Basic Operation

Quick Reference

  • Signal source results from atoms with unpaired protons.
    • Primarily Hydrogen
    • Also: O, F, Na, and K
  • Magnetic Field Strength: 0.1T - 7T (earth's magnetic field is 6.5x10-5T)
    • Increased field strength increases signal-to-noise ratio but also increases geometric distortions.
  • The precession frequency is given by the Larmor equation
  • Gradient coils are responsible for the loud noise an MRI makes.
  • Signal strength (S) is proportional to square of the magnetic field (B).
Contrast Weighting

Short TR (TR = T1)Long TR (TR > T1)
Long TE (TE = T2)Not clinically usefulT2 weighted image
Short TE (TE < T2)T1 weighted imageProton density image

  • T2 is much smaller than T1 (milliseconds versus seconds).
  • A reliable way to distinguish T1 and T2 weighting is to looks at fluid-filled structures which will appear dark on T1 weighted images and bright on T2 weighted images.

Theory of Operation

Basic Steps in Imaging

  1. A strong magnetic field (0.1T - 7T) is applied to the patient aligning hydrogen atoms (parallel and anti-parallel) with the field.
    • Parallel alignment is only slightly preferred (~3ppm/T). This small fraction will generate the detectable signal for imaging!
  2. Once aligned, the protons precess about their poles at a frequency given by the Larmor Frequency equation.
  3. A resonant radio frequency (RF) pulse of Larmor Frequency is used to excite the aligned atoms.
    • Excitation causes the amplitude of their precession to increase but does not impact the frequency.
  4. After the pulse, the excited atoms return to their lower energy state by emitting RF waves at their Larmor frequency.
  5. Receiving coils detect the emitted RF waves, recording their signal.

Voxel Encoding

The signal the receiving coils detects does not in itself contain any information except for the intensity and frequency of RF return. In order to use that signal to generate an image, one must be able to determine where that signal is coming from. Therefore, the signal is encoded along the three Cartesian axes as outlined below.

1. Slice Selection

The first method of encoding is to selectively excite only 1 slice at a time.

  • A gradient magnetic field is applied in addition to the uniform magnetic field. This causes the Larmor frequency to vary as a function of position along the direction of the field gradient.
  • Slices of variable width may be selected by changing the RF pulse frequency and bandwidth.
2. Phase Encoding

A temporary small gradient is briefly superimposed in a direction perpendicular to the slice selecting direction. This speeds the atoms on one side of the slice while slowing atoms on the other side. When this gradient is shut off, all the atoms return to their original precession frequency but they are now at different phases. At this point, both slice and row can be determined.

3. Frequency Encoding

The final step again uses gradient coils to create a magnetic field gradient along the axis perpendicular to both the slice selection axis and the phase encoding axis. This gradient is left on during readout.

Image Reconstruction

It is important to realize that, although each voxel’s location is encoded by phase and frequency, all voxels in a row are read out simultaneously. This data must be deconvolved by an Inverse Fourier Transform in K-space. K-space is an image that maps frequency rather than physical x-y coordinates. Once this is complete, the information may be used to generate an image.

For k-space maps, pixels near the center correspond to low-frequency data while pixels near the edge correspond to high frequency (edges).

Frequency data is collected in k-space and converted to an image using the inverse Fourier Transform.
K-space is a map of the frequency of MR signal. Higher k frequency at the edges of a k-space map represent sharp edges.

Source of Tissue Contrast (T1 and T2 relaxation)

There are two sources of signal that can be imaged. First, changes to the number of protons aligned parallel and anti-parallel to the magnetic field called longitudinal magnetization. Longitudinal magnetization rate is characterized by T1 time.

T1 (Longitudinal relaxation): This is the time it takes for the bulk magnetization to regrow to 63% of its equilibrium value. T1 relaxation is sometimes called spin-lattice relaxation because it is the time that it takes for the protons to realign with the magnetic field.

The second method is to measure the component of their magnetization that is orthogonal to the direction of the magnetic field (referred to as transverse magnetization). Normally the phases of precession are randomly distributed and the total transverse magnetization is zero. But immediately after excitation the phases become aligned and a signal can be detected. Random magnetic disturbances rapidly decay this signal in a precess called Free Induction Decay (FID). Rate of FID is characterized by the T2 time.

T2 (Transverse Relaxation): This is the time required after excitation for the transverse magnetization to decay to 37% of its maximum value.

Scan Parameters

Repetition time (TR): The time between excitation RF pulses. During this time, both T1 and T2 relaxation occur to varying degrees.

Echo time (TE): The time between excitation and signal acquisition.

Choice of TR and TE is driven by the T1 and T2 times of the tissues being distinguished. Timing is chosen to maximize the difference in signal intensity. For instance, if two tissues have very different T1 times we might use a TR time approximately equal to the shorter of the two tissues T1 time and a very short TE time. Reciprocally, if the two tissues have similar T1 times but differences in T2, we might use a long TR time to allow full recovery of longitudinal magnetization in both tissues but then choose a TE time that maximizes the difference in T2 decay.

Key Point: What is an echo?
An echo in MRI is similar to a sound echo in that it is a return of the FID signal at some time after the initial FID. The echo is generated by first using a 90-degree flip angle then, at time TE/2 employing a 180-degree flip angle. Echo imaging allows for rapid acquisition (multiple echoes) and fine control of contrast weighting.

Flip Angle: The angle by which the net magnetization is directed away from the magnetic field. Flip angle can be adjusted by changing the amount of energy transmitted by the RF pulse. Common flip angles are 90 degrees and 180 degrees. Flip angles are used to adjust contrast and to create echo time based weighting, such as Fluid Attenuation Inversion Recovery (FLAIR).

Time of Inversion (TI): Inversion recovery sequences operate by initially inverting the magnetization with a 180-degree pulse, then allowing some degree of T1 recovery prior to emitting a 90-degree pulse to yield contrast. TI is the time between this initial 180-degree pulse and the subsequent 90-degree pulse.

Key Point: TR controls the T1 weighting of the image. TE controls the T2 weighting of the image.

Scanner Components

Primary Magnet: Creates the static magnetic field (0.1 - 7T).

Gradient Coils: Creates a linearly varying magnetic field (order of mT) used for spatial encoding. Also used to produce contrast in diffusion/flow imaging. Gradient coils are responsible for the noise in an MRI.

RF Coils: Sends and receives the radio frequency signals (order of μT) used to excite hydrogen atoms to their Larmor frequency. Many different coil designs exist for imaging specific body locations.

Shim Coils: Magnetic structures that serve to homogenize the magnetic field of the primary magnet.

Helium Cooling System: The magnets in a modern MRI scanner must be superconducting and require extremely low temperatures (~3 Kelvin). Liquid hydrogen is used to maintain these cool temperatures.

Key Point: Quenching
In the event of emergency (over-pressure or when the magnetic field is endangering life), the liquid helium can be dumped (flash-boiled off). This process is known as quenching. Quenching will demagnetize the machine but will also damage the magnet.

Illustration of major MRI components.
Patient with full body RF coils.

Contrast and Weighting Schemes

T1 Weighting

T1 weighting attempts to maximize differences in T1 relaxation while minimizing the impact of T2 relaxation. This is accomplished by using a short TR and TE [i.e. TR approximately equal to T1 (500ms) while TE is much shorter than T2 (<15ms)]. T1 weighting offers good tissue contrast but sees less use in radiation therapy than T2 due to limited tumor contrast. When used in radiation therapy planning, T1 weighting is most often used to asses nodal invasion. The most common T1 weighted sequence is Spoiled Gradient Echo.

T2 Weighting

T2 contrast offers higher soft tissue contrast than proton density weighting or T1 weighting. For this reason, T2 weighting is the most widely used weighting scheme in radiation therapy planning. T2 weighting is achieved using a long TR (>2000ms) which allows near full T1 recovery while long TE times (=T2) creates contrast. These long TR times and low signal intensity per excitation make a T2 weighted scan slow compared with T1. The most common T2 weighted sequence is Fast Spin Echo.

Key Point: Cancerous tissue tends to have longer T2 times making tumors appear bright on T2 weighted images, especially when surrounded by edema. This is not always true, see image for examples of bright and dark tumors, but is a good rule of thumb.

Proton Density Weighting

Contrast in a proton density weighted MRI is derived from differences in the number of protons within a voxel. Because most soft tissue has similar proton density, proton density weighting is a low-contrast imaging technique but finds use in evaluation of menisci and brain structures. Proton density weighting is achieved using long TR times (>2000ms) to allow full T1 recovery and very short TE times (< T2) to minimize T2 contrast.

Illustration of Te and TR for proton weighted MRI contrast

Fluid Attenuated Inversion Recovery (FLAIR)

Like all inversion recovery sequences, FLAIR begins with a 180-degree RF pulse to invert the magnetic moment of the effected hydrogen. The 90-degree excitation pulse is timed to coincide with fluid T1 recovery crossing 0 magnetic moment thereby suppressing its signal. FLAIR is most commonly used to suppress cerebrospinal fluid (CSF) in brain scans which can obscure structural information of T2 weighted brain scans. FLAIR finds significant use in distinguishing cerebral edema. This is because cerebral edema appears bright in both T2 weighted and FLAIR scans but FLAIR suppresses CSF, making identification easier.

Diffusion Weighting

Diffusion weighted images register Bownian motion of individual water molecules. This is extremely useful in detecting necrotic regions of a tumor as necrosis appears brighter than the surrounding tissue. Diffusion weighted images may be referred to as an Apparent Diffusion Coefficient (ADC) map.

Contrast Agents

Gadolinium Contrast

Gadolinium is the most common contrast. Gadolinium works by shortening the T1 time in the tissues that absorb it. This causes tissues that absorb gadolinium to appear bright in T1 weighted images.

Gadolinium contrast enhances glioma in brain T1 weighted MRI imaging.

Paramagnetic Iron Oxide

Clinically referred to as SPIO (Super Paramagnetic Iron Oxide) or USPIO (Ultra Small Paramagnetic Iron Oxide), Paramagnetic Iron Oxide primarily impacts T2 times and produces a signal drop in T2* weighted images.  A normal liver will absorb SPIO but a metastasized liver will not. This will cause the normal liver to appear dark and the metastasized liver to appear bright.

SPIO darkens normal liver but leaves metastasis bright in T2 weighted image.

MRI Artifacts

MRI artifacts arise from a variety of sources including MR hardware, room shielding, patient motion, tissue heterogeneity, foreign bodies, sampling resolution and k-space errors. Ability to identify and explain basic MRI artifacts is important because it allows one to distinguish between anatomy and artifact. Further, understanding the source of an artifact makes eliminating or reducing the clinical impact of an artifact possible.

Aliasing (Wrap Around) Artifact

  • Anatomy outside of the field-of-view (FOV) is superimposed on the opposite side of the image.
  • Aliasing is most commonly seen in the phase encoding direction.
  • Sampling at below the Nyquist frequency (< twice the RF frequency of the voxel) causes the apparent frequency to be lowered in a process known as aliasing.
Artifact Reduction
  • Many scanners have automatic wrap around removal either by oversampling or by applying a low pass filter prior to analog-to-digital conversion.
  • Change phase encode direction.
  • Increase field-of-view.

Chemical Shift

  • Dark or light edges on borders of structures with different chemical compositions.


All chemical shift artifacts occur because of differences in the intrinsic shielding of bodily chemical structures. This causes apparent displacement in either the frequency or phase encoding directions.

  • Frequency encoding
    • Occurs as a direct result of small changes (3-4 ppm) in precession frequency resulting in signal being mapped to a different location in the frequency encoding direction.
    • Appears as displacement with a light edge on one side (overlap) and a dark edge on the other (signal void).
  • Destructive phase shifts
    • Occurs in gradient echo imaging as the result of constructive and destructive interference in dephasing and rephasing the echo.
    • Appears as dark border around the entire structure.
Artifact Reduction
  • Use higher bandwidth.
  • Use lower field strength scanner.
  • Avoid gradient echo imaging or adjust gradient echo timing.

Key Point: Chemical shift magnitude is proportional to field strength as governed by the Larmor frequency.

Geometric Distortion

  • Appears as physical distortion of the image.
  • May not be readily apparent in tissue but can be evaluated using grid phantom.
  • Geometric distortion may be the result of non-uniformity of the magnetic field leading to inaccurate encoding the spatial position.
  • May be caused by the machine itself or by areas of variable susceptibility within the patient (i.e. implanted devices).
Artifact Reduction
  • Use lower field strength scanner.
  • Place regions of interest near field center which is area of lowest distortion.
  • Regular quality control to assure proper shimming.

Key Point: A major challenge in using MRI for radiation therapy treatment planning is geometric distortion.

Herringbone Artifact

  • A repeating pattern superimposed over the image.
  • Caused by a bad pixel in k-space image usually due to a hardware fault.
Artifact Reduction
  • Repeat scan.
  • If artifact persists, scanner repair will be required.

Magnetic Susceptibility Artifact

  • Most often appears as darkening (signal dropout) in a portion of the image.
  • May also appear as brightening or spatial distortion.
  • Results from drastic changes to the local magnetic field arising from inclusion of ferromagnetic, paramagnetic, or diamagnetic materials in the scan.
  • Ferromagnetic materials (iron, nickel, etc.) cause strong artifacts like signal drop out.
  • Paramagnetic and diamagnetic materials (those weakly attracted to or repelled by magnetic fields respectively) cause lesser artifacts.
Artifact Reduction
  • Remove metal from scan.
  • Use short echo time image sequences.

Moiré Fringes Artifact

  • Repeated irregular bands of light and dark.
  • Signals from different phases are superimposed yielding constructive and destructive interference of signal intensity.
  • Non-uniformity of magnetic field.
Artifact Reduction
  • Improve field uniformity by shimming.

Motion Artifact

  • Faint copies of the image (known as ghosts) are superimposed on the image.
  • Ghosting typically occurs along the phase encoding direction.
  • Motion (including involuntary motion and blood flow) causes anatomy to be imaged at different locations.
Artifact Reduction
  • Reduce patient motion
  • Choose faster acquisition sequences
  • Use PROPELLER acquisition sequences

Truncation (Ringing) Artifact

  • Fine parallel lines adjacent to a high contrast interface.
  • Can appear as an anatomical structure, especially in the spinal cord.
  • Caused by the inverse Fourier transform used to convert k-space map into image.
  • A step gradient (or other very sharp edge) requires a very large (nearly infinite) number of varied frequency sin waves to construct. Practically, only a finite number of frequencies can be sampled causing an oscillation in the resulting image.
  • Since the Fourier series was cut short (truncated), the artifact is referred to as truncation.
Artifact Reduction
  • K-space post-processing can reduce Truncation artifacts.

Zipper Artifact

  • Interference across one or more rows, most commonly in the phase encode direction.
  • Most commonly caused for feed-through in which an outside RF signal is picked up by the receiver system.
    • Phase encode direction.
  • May also arise from stimulated echoes resultant of poor slice selection profiles or poorly adjusted RT-transmitters.
    • Frequency encode direction.
Artifact Reduction
  • Assure room door is closed for complete RF shielding.
  • Move patient monitoring system (e.g. anesthesia equipment) as far away from the scanner as possible.

Knowledge Test

1. How much larger is a 1.5T MRI scanner's magentic field than the earths magnetic field?

Question 1 of 3

2. How might the below artifact be reduced?

Question 2 of 3

3. What type of weighting was used in the below MRI image?

Question 3 of 3


Not a Premium Member?

Sign up today to get access to hundreds of ABR style practice questions.