Reference-scan-free method for automated correction of Nyquist ghost artifacts in echoplanar brain images.

One of the most common artifacts for echo-planar imaging is the Nyquist ghost, typically overcome with the aid of a reference scan preceding the actual image acquisition. In this work, a nonlinear phase correction obviating the need for a reference scan is proposed. The method is based on computing the second moments of the even and odd subdomains of the phase-encoding k-space data to retrieve the phase disparities between even and odd echoes. The method's underlying assumption is that the images are dominated by their low frequency and symmetric part. In vivo data demonstrate the effectiveness of the algorithm showing ghost suppression comparable to that achievable with the reference-scan method. Copyright 2004 Wiley-Liss, Inc.