Resting state functional MRI（rsfMRI）is a promising method for investigating brain disorders because it is non-invasive and does not require task performance as in typical task-fMRI studies. Since rsfMRI first description, many studies on brain disorders have demonstrated its potential as a detection or diagnosis tool. However, most conclusions are based on group studies. Their application to individuals requires overcoming some challenges such as reducing physical and physiological noises. Several methods of noise reduction for rsfMRI have already been established but further noise reduction is required for an individual application of rsfMRI. Here we propose a new approach in which we restricted the MRI measurement environment to the MRI scanner and one subject, from which we generated a noise set that was subtracted from rsfMRI data acquired with the same MRI scanner or subject. The present preliminary study examined the validity of the proposed approach using data from a phantom model and a human subject. Noises from rsfMRI data were removed by regression of previously prepared noise signals. The resulting signal presented properties similar to band-pass filtered one, suggesting the efficacy of the approach. In the future, the approach potential will be assessed using with more human subjects.
Sung et al. (Thu,) studied this question.