measurements from a marker-based motion capture system and to provide a method for non-linear scaling of muscle path geometries without medical imaging using shape model reconstructions. The second part introduces methods to reconstruct intact bone shapes from planar X-ray or 3D medical images of partial bones representing various levels of defect and evaluates the methods for use in clinical practice for surgical planning and intra-surgical assistance. As result, the described methods increased the accuracy of bone shape prediction from motion capture measurements from between 3.8 mm and 3.9 mm for linear scaling methods to between 2.6 mm and 3.0 mm average surface errors for shape model reconstructions. Muscle path geometry models using statistical shape models reduced the variability and the overall error of predicted attachment points by 9% to 20% compared to linear scaling methods. Predictions of intact bone shapes from 3D models were comparable to or better than using a mirrored model of the contralateral side for small defects. Reconstructions from planar X-ray showed average surface errors between 1.6 mm and 3.6 mm for defects between 0 and 50% of the distal end of the bone that was removed. The presented studies provide shape modelling methods with levels of accuracy that allow for application in clinical practice and provide models that may be used as part of multiscale musculoskeletal models for resource constrained clinical and research environments.
Supervisor: Bull, Anthony Sponsor: Beit Felloship for Scientific Research ; Engineering and Physical Sciences Research Council
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral