Neuorimaging - bioethics

This article in Science brings up some interesting points to ponder (Science, Vol 307, Issue 5715, 1548-1550).

"But the ability to watch the brain in action raises many questions about when, if ever, society has a right to know what someone is thinking. 'If some of these technologies become available, it could change how we live enormously,' says Henry Greely, a law professor at Stanford University in California who has written extensively about the legal and social implications of neuroimaging technologies. 'To the extent that small, easy-to-use devices could tell, either voluntarily or surreptitiously, what was going on inside someone's head, that could have enormous uses throughout society--and also what we today would consider abuses.'
Beyond personality assessment lies the prospect of detecting defects in brain functioning that could contribute to criminal acts. Imaging studies have shown that moral reasoning engages parts of the brain that are not involved in other forms of reasoning, and other studies have found reduced activity in some of the same brain regions among convicted murderers. One goal of 'forensic neuroimaging,' says Canli, is to determine whether individuals with a reduced ability to feel empathy, guilt, or remorse about criminal acts exhibit a unique neural signal. If so, this information could be used to monitor individuals at risk of carrying out a criminal act or in sentencing and parole decisions."

OK, clearly we're a long way from being able to do this effectively (technologically speaking) and in a robustly repeatable (results-section speaking) fashion. But let's assume for a moment that the technology were available. Let's say there were indications that we could reliably tell someone's likelihood of having committed a past crime or of committing future crimes. What technological threshold would this sort of technology need to pass to be implemented in courts of law? What ethical boundaries would we need to set to responsibly allow for rectification of false-positive results? How much weight would we allow neuroimaging to wield when it comes to allowing a jury to reach a guilty/not guilty verdict or a judge to sentence a convict? Any ideas?