Author : Jagdeep Kaur and Ruchika Chhabra

The FMRI images are spatially noisy and require a great amount of image pre-processing operations in order to extract the desired trend. Further, the direct implementation of spatial filter on fMRI images using loops is not computationally efficient and takes long time. Therefore, the dimension of the fMRI image data has to be brought down by using the principal component analysis. In fact, after using the PCA on fMRI data, the image data should come to the order from where they can be processed spatially using the image processing techniques. Therefore, specific spatial filters that filter out noisy pixels in 2-d in form of pixels and later on voxels in 3-d, at a high speed are the requirement of the presented work. This task is taken up by bringing the fMRI image to 2-d image frames. In the existing MRI analysis, it never gives estimation about the correlation between particular activity the person is engaged in, and the brain thematic region. However, this can be approximated by statistically analyzing the MRI images after stimulated by engaging the person under scanner towards a target functional activity e.g. by giving some logical calculation. Also, the brain mapping or functional MRI predictions is multivariate problem where, the observations are less than the variables comprising that observation. This problem has been framed in statistical domain and a penalized function methodology has been adopted to correlate the functional activity to the particular brain area.

Keywords: fMRI Image, Pixel Neighborhood Segmentation etc.

Article published in International Journal of Current  Engineering  and Technology, Vol.4,No.3 (June- 2014)