MRI physics you operate instead of read
Twenty-one units from a single hydrogen nucleus to a line of a real scanner protocol. Each unit asks one question and hands you a working instrument to answer it. Nothing moves forward until you have made it move.
Why another MRI explainer
Most of them show you a picture of a spin and hope. You nod along, close the tab, and a week later you still could not say why a long TR makes T1 contrast go away. The gap is not information. Reading about a rotating vector and turning one yourself are different acts, and only the second one stays.
So this course computes. Every curve on the page comes out of the Bloch equations while you watch, every image out of a two-dimensional Fourier transform of data the page just filled in, and the tissue values are the ones a scanner sees at 1.5 tesla. Drag TE past 200 milliseconds and the white matter really does sink into the noise, because the exponential says so and not because somebody drew it that way. When a unit claims something, you can go and break it.
The order is the argument. You cannot skip ahead, and every unit opens by restating what the last one established, so a term never arrives out of nowhere. Underlined words open their own definition, and a reference panel lists all twenty-two of them at any point. Nobody should have to scroll back to remember what TR was.
The twenty-one units
- 1
The nucleus. Why a body full of tiny magnets is not magnetic.
- 2
The field. How few nuclei it takes, and how few that really is.
- 3
Precession. Why the arrow circles instead of falling over.
- 4
Resonance. Miss the frequency and nothing happens at all.
- 5
The pulse. Tipping the magnetisation by an angle you pick.
- 6
The signal. What the coil actually sees.
halfway
- 7
T2. The decay you cannot undo.
- 8
T2 star. The decay you can.
- 9
The spin echo. Signal recovered from what looked like nothing left.
- 10
T1. The recovery clock.
- 11
TR. The waiting time, and what waiting costs you.
- 12
TE. The readout delay.
- 13
The four corners. Every weighting from two numbers.
- 14
The three gradients. What each one encodes.
- 15
k-space. Fill it yourself and watch the head appear.
- 16
Resolution, signal, time. Pick two.
- 17
Flip angle. The Ernst angle and the gradient echo.
- 18
Sequences are recipes. STIR, FLAIR, and why TI matters.
- 19
Contrast agents. Gadolinium gives off no signal at all.
- 20
Three artefacts. Each one simulated, none of them drawn.
- 21
Read a protocol. Account for every number on the line.
Who this is for
Radiology residents in their first year, radiographers who want the physics behind the console, medical students facing an MRI question, and anyone who has read three explanations of k-space and still cannot picture it. No physics degree is assumed. Bring arithmetic, curiosity and about two hours, although the course remembers where you stopped and most people come back to it three or four times.
It runs in any current browser, with no plugin and no login. A desktop screen is where it is happiest, since a slider and a curve both want room, but it works on a phone.
Hear when the next one is up
The same treatment is coming for CT, for ultrasound and for the parts of MRI this course leaves out, diffusion and perfusion and spectroscopy among them. One email per new piece, nothing else, and one click to leave.
The course itself stays free, all twenty-one units, no account and no payment.
Written by Josh Decker, MD, board-certified radiologist, at Rad Insights. The physics engine is cross-checked against analytic solutions of the Bloch equations and against a second, independent implementation. The anatomy is schematic and contains no patient data.
This is teaching material about how a scanner produces an image. It is not clinical guidance and not a basis for diagnosis or for setting protocols on a real system. Relaxation times are typical values at 1.5 T and vary with field strength, tissue, temperature and vendor.
