How Brain Imaging Works

Start with the magnet: how MRI maps anatomy

Standard MRI, magnetic resonance imaging, produces the crisp anatomical pictures most people picture when they think of a brain scan. It works without any X-rays or radiation. The body is mostly water, and water contains hydrogen. Each hydrogen nucleus behaves a little like a tiny spinning magnet. Ordinarily these point every which way, but inside the scanner's powerful magnetic field they line up. The scanner then sends in a brief pulse of radio waves that knocks them out of alignment, and as they settle back they give off a faint radio signal of their own.

The key is that different tissues, grey matter, white matter, fluid, a tumour, release that signal at slightly different rates. By carefully varying the magnetic field across space and reading back the timing and strength of the returning signals, the scanner works out how much signal came from each small location. A computer assembles all of this into a three-dimensional map of the brain's anatomy, slice by slice. Nothing in the image is a direct photograph; every point is a value calculated from the radio echoes. That is why MRI is so good at showing soft tissue in fine detail, and why it is the backbone of both clinical diagnosis and functional research.

The leap to function: the BOLD signal

Plain MRI shows structure. Functional MRI, fMRI, exploits a happy accident of physics to show activity as well. When a patch of brain becomes more active, it demands more oxygen, and the body responds by sending in a surplus of fresh, oxygen-rich blood, slightly more than the neurons actually consume. Oxygenated and deoxygenated blood have subtly different magnetic properties, so this local shift in the oxygen balance nudges the MRI signal up or down by a tiny amount. That change is the BOLD signal, short for blood-oxygen-level-dependent.

This is the pivot on which most of modern functional imaging turns, so it is worth being clear about what it means. fMRI does not measure neurons firing. It measures a change in blood oxygen that follows neural activity a moment later. It is an indirect proxy: a footprint in the sand rather than the foot. The link between neural activity and the BOLD response is real and well studied, work by Nikos Logothetis and others tied the signal most closely to the input and processing of a region rather than its output, but it remains a proxy, and reading fMRI honestly means never forgetting that.

How an fMRI activity map is made

A finished fMRI picture is the end of a pipeline, not a raw image. Understanding the steps demystifies both the power and the caveats of the method.

1. Scan repeatedly while conditions change

   The scanner takes a fresh whole-brain image every couple of seconds while the volunteer alternates between conditions, for example resting and then reading words. Each image is divided into tens of thousands of small blocks, and the BOLD signal in every block is recorded over time.

2. Clean up the raw data

   The data are corrected for the person's small head movements, aligned to a common anatomical template, and smoothed to reduce noise. These preprocessing choices are legitimate and necessary, but they are choices, and different choices can shift the final picture.

3. Compare conditions block by block

   For each block, software asks a statistical question: did its signal rise reliably more during reading than during rest? This produces a number for every block describing how strong and how trustworthy the difference is, not a yes or no.

4. Apply a threshold and colour the survivors

   Only blocks whose difference clears a chosen statistical threshold are kept. These are painted in colour and laid over a structural image. The bright patches are the blocks that passed the test, which is why raising or lowering the threshold changes how much of the brain appears to light up.

Seen this way, a brain activation map is a coloured summary of thousands of small statistical comparisons, laid on an anatomical backdrop. It is informative and often robust, but it is a reconstruction shaped by method at every stage, not a literal photograph of thought.

The words you will keep meeting

A handful of terms recur across every discussion of imaging. Getting them straight makes the rest of this library much easier to read.

Why the delay, and why it matters

The single most important limit of fMRI follows directly from what it measures. Neurons communicate in milliseconds, thousandths of a second. The blood-flow response that produces the BOLD signal is glacial by comparison: it takes a couple of seconds to build after the neurons fire, peaks around five or six seconds later, and then slowly fades. So fMRI is not watching the brain in real time. It is watching a slow, smeared echo of activity that has already happened, like judging where a runner was by watching ripples spread across a pond several seconds after they splashed in.

Where to go next

Now that the mechanism is clear, compare the methods that use it in the types of brain scans guide. To see how these signals are interpreted, and mis-interpreted, read what brain imaging reveals. Or step back to the overview for the big picture.