|
|
 |
|
|
|
My advisor is Professor Jiayang Sun. I also work with Dr. Kath Bogie (The Cleveland functional electrical stimulation(FES) Center and Department of Orthopedics, CWRU); Dr. Guang H. Yue (Department of Biomedical Engineering, The Cleveland Clinic Foundation); Dr. Ken Gustafson (Neural Engineering Center, Department of Biomedical Engineering, CWRU). Mixture Estimation and Bump Hunting with Measurement Error || Data Mining with Biomedical Image Data || Computational Neuroscience Mixture Estimation and Bump Hunting with Measurement Error The effects of measurement error are well-known. Analyses that ignore measurement errors could be misleading. For example, if the true density is bimodal, the density of the data measured with measurement error is unimodal. In this research, we would like to recover the density of the interest based on observations with measurement error when a direct observation is not possible. We develop a new non-parametric method to solve the deconvolution problem.
Data Mining with Biomedical Image Data This project is concerned with the development of an effective clinical protocol for the qualitative evaluation of interface pressures obtained from seated wheelchair users through systematic analysis of image data. The data utilized in the development of this protocol was obtained as part of a study to determine the effects of gluteal neuromuscular electrical stimulation (NMES) on tissue health [Bogie & Triolo, 2003]. However, the principles are valid for any series of interface pressure measurements obtained over time, such as when monitoring seating needs for an individual with mobility impairment. This experimental protocol leads to a challenging condition for data mining because the data set is large in dimension and small in sample size. In addition, if subjects do not sit at the same orientation for each session, images may not align spatially. There may be further artifacts from differences between stimulation movies if images from different phases of the stimulation are not aligned temporally. We have developed spatial and a temporal registration schemes so that potential image differences due to an intervention can be examined after registration. These registration methods are applicable to typical image data obtained from clinical seating assessments. We propose a statistical mapping method for analyzing large image data sets that leads to an efficient procedure for computing a False Discovery Rate (FDR) map. The FDR map allows us to determine whether a change is clinically relevant or spurious. We have applied our registration and statistical methods to data obtained from the gluteal NMES study and have defined a measurement protocol for the effective detection of significant changes in seating interface pressures.
The spinal cord contains the neural circuitry required to generate coordinated movements, but detailed maps of the spinal neural circuitry are not available. A spatial organization of receptive fields and a modular organization of the flexion withdrawal reflex system have been recently supported. However, the location and topographical organization of interneurons involved in flexion reflex pathways have not been systematically examined. In this project, we shall determine the anatomical locations of spinal neurons involved in the hindlimb flexion withdrawal reflex using expression of the immediate early gene c-fos. A Spatial regression model is developed to fit the data. A parametric mapping with a FDR controlling procedure is given to determine locations within the spinal cord where stimulation caused a significant increase in the number of active spinal neurons. The FDR controlling procedure overcomes the multiplicity effect from the hundreds of null hypotheses involved above. The results suggest that the region of increased spinal activity upon stimulation is similar in shape to a traditional motoneuron population. Parallel 3-D distributions are observed for the two stimulated nerves in the upper laminae. The results also support the spatial/modular organization of the flexion withdrawal reflex system within the spinal cord.
Increased fatigability occurs in every patient with muscle weakness, regardless of whether the weakness is due to a central or peripheral neurological disorder. The underlying mechanisms are not well understood and there is a need to study fatigability systematically in neurology and rehabilitation. The behavior of the peripheral neuromuscular system during muscle fatigue has been studied extensively, but the role of the central nervous system in muscle fatigue is largely unknown. We believe that without a good understanding of mechanisms of fatigue in health, an assessment of mechanisms contributing to increased fatigability in neurological disorders is difficult. We are investigating changes in brain activity during motor performance from non-fatigued to moderately fatigued to severely fatigued conditions in healthy volunteers (NIH grant NS37400). A series of Multivariate time Series Models are used to fit the data.
| |||||
|
|
