Types of Brain Scans

The central trade-off

Before the individual methods, one idea organises them all. Every technique measures a different physical thing, and that choice fixes its strengths and weaknesses. A method can excel at saying where in the brain something happened, its spatial resolution, or at saying exactly when it happened, its temporal resolution, but almost never both at once. The reason is physical: the signals that pin down location well, like blood flow, are slow, while the signals that pin down timing well, like electrical fields, are hard to trace precisely back through the skull to their source.

So the toolkit is not a ladder from worse to better. It is a set of specialists. When you understand which trade a method makes, you understand which questions it can answer. Reading a scan wisely starts with knowing what its numbers actually came from.

The methods at a glance

This table summarises the six most common methods. Treat the resolution columns as broad characterisations rather than exact figures, since they vary with the specific scanner and protocol.

The structural pair: MRI and CT

Two methods answer the question of what the brain physically looks like. MRI, described in detail on the how it works page, gives exquisite pictures of soft tissue without any radiation, which makes it the first choice for detailed anatomy and for spotting subtle changes in tissue. Its drawbacks are that it is slower, noisier, more expensive, and unsuitable for people with certain metal implants, since it relies on a very strong magnet.

The two are complementary rather than rivals. In an emergency, CT wins on speed and availability; for a careful look at soft tissue over time, MRI wins on detail. Choosing between them is a practical decision about the clinical question, not a ranking of one as simply superior.

The functional methods and their bargain

Four methods track activity rather than anatomy, and they divide neatly by the trade-off above. On one side sit fMRI and PET, which follow slow, blood-and-metabolism-related signals. They locate activity reasonably well in space but are sluggish in time. fMRI infers activity from the BOLD signal and needs no injection; PET follows an injected radioactive tracer and can be tuned to track specific molecules, such as a particular receptor, which fMRI cannot do, at the cost of a small radiation dose.

On the other side sit EEG and MEG, which listen to the electrical activity of neurons directly, either as voltages at the scalp or as the faint magnetic fields that activity produces. Because they read the electrical signal itself, their timing is superb, down to the millisecond, which makes them ideal for questions about the order and speed of mental events. Their weakness is location: working out exactly where inside the head a scalp signal originated is a genuinely hard problem, and EEG in particular blurs it badly. MEG localises somewhat better than EEG but needs a large, heavily shielded machine.

Reading resolution figures wisely

It is tempting to treat spatial and temporal resolution as fixed specifications, like the megapixels of a camera, but they are more like ranges that depend on the machine, the settings, and what you are willing to trade away. A research fMRI scanner can be pushed to finer spatial detail at the cost of slower, noisier images; an EEG setup can gain a little localisation by adding many more electrodes and heavy computation. So the honest way to read the table above is as a description of each method's typical bargain, not a promise of exact numbers.

There is also a subtler point hidden in the word resolution. High spatial resolution tells you how finely the machine samples space, but the meaningful unit is still a voxel containing many thousands of neurons, so even a sharp fMRI image is coarse compared with the cells doing the work. Likewise, EEG's millisecond timing is genuine, yet the same signal could have come from several possible sources, so its precision in time sits alongside real uncertainty about place. Keeping both halves of each method in mind, what it nails and what it blurs, is what separates a careful reading of a scan from an impressed one.

Why combining methods is the future

Because no single method is strong on every front, the most informative studies increasingly pair methods with complementary weaknesses. The classic combination joins the millisecond timing of EEG or MEG with the spatial detail of fMRI, so that a researcher can say both roughly where an event happened and precisely when. Others overlay a functional activity map on a high-resolution structural scan, or use PET to identify a specific molecule while MRI supplies the anatomy around it. None of this makes any one method obsolete; it plays their strengths together. The practical lesson for reading brain research is simple: ask what was measured and what was traded away, and you will understand far more from a study than its headline conveys.

Where to go next

Now that you can tell the methods apart, see how their signals get turned into claims, and the myths that follow, on what brain imaging reveals. For the physics behind the two headline methods, revisit how brain imaging works. Or return to the overview.