Functional magnetic resonance imaging (tMRI) is a method for detecting brain regions that are activated when a certain task is carried out. The method is useful in planning of neurosurgical procedures, where knowledge of the exact locations of important functions is needed to avoid damage to these regions. It is also an important tool in neurological research, where it is used to investigate the function of the human brain.
To find the activated regions, a sequence of images of the brain is collected while a patient or subject alters between resting and performing the task. The variations in image intensity over time is then compared to a model of the variations expected to be found in active parts of the brain. Locations where the intensity variations are similar to the model are considered to be activated by the task.
Since the images are very noisy, filtering is needed before the detection of activation. If adaptive filtering is used, i.e. if the filter at each location is adapted to the local neighborhood, very good detection performance can be obtained. This thesis presents two methods for adaptive filtering of fMRI data. One of these is based on canonical correlation analysis (CCA), and is an extension of a previously proposed CCA-based method. As in the old method, CCA is used in each neighborhood to find a spatial fi lter that maximizes the correlation to the model of the intensity variation. A novel feature of the presented method is that it is rotationally invariant, i.e. that it is equally sensitivelo activated regions in different orientations.
The other method is based on bilateral filtering. This method creates spatial filters which averages pixels with similar intensity variation. Since these filters are not optimized to maximize the similarity to the model of activated signals, the risk of declaring inactive pixels as active is lower compared to CCA-based methods.
Linköping: Linköpings universitet , 2005. , 57 p.