Brain Scanners are still far away from providing useful diagnoses

Wired Magazine - Brain Scans as Mind Readers? Don't Believe the Hype

This article explores a few different places where brain scans like fMRI and EEG are being used in businesses, from personalized brain scans (costing over 3K) to companies trying to make a business out of fMRI lie detectors. The problems with the brain scanning devices is as follows: though they find statistical correlations between clinical diagnoses (depression, anxiety, excessive anger) and brain activity, there are still those with the diagnosis who have none of the relevant brain activity, and there are those with relevant brain activity without any diagnosis. What the science finds is statistically significant correlations, but not type-type correlations. So there are plenty of outliers, those who exhibit all the symptoms of depression who don't have the relevant brain activity, and those who report no problems with depression who have the relevant brain activity. A proposed fair way of determining brain scan success is to have a blind-detection process: to look at brain scans and make diagnoses that jibe with clinical ones. This work has yet to succeed. The one place where this kind of work is especially promising is in lie detection. A kind of super-lie-detector might be the fMRI, which tries detects the extra brain work needed to lie. One company posts a 90% success rate.