Project Summary Understanding functions of human brains is a very challenging yet crucially important task. While modern technologies allow us to investigate how our brains work to some degree, many open questions await to be answered as highlighted in the recently unveiled Brain Research through Advancing Innovative Neurotechnologies (BRAIN) Initiative. Building upon the PIs experience and interest, the proposed project addresses a newly emerging line of statistical research on this area, namely experimental designs for pioneering neuroimaging techniques. As one of the key steps of scientific investigations, selecting high-quality experimental designs for collecting informative data to make precise and valid statistical inference is crucial. However, research on this area is very scarce, hindering the usefulness of the advanced brain mapping techniques. The proposed project will address this scarcity by developing novel approaches for obtaining good designs. The results are expected to greatly improve the quality of neuroimaging experiments to allow a better understanding of brain functions. In addition, many new, challenging statistical research problems are expected to arise in this new area. The outcome of this proposal will also serve as a stepping stone for future research. This is an opportune time for making important contributions. Intellectual Merit. The first part of the proposed project addresses experimental design issues for one of the most dominant brain mapping technologies, namely functional magnetic resonance imaging (fMRI). Recent years have seen an explosive growth in fMRI data analysis. However, research on fMRI designs has not kept up; many statistical challenges remain to be overcome. First, existing studies focus mainly on computational methods. Fundamental theories, while crucial, are very rare. This proposal will address this by deriving novel analytical results. The results will not only contribute to basic statistical research, but also help to guide the selection of good fMRI designs for real experiments. In addition, current results on fMRI designs are only for basic settings with simple statistical models. Sophisticated experimental settings with advanced statistical techniques become increasingly important for answering complex scientific questions. For these experiments, new fMRI designs are required. To fulfill this need, the PI will develop approaches to obtain optimal fMRI designs for various practical situations. He will work on cases with (i) a linear statistical model whose model matrix depends on the experimental subjects probabilistic behavior; (ii) a nonlinear mixed effects model allowing for time-varying shape of the hemodynamic response function (HRF, which models subjects brain activity in response to a mental stimulus); and (iii) flexible semi- and non-parametric methods for yielding a better interpretation of complex fMRI data. Moreover, the PI will extend the obtained results to cases involving other pioneering brain mapping techniques, including functional near-infrared spectroscopy and electroencephalography. There literally is no systematic study on design selection for such experiments. Furthermore, the PI will develop free, user-friendly computer software packages for researchers to easily obtain designs suited to their experiments. Broader Impacts. The results of the proposed project will be disseminated in conferences and journal publications. They will be of great practical importance by allowing us to collect informative data to improve our knowledge on brain functions, and to help a better understanding of, say, brain disorders such as Alzheimers disease. These results are thus expected to have a significant impact to the society. In addition, conducting high quality research is not the only emphasis of this proposal. Education also plays an important role. The PI will continue to train and educate students on conducting research projects, and to hold regular meetings for a focused study group to discuss and brainstorm new ideas. He will also develop new courses on experimental designs, and give lectures in neuroscience courses. Results of this proposal will be an integral part of the course materials. The PI belongs to one of the leading groups that have made important contributions to this new research area. He is highly qualified and ready to perform the tasks described in the proposal. The results are expected to help to move this field forward, and invite much future research on this and related areas.
|Effective start/end date||7/1/14 → 6/30/20|
- National Science Foundation (NSF): $400,000.00
Design of experiments
Near infrared spectroscopy
Magnetic Resonance Imaging