Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Development and validation of a new scaling method to enable clinical use of advanced in silico bone models
Our in silico models accurately evaluate bone strength based on patient scans and simulate disease and treatment but are not applied widely as they are computationally intensive. This project aims to develop a method to select representative subregions to facilitate clinical use.
Keywords: Finite Element, HPC, image processing, in silico, bone, osteoporosis,
The 9 million osteoporotic fragility fractures that occur every year currently lead to over 400,000 deaths and health care costs in excess of US$80 billion. Recent advances in medical imaging have enabled the assessment of bone structure at successive time points. We at the laboratory for bone biomechanics at the ETH Zurich have developed a variety of computational models that when combined with HR-pQCT scans can accurately estimate bone strength and fracture risk, and that seek to predict how the bone microstructure and strength will change as a result of bone diseases and their treatments. These models have proven capable of predicting changes in bone over 10 years of osteoporosis and its treatments but they have yet to be used in a clinical setting.
The goal of this project is to remove the most important barrier to the clinical use of our models: the high computational power requirement. Models accepting as input the entire HR-pQCT scans can only be run on the Swiss National Supercomputer (CSCS in Lugano) and even thus, they run over hours (advection models) to weeks (fully parallelized multiphysics models). We need to demonstrate that subregions of the whole scans can be chosen in such a way as to be representative of microstructural changes in the whole scans. If we can obtain similar results at significantly reduced scales we can create a tool for doctors and patients to predict fracture risk, assist in the selection of therapeutics and in the design of clinical trials.
We seek a motivated student to help us bring our computational innovations to clinics and reduce the disastrous consequences of osteoporosis. Don't hesitate to contact us with any questions!
If you happen to have experience in python, C++, high performance computing clusters (e.g. Euler, CSCS), knowledge of bone biology, familiarity with image-processing techniques that's great but not a requirement, you can also learn on the job!
The 9 million osteoporotic fragility fractures that occur every year currently lead to over 400,000 deaths and health care costs in excess of US$80 billion. Recent advances in medical imaging have enabled the assessment of bone structure at successive time points. We at the laboratory for bone biomechanics at the ETH Zurich have developed a variety of computational models that when combined with HR-pQCT scans can accurately estimate bone strength and fracture risk, and that seek to predict how the bone microstructure and strength will change as a result of bone diseases and their treatments. These models have proven capable of predicting changes in bone over 10 years of osteoporosis and its treatments but they have yet to be used in a clinical setting.
The goal of this project is to remove the most important barrier to the clinical use of our models: the high computational power requirement. Models accepting as input the entire HR-pQCT scans can only be run on the Swiss National Supercomputer (CSCS in Lugano) and even thus, they run over hours (advection models) to weeks (fully parallelized multiphysics models). We need to demonstrate that subregions of the whole scans can be chosen in such a way as to be representative of microstructural changes in the whole scans. If we can obtain similar results at significantly reduced scales we can create a tool for doctors and patients to predict fracture risk, assist in the selection of therapeutics and in the design of clinical trials.
We seek a motivated student to help us bring our computational innovations to clinics and reduce the disastrous consequences of osteoporosis. Don't hesitate to contact us with any questions!
If you happen to have experience in python, C++, high performance computing clusters (e.g. Euler, CSCS), knowledge of bone biology, familiarity with image-processing techniques that's great but not a requirement, you can also learn on the job!
The goals of the project would be to:
Investigate sample heterogeneity in regards to morphological parameters
Find an extraction strategy
H1: Extraction bias: Properties are altered due to extraction (e.g. Load distribution of simple compression will change due to extraction, should we add padding aso.)
H2: Biopsies that reflect certain global Mechanobiological measures can be extracted (e.g. Load Distribution, Initial Seeding, Mechanoregulation)
H3: Find 1 biopsy that most closely reflects all measures (e.g. using weights)
The goals of the project would be to: Investigate sample heterogeneity in regards to morphological parameters Find an extraction strategy H1: Extraction bias: Properties are altered due to extraction (e.g. Load distribution of simple compression will change due to extraction, should we add padding aso.) H2: Biopsies that reflect certain global Mechanobiological measures can be extracted (e.g. Load Distribution, Initial Seeding, Mechanoregulation) H3: Find 1 biopsy that most closely reflects all measures (e.g. using weights)
charles.ledoux@hest.ethz.ch
Charles Ledoux
PhD Student
ETH Zürich, HCP H 18.2
Leopold-Ruzicka-Weg 4
CH - 8093 Zürich
charles.ledoux@hest.ethz.ch Charles Ledoux PhD Student ETH Zürich, HCP H 18.2 Leopold-Ruzicka-Weg 4 CH - 8093 Zürich