In which ways is a stronger magnet better for magnetic resonance imaging? I read that:

The field strength of the magnet will influence the quality of the MR image regarding chemical shift artifacts, the signal to noise ratio (SNR), motion sensitivity and susceptibility artifacts.

but I don't understand why. I suppose the dephasing is quicker, reducing the transverse net magnetization more quickly. Does this go for the longitudinal net magnetization too? How would this reduce motion sensitivity and SNR?

  • $\begingroup$ this is not a a bad question for biology - i was just answering it when you closed it. Mark the v short answer is that stronger magnetic fields allow the machine to distinguish between the biological state of the atoms they are reading. lots of biologists use MRI and other physical methods to investigate biology. where do all those protein structures come frome? $\endgroup$
    – shigeta
    Mar 3, 2012 at 15:03
  • $\begingroup$ BTW Mark, If you want to reestablish control as the poster of the question create an account here on Physics.SE and link it with your Biology account. And welcome aboard. $\endgroup$ Mar 3, 2012 at 17:56

2 Answers 2


The magnetic moment of the nuclei being read by the MRI instrument does have a stronger coherence when the magnet is stronger - the energy difference between the excited and ground state of the atom's nuclear spin is larger.

This does lengthen the relaxation time as you say, but the really important effect is that the resolution of the MRI instrument becomes greater. When differentiating between nuclear spins of protons in cancer tissue vs healthy tissues is very small - but fortunately it is there. Most of the medical applications need as much resolution as they can get because there is a massive amount of protons in water or phosophorus the machine has to differentiate over - most if the 'signal' is miniscule compared to native state biological atoms the MRI is also seeing.

@DrSAR I'm sort of confused by your comment. Not sure what you are saying.

1) MRI is one of the more common ways to identify cancer. Maybe if you could be more explicit in why you do not believe NMR (I use this term interchangably with MRI) can differentiate physiological state of tissues? This has been pretty well discussed since the 1980s.

this quote passed wikipedia's editorial process okay:

MRI provides good contrast between the different soft tissues of the body, which makes it especially useful in imaging the brain, muscles, the heart, and cancers compared with other medical imaging techniques such as computed tomography (CT) or X-rays. Unlike CT scans or traditional X-rays, MRI does not use ionizing radiation.

2) MRI does not need protons, any nucleus with a spin =1/2 is a decent target for an experiment. again P31 is one of the more popular targets because it has a high natural abundance, but O17 and C13 are also accessible through most machines (and their high intensity magnetic fields) today. More exotic spins (3/2 5/2 etc) can also be used, but the sensitivity and analysis of the returned signal are more complicated. So many different kinds of atoms can and are used in MRI experiments.

  • 1
    $\begingroup$ Sorry, but there are multiple mistakes or at least misleading statements: differentiating cancer from healthy tissues based on either relaxation times or spectral resolution (not clear which you propose) is not that easy at best, impossible most of the time. There are also no protons in phosphorus and atoms are not classified based on being biological or not. $\endgroup$
    – DrSAR
    Oct 24, 2012 at 7:47
  • $\begingroup$ The main magnetic field of an MRI machine does NOT result in higher resolution. The resolution is given by the gradient field strengths and the time for how long you can read out the signal before it decays (mainly $T2^*$). Hence, larger $B_0$ does mean more signal, but also mostly shorter $T2^*$. While it is true, that today the highest resolution is achievable at the highest fields, this is a secondary effect resulting from the higher SNR. So it is indirectly coupled to $B_0$. $\endgroup$
    – M529
    Jan 23, 2021 at 21:01
  • $\begingroup$ the atomic nuclei scanned by MR dont have to be hydrogen and not all hydrogens respond to MR. MR reads on the so called nuclear magnetic moment (spin). This is strongest when the spin is 1/2 as it is for 1H 31P 19F 13C. theoretically with a strong enough magnet you can create a resonance with any nucleus. Medical MR for iron has been done as well as carbon and phosphorus. $\endgroup$
    – shigeta
    Jan 29, 2021 at 17:22
  • $\begingroup$ I refer any curious persons to see the early work on Cancer and relaxation times of water to this 1975 paper "Nuclear Magnetic Resonance Investigations of Human Neoplastic and Abnormal Nonneoplastic Tissues" or the tens of thousands of papers that have come after. $\endgroup$
    – shigeta
    Jan 29, 2021 at 17:23

The strength of the main magnetic field of the magnet, $B_0$ changes lots of things in MRI. The above answer is somewhat incomplete. Briefly, and I hope vaguely understandably:

  1. In almost all circumstances, the underlying signal is proportional to the nuclear polarisation, $P$, of the spin-non-zero matter placed within the scanner. This is a form of paramagnetism and arises due to thermodynamics: there's an energy of interaction between the nuclear spins and $B_0$ and yet an entropic cost to having all spins aligned in the same state. The nuclear polarisation $P$ is easily determined as $P=\tanh\left(\frac{\hbar \omega}{2k_B T}\right)$, where $\hbar\omega$ is the energy of the photon that needs to be absorbed for magnetic resonance to happen, $k_B$ is Boltzmann's constant, and $T$ is temperature -- usually about 300 K for living things. What is $\omega$? Well, it can be said that the condition for resonance is the Larmor equation, $\omega=\gamma B$, where $\gamma$ is a constant known as the gyromagnetic ratio (constant for a given nucleus -- about 40 MHz/T for protons, for example). So, all other things being equal as it turns out that $\hbar \omega << k_B T$ at 300 K, the argument of $P$ is small and in fact it can be Taylor expanded to good accuracy and shown that $P\propto B$. So, to a very good approximation, increasing the magnetic field strength of the scanner increases the polarisation you have.
  2. That underlying polarisation generates a magnetisation, $M$, which is proportional to it multiplied by the concentration of the nuceli in question. If the scanner "does stuff" -- plays a sequence of RF pulses and magnetic field gradients that constitute running a scan -- $M$ moves in time accordingly, as determined by the Bloch equations. We can only detect if $M$ changes over time, and we do so via a loop of wire (known as an RF coil) placed close to the sample/patient. That loop of wire also detects thermal noise from the sample (in fact, as $\propto \int \sigma |\mathbf{E}|^2 dV$) and it transpires that combining everything together yields a behaviour that scales as $\omega M_0$ or, all other things being equal, to $B_{0}^{2}$.
  3. Separately to this, the nuclear decoherence times $T_1$ and $T_2$ that vary greatly between tissues and give MRI its high contrast-to-noise ratio differ as a function of field. The behaviour of these with field is more complicated but can be briefly understood by thinking that a common source of relaxation is molecular tumbling: if the thermal motion of molecules with a dipole moment "matches up" with the Larmor frequency $\omega=\gamma B$, then it will promote transitions towards thermal equilibrium. So, as fields change, what happens to $T_1$ in particular and $T_2$ depends on the molecule in question (and its diffusion coefficient).
  4. Because of Maxwell's equations, at an interface between two different materials of different electric properties, a discontinuity in the magnetic field occurs in one of its components. This leads to a further source of relaxation in MRI, $T_{2}^{*}$ relaxation. This gets worse at high field -- i.e. static field inhomogeneities are greater.
  5. The wavelength of $\omega$ decreases at higher fields. This means that it is harder to generate a uniform RF field over a person. At 3T this isn't a problem, as the wavelength is much longer than a person, but at 7T and above it's a big issue and can cause signal voids / nonuniformities if not dealt with.
  6. The resolution of an MRI scan is determined by how far out in (Fourier) $k$-space you can sample and obtain signal (and not noise). Magnetic field gradients are played by the scanner in order to physically do this; at higher fields further dispersion is required to reach the same place and thus gradients have to be stronger. This isn't in practice a problem as usually, for human applications, SNR is limiting, or the physiological limits of humans are limiting the rate at which gradients can be turned on or off.
  7. As $\omega=\gamma B$, nuclei in different molecular environments will experience a greater absolute difference of frequencies at higher fields. This means that the NMR spectral resolution increases -- great for chemists, who want to resolve high-resolution spectra of liquids in tubes. In a person, imaging protons, the largest two chemical shifts that are resolvable are between "water" (at 4.4 ppm) and "fat" (at ~1.3 ppm). This causes an artefact in conventional proton imaging (known as chemical shift artefact) but is of obvious utility in MR spectroscopy, where it is desired to record spectroscopic information from living systems from spin $\neq0$ nuclei in vivo. Higher fields are usually absolutely required to do this well.
  8. Finally, because the frequency of the RF increases, the degree of RF-induced heating increases and MR safety concerns become more prevalent at higher fields. The electrodynamic environment changes quite considerably, and so medical devices with conductive parts are typically only certified as safe at well-defined, well-measured frequencies corresponding (usually) to 1.5 T and/or 3 T. Research MRI scanners operate at many other field strengths, from low field scanners (~0.4T) to ultrahigh fields (above 11.7 T).

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.