_Magnetic Resonance Imaging （fMRI）功能性磁共振成像（fMRI）课件.ppt

Functional Magnetic Resonance Imaging (fMRI) Alexander Wolf Supervisor : Dr. Peter Tischer Presentation Overview Project Objectives Project Background Results Conclusion Future Work Project Objectives - Investigate filtering of fMRI data. Develop new filtering techniques. Develop software tools for fMRI analysis. fMRI Introduction fMRI is a sequence of MRI’s over time. Images contain White Matter, Grey Matter, Cerebral-Spinal Fluid and Other. Basic unit in fMRI is the volume element (voxel). fMRI Importance Neurologists use them to determine what regions of the brain where active/inactive for a known task performed by the subject. 3D fMRI image is segmented into active/inactive regions. Filtering fMRI Anomalous values can occur due to motion, image capture. Leads to incorrect analysis, false conclusions. To account for anomalous values, consider filtering approach. Classical methodologies in fMRI such as FSL(FMRIB Software Library) and SPM(Statistical Parametric Mapping) have not considered filtering of any kind. Filtering Objectives Recognition of anomalous data. Determine unreliable and reliable data. Determine whether voxel represents the same kind of voxels that belong together. Filtering Importance Can determine whether we have enough data to work with so we can eliminate the need for motion compensation altogether. Determines proportion of voxels that are affected by motion compensation. Replace voxels by more meaningful values. Determine which voxels represent activation. Filters Investigated Vector Median Filter(VMF) Tischer Least Change Filter(TLC) n-preserve Filter n-preserve Filter(New method). - Least Change Filter(New method) Vector Median Filter Replaces the center pixel by the pixel in the mask whose value is most inlying. Inlying is the value whose total sum of the absolute difference to every other pixel in the mask is minimised. Vector Median Filtering Vector Median Filtering Vector