Abstract #0239

Atkinson I, Claiborne T, Flannery M, Thulborn K

The substantial acoustic noise produced by EPI sequences often used for fMRI make functional studies involving analysis of participant speech challenging.We proposed an adaptive technique for suppressing gradient noise from human speech.Experimental results indicate that proposed method is able to effectively cancel the noise induced by the switching gradients while preserving frequencies in typical human voice.When applied to human speech, the proposed scheme increased the ability of a nave listener to correctly identify individual English words spoken during fMRI by an average of 21.6% and decreased the time required to make this identification by an average of 17.5%.