Abstract: In a method for magnetic resonance imaging, k-space is sampled using density weighted data acquisition or acquisition weighted data acquisition, with at least one region of k-space being intentionally undersampled with respect to the Nyquist criterion. Reconstruction of the MR image is performed using a parallel imaging method, with the influence in the reconstructed image of the missing k-space data, as a result of the undersampling, being reduced.

Abstract: In a method for magnetic resonance imaging, k-space is sampled using density weighted data acquisition or acquisition weighted data acquisition, with at least one region of k-space being intentionally undersampled with respect to the Nyquist criterion. Reconstruction of the MR image is performed using a parallel imaging method, with the influence in the reconstructed image of the missing k-space data, as a result of the undersampling, being reduced.

Abstract: In order to acquire data for an image representation which shows the spatial distribution of the magnetic resonance behavior of an object within a selected localized region, the relevant localized region is arranged in a stationary magnetic field and is subjected to a sequence of high frequency pulses and pulses of magnetic field gradients in different spatial directions, such that a succession of magnetic resonance signals appears. These signals are read out by scanning under the influence of a read gradient and are used as sets of data for the filling of the K-space. In accordance with the invention, at least two different sequences of high frequency pulses and magnetic field gradient pulses are used, which differ in at least one of the features of the signal creation responsible for different aspects of the image quality. These sequences are executed one after another in time, with the magnetic resonance signals read out for the different sequences being collated in different bands of the K-space.