Spatial Encoding

How does the scanner know the locations of all the MR signals? 
A typical MR scan may contain over 4 million voxels, each generating its own tiny MR signal. With such a huge array to analyze, how does the scanner possibly sort out where each signal is coming from? 

The answer to this question is not simple, and several dozen Q&A's in the next few tabs will be devoted to exploring various imaging strategies. In brief, the following methods are employed, often in combination:
  1. Differences in frequency. Magnetic field gradients are applied that alter the main magnetic field in a spatially specific pattern, changing the local resonance frequency by a few hundred to a few thousand Hz. Frequency localization methods are commonly used for: a) slice selection in 2D imaging; and b) spatial position encoding within a given slice. 
  2. Differences in phase. Applying an imaging gradient results in a gain or loss in phase that persists even after the gradient has been turned off. Phase localization methods are used for: a) slice selection in 3D imaging; and b) spatial position encoding within a given slice.
  3. Differences in signal timing. Pulse sequences often have considerable amounts of "wait time" after signal collection and before the next RF-excitation. This provides a window for stimulating other slices or parts of the imaged volume and collecting signals at different time offsets. 2D-multislice imaging uses this method.
  4. Distance from receiver coils. Proximity to surface receiver coils provides information about a voxel's position. Coils near a voxel record stronger signals than coils farther away. This difference in signal intensity among receiver coils can be used as an aid to signal localization. This is the principle of parallel imaging, the subject of multiple later Q&A's. 

Advanced Discussion (show/hide)»

References
     Elster AD, Burdette JH. Questions and Answers in MRI, 2nd ed. St. Louis: Mosby, 2001;pp 72-101
Related Questions
     How does frequency-encoding work?