Dr. Stephen Strother
Senior Scientist,
Rotman Research Institute, Baycrest,
Professor, Department of Medical Biophysics, University of Toronto
My research is directed at developing and testing a set of optimal experimental planning, standardization and analysis tools for neuroimaging researchers that are coupled with state-of-the-art neuroimaging research databases. The overall goal is to yield new insights into human mental functions, how they are changed with normal aging, and compromised by damage and disease across the lifespan. We pursue this by merging and formally optimizing multi-modality neuroimaging (e.g., positron emission tomography, structural and functional magnetic resonance imaging, electroencephalography) and meta-data using machine learning and modern biostatistical techniques.
A major focus of this work involves multi-institutional support of the Brain-CODE data repository and analytics infrastructure at HPCVL supported by the Ontario Brain Institute (OBI). In particular we are co-investigators within the multi-institutional, independent disease programs in neurodegeneration and depression funded by OBI, and the multi-institutional stroke program of the Canadian Partnership for Stroke Recovery.
My goals require optimization of the experimental designs, data-analysis algorithms and associated software tools being developed for cognitive neuroscience and clinical neuroimaging as a function of age and disease, in the so-called "processing pipeline". Processing pipeline choices in neuroimaging pose a range of critical issues, particularly which tools/packages to use for a particular experiment. However, there is growing evidence that new insights into human brain function may be obscured by poor choices in the image-processing pipeline particularly as a function of age and disease. We are focusing on using HPC to optimize pipelines for measuring brain networks with multivariate machine learning models and resampling techniques from the field of statistical learning theory. In particular we are developing a quantitative, optimization framework called NPAIRS that focuses on using prediction and reproducibility metrics. An initial software package is distributed as an opensource Java package, and several other packages are under development and undergoing evaluation for commercialization.
For more information on my science background please click here.
A major focus of this work involves multi-institutional support of the Brain-CODE data repository and analytics infrastructure at HPCVL supported by the Ontario Brain Institute (OBI). In particular we are co-investigators within the multi-institutional, independent disease programs in neurodegeneration and depression funded by OBI, and the multi-institutional stroke program of the Canadian Partnership for Stroke Recovery.
My goals require optimization of the experimental designs, data-analysis algorithms and associated software tools being developed for cognitive neuroscience and clinical neuroimaging as a function of age and disease, in the so-called "processing pipeline". Processing pipeline choices in neuroimaging pose a range of critical issues, particularly which tools/packages to use for a particular experiment. However, there is growing evidence that new insights into human brain function may be obscured by poor choices in the image-processing pipeline particularly as a function of age and disease. We are focusing on using HPC to optimize pipelines for measuring brain networks with multivariate machine learning models and resampling techniques from the field of statistical learning theory. In particular we are developing a quantitative, optimization framework called NPAIRS that focuses on using prediction and reproducibility metrics. An initial software package is distributed as an opensource Java package, and several other packages are under development and undergoing evaluation for commercialization.
For more information on my science background please click here.
|
Rotman Research Institute | Baycrest
3560 Bathurst Street, North York Ontario, Canada M6A 2E1 |
