fMRI Image Processing and Functional Region Detection in the Wavelet Domain

The traditional Gaussian smoothing usually cannot effectively reduce the correlated noise in fMRI data, which inevitably affects the final detection results.In order to detect and locate functional active regions more accurately, a method based on wavelet transform is proposed in this work.At first, fMRI data is de-noised in the wavelet domain by soft-thresholding.Then, the activated spots are detected on the de-noised simulated time series using the false discovery rate similarly in the wavelet domain.The detection results show that the proposed method can reduce the number of false positive spots while maintain the detection sensitivity, and manifests a better specificity and reliability than the traditional Gaussian smoothing does.