Principles of MRI & Instrumentation

In MRI, the magnetic field strength applied in space is varied to identify the spatial origin of the emitted signal.
By changing the magnetic gradients(i.e. its magnitude) along the x, y and z directions, the positions along the relevant axis could be computed by measuring the emission frequency of photons.
Frequency is measured indirectly by measuring the strength of the FID- pin echo as a function of time.
Protons at different locations emitting at different frequencies contribute to the measured signal.
Fourier Transformation is used to separate out the total signal and the frequencies and relative strengths of various sources are identified.

The above diagram indicates how signals could be separated out from the three sources, each of which contains Hydrogen atoms using Fourier transformation, the actual integral equation of which is too complex for any discussions here.
Readout gradient is the jargon referring to the gradient that causes the spatially variant frequencies during signal detection.
The technique, Selective irradiation determines the slice thickness.
Having applied a gradient along z, if the 90-degree resonance pulse contains a narrow range of frequencies, only a small region along z is resonating.
Only protons in this excited plane contribute to the image. Changing the irradiation frequency adjusts the slice position.
Projections at different angles are obtained by rotating the readout gradient.
Image could be reconstructed using standard Computer Tomography algorithms.

New Method

Phase encoding, or spin-warp is used nowadays for image reconstruction.
To obtain an image with resolution n along the x-axis, n separate phase-encoded projections is needed.
TR, the waiting time between projections is longer than the time needed for obtaining spin echoes. Selective irradiation allows obtaining projections from other planes during TR. A Multi-slice image could be obtained like the following 15-sections image through the head obtained in 6.5 mins with 1.5s TR.

Any axis can perform any of the three gradient functions allowing for direct accumulation of images is sagittal and coronal orientations as well as the transaxial view obtainable from CT scans.

A magnet large enough for human whole-body imaging is required as well as magnetic field uniformity and stability.
Solenoidal superconducting magnet is used in high-quality imaging devices.
Its electrical windings are cooled at liquid helium temperature (c. 4Kelvins) to make electrical resistance negligence. Hence, once a current is established to set up the desired magnetic field, no additional power is needed to maintain it.
For field strength below 0.25 Tesla, resistive magnets could be used, which is cheaper initially, but requires electrical power more expensive than the cost of liquid helium used for cooling the superconducting magnet.
The determinant of a MRI imager's quality is the gradient system. The coils for the x, y and z directions magnetic fields are placed inside the magnet's bore. They are powered by audio amplifiers of very high quality and power.

Eddy currents is produced the rapid changing magnetic fields. The gradient power amplifiers have to shape waveforms compensating for these eddy currents.
Radio-frequency(RF) energy is transmitted to and received from the patient by a RF coil around the patient.

A common RF system involves a cross-coil system where the separated transmitted and receiver coils are orthogonal to each other, which could reduce direct coupling between them.

Computers are used to control the gradient fields and RF pulses with precise timing and accurate amplitude, whose patterns is known as the sequence.
Computers must also be able to be reprogrammed for different imaging protocols like that for transaxial and saggital images.
Computer Storage is also crucial as well as processor powers since many images are needed to be stored in a short time (e.g. 40 images in 6.5 minutes for a head scan)
An array processor (a computer performing many identical arithmetic calculations in parallel) is needed for data reconstruction.
As the reconstruction algorithm(Fourier transformation) is two-dimensional, no special-purpose hard-wired back projector is required as in CT.
Changing acquisition parameters affects contrasts, and colour is useful in MRI due to its wide latitude in contrast.
Images of calculated parameters like T1, T2 and H could be produced to supplement the acquired images. These fundamental images allow calculation of TE and TR values not actually acquired.

It is critical for data analysis computing not to interfere with the data acquisition for the next patient to be diagnosed; hence it is sensible for separate consoles with different functions- physician display console, data acquisition and image reconstruction.