A magnetic 
resonance (MR) imaging apparatus and technique exploits spatial information inherent in a 
surface coil array to increase MR 
image acquisition speed, resolution and / or 
field of view. Magnetic 
resonance response signals are acquired simultaneously in the component coils of the array and, using an autocalibration procedure, are formed into two or more signals to fill a corresponding number of lines in the 
signal measurement 
data matrix. In a Fourier embodiment, lines of the k-space matrix required for image production are formed using a set of separate, preferably linear combinations of the component coil signals to substitute for spatial modulations normally produced by phase encoding gradients. One or a few additional gradients are applied to acquire autocalibration (ACS) signals extending elsewhere in the 
data space, and the measured signals are fitted to the ACS signals to develop weights or coefficients for filling additional lines of the matrix from each measurement set. The ACS lines may be taken offset from or in a different orientation than the measured signals, for example, between or across the measured lines. Furthermore, they may be acquired at different positions in k-space, may be performed at times before, during or after the principal imaging sequence, and may be selectively acquired to optimized the fitting for a particular tissue region or feature size. The 
in vivo fitting procedure is readily automated or implemented in hardware, and produces an enhancement of image speed and / or quality even in highly heterogeneous tissue. A dedicated coil 
assembly automatically performs the calibration procedure and applies it to measured lines to produce multiple correctly spaced output signals. One application of the internal calibration technique to a subencoding imaging process applies the ACS in the 
central region of a sparse set of measured signals to quickly form a full FOV 
low resolution image. The full FOV image is then used to determine 
coil sensitivity related information and dealias folded images produced from the sparse set.