Lesson 3 of 10 · 18 min
Neuroimaging with PET, MRI, fMRI and EEG
Explain what PET, MRI, fMRI and EEG each measure, whether they show structure or function, how precisely they show where and when, and how they link the brain to cognition, emotion and behaviour.
What you will be able to do
- Explain what PET, MRI, fMRI and EEG each detect and what they show.
- Distinguish techniques that show structure from those that show function.
- Compare the techniques' spatial and temporal resolution, cost and risks.
- Choose and justify a technique for a research question about cognition, emotion or behaviour.
Gall could only feel the skull. Flourens and Penfield had to open it. EEG has recorded the brain's electrical activity through the scalp since the 1920s, and from the 1970s to the 1990s PET, MRI and fMRI followed. Together, neuroimaging lets researchers study the living brain from outside the head, in healthy people as well as patients, while they think, feel and act. That has done more than any earlier method to connect the brain's anatomy (its structures), its physiology (what those structures are doing) and psychology (the thinking, emotion and behaviour that results).
The techniques answer two different questions. Some show structure: what the brain looks like and whether anything is damaged. Others show function: which areas are active, and when.
Definition
Spatial and temporal resolution
Spatial resolution is how precisely a technique shows where something is (a few millimetres is fine; several centimetres is coarse). Temporal resolution is how precisely it shows when activity happens (milliseconds is fine; minutes is coarse). No technique is best at both.
Positron emission tomography (PET). The person is injected with a radioactive tracer, usually a form of glucose. Active neurons use more glucose, so the tracer collects in the most active areas. As it decays it gives off positrons, which produce gamma rays that detectors around the head pick up, and a computer builds a colour-coded image of activity. PET shows function, and with other tracers it can show where particular neurotransmitters act. Its drawbacks: it is invasive (an injection of radioactive material, so scans cannot be repeated often), expensive, and each image averages activity over a long time: about a minute with a tracer that follows blood flow, and half an hour or more with radioactive glucose.
Magnetic resonance imaging (MRI). The person lies inside a very strong magnet. Radio waves make the hydrogen atoms in the brain's water give off signals, and different tissues give different signals. The result is a detailed, three-dimensional picture of structure, sharp to about a millimetre, with no radiation. MRI shows tumours, damage from a stroke, and differences in the size of structures, but not activity. The person must keep still in a narrow, noisy tube, and some metal in the body (many older pacemakers, some implants and metal fragments) rules a scan out.
Functional MRI (fMRI). The same scanner, used to detect changes in the oxygen level of the blood. When an area becomes active, more oxygen-rich blood flows to it, and oxygen-rich and oxygen-poor blood respond differently to the magnet (the blood-oxygen-level-dependent, or BOLD, signal). fMRI shows function and structure together: which areas are active while the person does a task, on a detailed image of the brain. It has good spatial resolution (a few millimetres) and no radiation, so it can be repeated, but it is indirect and slow: blood flow rises a few seconds after the neurons fire.
Electroencephalography (EEG). Electrodes on the scalp detect the tiny voltages produced when millions of neurons in the cortex are active together, recorded as brain waves on an electroencephalogram. EEG records activity as it happens, to the millisecond, and is cheap, portable and harmless. But it has poor spatial resolution: each electrode picks up a wide patch of cortex, and activity deep in the brain barely reaches the scalp. It is used to diagnose epilepsy and to study sleep and alertness (Topic 3).
| Technique | What it detects | Shows | Spatial resolution | Temporal resolution | Main drawbacks |
|---|---|---|---|---|---|
| PET | A radioactive tracer (usually glucose) taken up by active areas | Function | Moderate (about 5 mm) | Poor (about a minute for blood flow; 30 minutes or more for glucose) | Radioactive injection; expensive; indirect |
| MRI | Signals from hydrogen atoms in a magnetic field | Structure | Very good (about 1 mm) | None: no activity shown | Must keep still; no metal; noisy |
| fMRI | Changes in blood oxygen level (BOLD) | Function and structure | Good (a few mm) | Moderate (seconds) | Indirect and lags activity; expensive; no metal |
| EEG | Electrical activity of the cortex at the scalp | Function | Poor (centimetres; surface only) | Excellent (milliseconds) | Cannot locate activity precisely or deep in the brain |
Where and when, approximately. Lower is finer detail in space; further left is finer detail in time. PET is plotted with a blood-flow tracer; with radioactive glucose each image takes half an hour or more, off the right of the chart. MRI is not on the chart because it shows no activity.
Which technique shows the structure of the brain in fine detail but not its activity?
Compare what each technique measures, then match it to the question being asked.
What imaging has shown. Neuroimaging links brain to mind in people who have no brain damage at all:
- Cognition: fMRI shows the left frontal and temporal language areas becoming active as people read and listen, and different areas becoming more active for faces, places and words.
- Emotion: fMRI shows the amygdala, deep in the temporal lobe, responding to frightened faces even when they are flashed too quickly to be seen consciously.
- Behaviour: in a PET study, Raine, Buchsbaum and LaCasse (1997) compared 41 people charged with murder who pleaded not guilty by reason of insanity with 41 people matched for age and sex. During an attention task, the prefrontal cortex of the people charged with murder used less glucose on average.
These are differences in groups and links between activity and behaviour. Raine's result does not show that low prefrontal activity causes violence, and it could not be used to diagnose an individual; the authors cautioned against both readings.
Worked example
Choose a neuroimaging technique for each question
For each research question, choose PET, MRI, fMRI or EEG, and justify your choice with a property of the technique.
(a) Is the hippocampus of people with long-term epilepsy smaller than that of people without it? (b) Which areas of the brain become more active when people look at photographs of their own family compared with strangers? Participants include 15-year-olds who will be scanned several times. (c) How many milliseconds after a word appears does the brain respond differently to real words and made-up words?
Step 1: (a) Decide whether structure or function is needed
The size of a structure is a question of structure. MRI shows structure in fine detail (about 1 mm) with no radiation, so the hippocampus can be measured accurately.
Step 2: (b) Function, where, and repeated scans of young people
The question asks where activity rises during a task: function with good spatial resolution. fMRI detects changes in blood oxygen to a few millimetres and uses no radiation, so it is safe to repeat with young participants. PET would also show activity but needs a radioactive injection each time.
Step 3: (c) A question about timing
The question is about when, to the millisecond. EEG records electrical activity as it happens, with millisecond temporal resolution; fMRI and PET are seconds to minutes too slow.
Try it yourself first, then check the working one step at a time.
Answer
(a) MRI: shows structure in fine detail. (b) fMRI: shows where activity increases, to a few millimetres, without radiation, so it can be repeated. (c) EEG: records activity to the millisecond.
Method
Decide whether the question needs structure or function.
For function, decide whether it asks where (spatial) or when (temporal).
Check for risks: radiation (PET), metal or a narrow scanner (MRI and fMRI).
Justify with the property, not with "it is best".
Watch out for
Choosing fMRI for (a). It would work, but a size question needs only structure, which MRI gives.
Choosing PET for (b) without mentioning that it needs a radioactive tracer, which matters for repeated scans of teenagers.
Why is EEG a poor choice for studying the activity of the hippocampus, a structure deep inside the temporal lobe?
Common mistake
Bright colours are not the brain "lighting up"
The colours on a PET or fMRI image are added by a computer to show where a measure (glucose use, blood oxygen) is higher than in a comparison condition. The rest of the brain is not switched off; every area is active all the time. Say "showed greater activity in" rather than "was the only area working".
Exam tip
Justify with the property
When a question asks which technique to use, name it and justify it with the property the question needs: structure or function, where (spatial resolution) or when (temporal resolution), and any risk (PET's radiation, MRI's magnet). "fMRI, because it shows which areas are active during the task with a spatial resolution of a few millimetres and no radiation" earns the marks; "fMRI, because it is the best" does not.
Summary
PET detects a radioactive tracer (usually glucose) to show active areas: function, moderate where, poor when, and it involves radiation.
MRI uses a strong magnet and radio waves to show structure in fine detail, with no radiation, but no activity.
fMRI detects changes in blood oxygen to show which areas are active during a task, on a detailed structural image: good where, moderate when.
EEG records the cortex's electrical activity at the scalp: excellent when (milliseconds), poor where.
By studying living, healthy brains, neuroimaging has linked anatomy, physiology and psychology, but brain differences between groups do not on their own show cause.
Finished the lesson?
Try the 2 quick checks above and the lesson completes itself, or mark it complete now.