Roll up your sleeves; this is a no-nonsense article, in which we will lay out for you all the neuroimaging modalities currently available, with a little explanation on how each works, and a tidbit or two on how it can (or can't) be used in psychiatry.

CT (Computed Tomography) Scan
What it is. If you shoot an X-ray through the brain, denser structures will impede (or "attenuate") the radiation more than less dense structures. The more of the Xray that passes through, the more exposed, or darker, that part of the image. Bone is very dense, and blocks all of the radiation, so it looks white. Water is least dense, so CSF is dark on CT and brain parenchyma is in between.
Uses in Psychiatry. Few to none. The only time you might consider ordering a CT scan is in ruling out a tumor or a stroke in someone with an atypical psychiatric presentation, but even in these instances you should choose an MRI over CT if at all possible. ER docs use CT all the time because it's cheap, quick and good for picking up acute bleeding.

MRI (Magnetic Resonance Imaging)
What it is. Put a brain in a strong magnetic field, and the protons of the hydrogen atoms will align in the direction of that field. An MRI signal ("radio pulse") disturbs that alignment, and as the protons realign themselves with the magnetic field, they release energy. This energy, analyzed and transformed by computers, is the basis for the images we see in an MRI. By controlling the type of radio pulse delivered, the MRI tech can "interrogate" different types of tissues. Thus, "T-2 weighted images" are best for picturing water, which looks white on the film. Since most brain pathology involves edema, T-2 images are preferred for detecting pathology, while T-1 images are better for normal anatomy.
Uses in Psychiatry. Debatable. Early research led to excitement about "white matter hyperintensities" in psychiatric patients, but later studies found these same blips in normal brains. At this point, the main use of MRI is to exclude potentially reversible causes of psychiatric symptoms (such as tumors and normal pressure hydrocephalus). However, the yield is extremely low. In a survey of 6200 psychiatric inpatients at McLean Hospital, 99 patients, or 1.6%, had MRI findings that might theoretically have led to a change in clinical management (Renshaw PF, Rauch SL: Neuroimaging in clinical psychiatry. In: Nicholi AM, Jr, ed. The Harvard Guide to Psychiatry, Third Edition. Cambridge, Massachusetts: Belknap Press; 1999.) A shotgun approach to ordering MRIs is likely to lead to coincidental findings of unclear significance that might lead patients down a road of unneeded diagnostic procedures. Most authorities limit their MRI recommendations to the following scenarios: pre-ECT workup, atypically late age of onset of a psychiatric condition, history of head trauma, presence of delirium, and focal neurological signs accompanying psychiatric symptoms.

PET (Positron Emission Tomography)
What it is. See this issue’s interview with Dr. Darin Dougherty for plenty of PET information. PET scanners use radioactive isotopes to measure metabolic activity or to identify neurotransmitter receptor sites. The downside of PET is that you need an on-site cyclotron, a very large and expensive ($2 million) machine that allows you to accelerate protons and smash them into target atoms in order to create the isotopes required for PET scanning.
Uses in Psychiatry. At this point, there is no reimbursable use of PET scan in psychiatry. In neurology and oncology, PET is covered for localizing the focus of a seizure before planned neurosurgery and for diagnosing and staging most cancers. Many researchers believe that it should be approved for early diagnosis of Alzheimer's Dementia, but the data are not yet compelling enough to convince either TCR or the Gods of Medicare (see the article in this issue).

SPECT (Single Photon Emission Computed Tomography)
What it is. SPECT can be thought of as PET's poorer cousin. Like PET, it depends on radiopharmaceuticals to detect brain activity, but it measures photons shot in one direction rather than gamma rays shot in two directions. This means that SPECT cannot produce the same degree of spatial resolution as PET. On the positive side, SPECT scans are cheaper, because they can use isotopes with longer half-lives, and therefore don't require an on-site cyclotron.
Uses in Psychiatry. Like PET, there are no reimbursable uses in psychiatry.

fMRI (Functional Magnetic Resonance Imaging)
What it is. This is an adaptation of basic MRI techniques that uses a measure of blood oxygenation to visualize blood flow and neuronal activity. Unlike PET, which does basically the same thing, fMRI is noninvasive, and therefore easier to perform and less risky for patients.
Uses in Psychiatry. There are no clinical uses in psychiatry, but it is used frequently in research. One of the more fascinating fMRI experiments was one published recently in Science, in which subjects were given a list of words and then asked to suppress the memory of certain words. fMRI was used to identify the neural substrate of suppression; perhaps predictably, many prefrontal brain regions were more active during memory suppression than retrieval (Science 2004 Jan 9; 303:232-5).

MRS (Magnetic Resonance Spectroscopy)
What it is. MRS measures brain chemistry safely and noninvasively. The technique allows you to measure levels of specific compounds in specific brain regions. Each compound has its own "frequency signature," and amazingly enough, MRS can detect these compounds without the need to inject any radioactive tracers.
Uses in Psychiatry. No approved clinical uses yet, but you'll start to see more and more studies reporting levels of various compounds in the brains of patients with various disorders. Understanding these results, however, is not for the biochemically faint of heart. A recent study, for example, found elevated lactate levels in the gray matter of medication-free patients with bipolar disorder. The authors concluded that the findings "suggest a shift in energy redox state from oxidative phosphorylation toward glycolysis" (Arch Gen Psychiatry. 2004;61:450-458). Toto, I have a feeling we're not in Kansas anymore.

TCR VERDICT:
Neuroimaging: Use it for Ruling Out, Not for Ruling In