Knoops, Paul G.M.; (2019) Physical and statistical shape modelling in craniomaxillofacial surgery: a personalised approach for outcome prediction. Doctoral thesis (Ph.D), UCL (University College London).

Preview

Text thesis_final_PK.pdf - Accepted Version Download (5MB) | Preview

Abstract

Orthognathic surgery involves repositioning of the jaw bones to restore face function and shape for patients who require an operation as a result of a syndrome, due to growth disturbances in childhood or after trauma. As part of the preoperative assessment, three-dimensional medical imaging and computer-assisted surgical planning help to improve outcomes, and save time and cost. Computer-assisted surgical planning involves visualisation and manipulation of the patient anatomy and can be used to aid objective diagnosis, patient communication, outcome evaluation, and surgical simulation. Despite the benefits, the adoption of three-dimensional tools has remained limited beyond specialised hospitals and traditional two-dimensional cephalometric analysis is still the gold standard. This thesis presents a multidisciplinary approach to innovative surgical simulation involving clinical patient data, medical image analysis, engineering principles, and state-of-the-art machine learning and computer vision algorithms. Two novel three-dimensional computational models were developed to overcome the limitations of current computer-assisted surgical planning tools. First, a physical modelling approach – based on a probabilistic finite element model – provided patient-specific simulations and, through training and validation, population-specific parameters. The probabilistic model was equally accurate compared to two commercial programs whilst giving additional information regarding uncertainties relating to the material properties and the mismatch in bone position between planning and surgery. Second, a statistical modelling approach was developed that presents a paradigm shift in its modelling formulation and use. Specifically, a 3D morphable model was constructed from 5,000 non-patient and orthognathic patient faces for fully-automated diagnosis and surgical planning. Contrary to traditional physical models that are limited to a finite number of tests, the statistical model employs machine learning algorithms to provide the surgeon with a goal-driven patient-specific surgical plan. The findings in this thesis provide markers for future translational research and may accelerate the adoption of the next generation surgical planning tools to further supplement the clinical decision-making process and ultimately to improve patients’ quality of life.

Download activity - last month

Export as XMLCSVJSON

Download activity - last 12 months

Export as CSVXMLJSON

Downloads by country - last 12 months

Export as JSONCSVXML

Archive Staff Only

View Item