K-space as encoded spatial information

K-space stores complex measurements of spatial-frequency content before Fourier reconstruction. Its coordinates describe frequencies across space, rather than anatomical positions. Each measurement contributes a pattern extending across the image, and the reconstructed anatomy emerges from combining those patterns with their measured amplitudes and phases.

Central k-space contains low spatial frequencies and strongly influences overall signal and contrast. Peripheral data provide higher spatial frequencies needed for sharp boundaries and fine detail. Removing peripheral data therefore blurs edges; corrupting a single point can produce stripes across the image.

For simple Cartesian imaging, one phase-encoding step acquires one k-space line. With one line acquired per TR, scan time is approximately TR × phase-encoding steps × averages. This timing expression needs modification for echo trains, parallel imaging and other accelerated schemes. A larger matrix changes sampling; it does not turn k-space into a one-point-per-pixel anatomical map.