MRI - K-space & Image-space

Discussion in 'Physics' started by blah2222, Apr 18, 2014.

  1. blah2222

    Thread Starter Well-Known Member

    May 3, 2010

    Not sure if this is the best medium to seek answers for a MRI physics, but heck, I'll try.

    Unfortunately, I am having trouble grasping the idea of K-space.

    From what I have understood, the echo of the MR signal is measured while the frequency encoding gradient is on. This whole process (including the 90deg RF pulse, 180deg pulse, and slice-select) is repeated for every phase encoding gradient increment.

    A given row in K-space then corresponds to an array of samples that is a digitized version of the MR signal measured during the frequency encoding stage.

    What confuses me is when the sources that I have read state that K-space is a collection of raw complex data. How can it be complex if each row is a sampled real MR signal? Shouldn't the image be complex because you take the 2D-FFT of the real K-space to obtain an image?

  2. ramueller11

    New Member

    Dec 13, 2013
    I'm not familiar with the explicit details processing of MRI but I have a bit of experience with NMR.

    That said, NMR signals (MRI is an application of NMR) are measured in quadrature, i.e. the transverse magnetization is measured in two orthogonal directions one "in-phase" and the other "anti-phase" with respect the spectrometer reference signal. This is commonly called the 'X' and 'Y' directions in the rotating coordinate system commonly used in NMR. Therefore, your assumption of 'real MR signals' is not correct, when in fact the X and Y components are recorded in the real and imaginary components, respectfully.

    I'm not sure, but I suspect this phase information is discarded when an image is produced to show the magnitude of the complex values.