Enhancing functional neuroimages: wavelet denoising as an alternative to Gaussian smoothing
Alle Meije Wink and Jos B. T. M. Roerdink. Enhancing functional neuroimages: wavelet denoising as an alternative to Gaussian smoothing. In: Proc. Intern. Conf. on Computer Vision and Graphics (ICCVG 2002), September 25-29, Zakopane, Poland, 2002, pp. 787-792.
We present a general wavelet-based denoising scheme for functional neuroimages and compare it to Gaussian smoothing, the standard method in functional neuroimaging. We adapted WaveLab thresholding routines to 2D data, and tested their effect on the signal-to-noise ratio of noisy images. In a simulated time series test, we also investigated the shapes of detected activations after denoising.
Download in pdf format