Skip to main content
This site will look better in a browser that supports web standards, but it is accessible to any browser or Internet device.

» System Architecture

The hardware components of an MRI scanner are displayed in this learning tool. Select a component for more information about its function.

« Tools List

System Architecture

Hover your mouse over the hotspots (point (1K)).

Superconducting magnet.
Other Higher field strengths can be obtained with the superconducting design because a larger current is achievable. I have used a superconducting magnet as the example in the diagram, because these are the most common type. Superconducting coils sit in a cryogen bath (liquid helium, at less than 4.2K). A large current passes through the coils to create the magnetic field (it's basically a solenoid). This current draws no power because the coils offer no electrical resistance. The magnet, therefore, is always "on".
…Read more about the magnet.
Surface coil.
Receive chain Surface coils are often used for better signal reception. Although they are placed on or around the surface of a patient, they may be optimised to image deep-body structures. The signal from the RF power amplifier drives the transmit coils such that the desired RF magnetic field is produced. Note that it is not the transmission and reception of radio waves which is used in MRI, but magnetic induction from the oscillating magnetic fields (Hoult. The magnetic resonance myth of radio waves. Concepts in Magnetic Resonance 1: 1-5, 1989).
…Read more about surface coils.
Gradient coils.
Gradient control path The gradient coils are used to spatially encode the positions of the MRI spins by varying the magnetic field linearly across the imaging volume such that the Larmor frequency varies as a function of position.
…Read more about gradient coils.
RF screen (Faraday cage).
Other The RF screen is a copper lining placed in the ceiling, walls and floor of the magnet room (the Inner Controlled Area). It keeps out spurious signals both from foreign transmitters and from the driving electronics of the MR system itself. This is required because the signal returned from the patient is very small.
…Read more about RF screens.
Quad hybrid.
RF transmitter path After RF power amplification, a quad hybrid splits the power into two signals ("real" and "imaginary", or "I" and "Q", but don't worry about that now) which have a 90° phase difference. This is a significant improvement on so-called linearly polarised transmit / receive coils. Quadrature transmit coils require less RF power and are almost twice as efficient.
RF power amplifier.
RF transmitter path The amplitude-modulated RF signal is amplified from approximately 10 mW to the order of 16 kW (peak). The power amplifier is the biggest (and most expensive) part of the transmit chain. It is at this point that the power output is monitored and compared to SAR limits. If they are exceeded, the power monitor will prevent the scan from taking place.
RF modulator.
RF transmitter path The RF modulator receives the harmonic signal with frequency ω0 from the synthesizer and generates a RF pulse from this continuous signal (that is, when to stop and start, and the pulse envelope).
Receive chain The very small MRI signal detected by the surface coil is boosted here.
Quadrature demodulator.
Receive chain The quadrature demodulator converts the signal received from the pre-amplifier from the MHz range down to the kHz range. It mixes the reference signal ωref (from the synthesizer) with the received signal ωrec. The result is a mix of signals at different frequencies, but without distinction between ωrecref and ωrefrec. We are interested in the latter. To solve this problem, it first splits the received signal into the real and imaginary components, and does the mixing with two different reference signals that are 90° out of phase. This maintains information about the frequency spectrum of the signal. The unwanted signal is rejected after digitisation in the ADCs.
Both receive and transmit The synthesizer is the source of the frequency of the RF signal transmitted (the "centre frequency"). It is also used in the demodulation of the received signal to centre around zero, instead of as it is received, around the centre frequency.
…Read more.
Gradient control timing.
Gradient control path The gradient system must be able to switch the direction of the linearly varying magnetic field very quickly to be able to implement some modern pulse sequences (e.g. echo planar imaging). Common high performance gradients can achieve a maximum strength of 30 mTm-1, although higher performance gradients are available. The slew rate is the maximum gradient strength divided by the time it takes to reach that strength (e.g. 150 Tm-1sec-1).
Digital-to-analogue converter.
Gradient control path The digital-to-analogue converter converts digital signals to analogue waveforms.
Pulse sequence control / timing
Receive chain This component is sometimes referred to as the Front End Controller. It receives scan details from the software on the scanner, and controls the magnet, gradients, RF transmitter and receiver, RF coil switches etc. It is part of what is referred to as the spectrometer, which also includes control over the data acquisition of the receive chain and the ADC (control lines not shown). A scan initialisation phase occurs before the scan proper, also controlled by the spectrometer. This includes processes such as the tuning of the synthesizer frequency to the resonant frequency (accounting for drift and the presence of the patient in the scanner), the tuning of the RF coils, adjustment of the RF transmitter voltage (e.g. to give a 90° pulse), adjustment of the receiver gain for the measurement to be made, etc.
Receiver analogue-to-digital converter.
Receive chain The receiver ADC converts the demodulated data from the mixer into a digital array and feeds these data immediately into the reconstructor for complex FFT (fast Fourier transform). The sampling frequency is determined by the receiver bandwidth (in kHz range after demodulation).
Receive chain The reconstructor, or image array processor, performs a 2D or 3D fast Fourier transform (FFT) on the digital k-space data it receives, to create the MR images.
Host CPU.
Other The computer workstation for scanner control.
User input.
Other Patient data is entered into the software by the user, including, crucially, the patient's weight. This information is used to calculate the Specific Absorption Rate for pulse sequences. User also specifies the geometrical parameters, the imaging method and the sequence timing. These are usually stored as preset procedures.
Other Think of this as the hard drive in the control computer. This is where patient data and images are stored, before being networked or archived.
Data storage.
Other Data storage available to back up to removable media.
Visual display unit.
Other The monitor may have magnetic shielding if it is a cathode ray tube ( most TVs) so it does not display distorted images. Liquid crystal displays (LCD) avoid this problem and are more common in recent installations.
Other devices.
Other Images may be sent via hospital networks to image storage systems such as PACS (Picture Archiving and Communication Systems), post-processing workstations, or for film hardcopy.

A larger diagram image is also available.
The diagram is colour-coded as follows:

  •     RF transmitter path
  •     Gradient control path
  •     Receive chain
  •     Both receive and transmit
  •     Other

Further reading on this topic:
Books: MRI From Picture to Proton: 23 pages, Q&A in MRI: 17 pages, Magnetic Resonance Imaging: 45 pages, MRI: Physical Principles & Sequence Design: 33 pages

« Back to Tools list

(This page: