I want to pursue rads later on, and for the life of me MRI must be the most complex thing I've ever tried to look at. I cant wrap my mind around it.
Can somebody give me a general technical overview of how MRI works? How much of that am I going to have to know as a radiologist?
thanks
spinning protons 
here are some links to sites that explain - from very techinal to for the rest of us - including some animations.
MRI Simply Physics http://www.simplyphysics.com/MAIN.HTM
PPT presentation MRI http://rad.usuhs.mil/rad/handouts/fletcher…cher/sld001.htm
MRI E-Text http://www.cis.rit.edu/htbooks/mri/inside.htm
MRI Tutor http://www.mritutor.org/mritutor/index.html
How Stuff Works: MRI http://electronics.howstuffworks.com/mri.htm/printable
LisaS is right, its all about nuclear spin physics. Protons have an intrinsic nuclear magnetic spin vector and angular momentum that responds to a magnetic field. In MRI, hydrogen protons from water are imaged most commonly, although technically you can image any nucleus. Hydrogen protons “spin” in a 3D frame of reference and the 3D orientation can be changed by applying time-variant magnetic fields superimposed on top of a static magnetic field, which is usually 1.5T or 3T for human clinical scanners.
MRI manipulates the magnetic fields created by spinning charged particles and then records changes in the magnetic field that are emitted from the tissue sample.
When a static magnetic field is applied to a sample of tissue, a small minority of spin vectors (proportion is determined by intrinsic properties of the material as well as the static magnet field strength) line up colinear to the field. Keep in mind that we are talking on the order of only 1 in a million spins excess, so not every proton spin vector lines up in the field. Most clinical MR scanners are 1.5T, but they can vary from 0.5T for open MR scanners to an experimental 7T human clinical scanner that the NIH is developing in combination with Johns Hopkins. Small animal scanners can go all the way up to 15T.
As you increase the field strength, the signal from the sample increases because a 15T static field causes a larger proportion of hte spins to line up in the field than a 1.5T magnet does. Therefore, image quality in general (signal to noise ratio specifically) is better in higher field strength scanners. The problem with high field scanners is that they are very hard to design. Its complicated, but the end result is that magnetic field homogeneity suffers as field strength is increased.
Basically, there are 2 core components to MRI: RF pulses and gradient fields.
RF pulses in a static magnetic field environment cause the nuclear spins to change their 3D orientation. If the static field is applied in the vertical (+z) direction, then a 90 degree RF pulse applied along the +z axis will cause all spin vectors that are congruent to the +z axis to “flip” into the xy plane.
RF pulses by themselves dont give any spatial information. Gradient fields, which are just small position-dependent magnetic fields are created by 3 coils (one for each axis) in the MR scanner. With gradient fields, you can “encode” the spin information into a 2D or 3D image by causing the spins to lose coherence with each other in a position-dependent manner.
An RF pulse applied simultaneously with a gradient field will “excite” only one portion of the tissue. For example, if I use an RF pulse with a +z gradient (assume z is vertical axis) then only a vertical slice of the tissue will be excited, due to the fact that hydrogen protons precess at 42.5 MHz/T and the gradient field ensures that only that particular slice of the object is in resonance during the time of the RF pulse.
When you change the gradient fields, you can generate a multislice image which is the most common setup. The precessing spin vectors from the tissue sample cause current to flow in a receive coil (could be a complex birdcage head coil, simple loop surface coil, pelvic phased array, or custom coil) by Biot-Savart’s Law. After analog anti-aliasing prefiltering and A/D conversion, the computer does an inverse fourier transform. The IFT converts the frequency data to a spatial map of pixel intensities which is the image you look at.
There are basically two kinds of contrast in MRI. One is T1 weighted, which refers to longitudnal relaxation time. After hte RF pulse is applied and the spins are flipped into the transverse plane (xy), they will try to recover to the original 3D vector orientation due to the presence of the 1.5T static field. You can arrange RF pulses and gradients such that the contrast in the image corresponds to different tissues that take varying amounts of time to return to the pre-RF condition. For example, CSF in the brain has a T1 value of 2200 ms, whereas white matter = 500 ms, and gray matter = 750.
T2 contrast involves transverse magnetization relaxation. When the spins are “flipped” into the xy plane, they begin to relax both in the xy planes and the z axis. The xy plane component of the spin vectors gradually loses coherence with other spin vectors in the same area. This information can be encoded to produce a T2 weighted image. Typically, T1 weighting is used for basic anatomical outlines with poor intra-organ contrast (i.e. trauma cases) whereas T2 weighted images are better at picking up organic disease states (Alzheimers, cancer, etc).
As a radiologist, to work with MRI you need to understand the basics of how an image is created, but you dont need to know it to the level of detail on those websites. As a clinical radiologist, you need to know which kind of scan to request for different clinical scenarios. For example, if you are imaging the prostate, you want a T2 weighted, not a T1 weighted image. Interventional radiologists use MRI in conjunction with procedures to guide things like needle biopsy, drug injection, or cardiac stent placement.
Let me know if you have any questions. This stuff can very difficult to conceptualize. Nowadays MRI involves everything from MR angiography to monitoring the migration of magnetically labeled stem cells. Its an extremely diverse area.
Howdy y’all!
While I normally respect MD/PHDs opinion, here he’s clearly wrong.
MRI is, fundamentally… magic. Radsdoc got it right to begin with.
Sorry, I’m still in my post-Step I refusal to acknowledge basic science. Give me a couple more months and I’m sure I’ll be back to normal. In the meantime, MRI stands for Magically Reproducable Images and the mechanism of action of all known drugs involves a magic wand.
Take care,
Jeff “I’ll shoot the next person who mentions intracellular signaling” Jarvis
MS-III
UTMB