When a person is telling the truth they use different parts of their brain than when people lie, the Temple University team said.
These changes were detected by functional magnetic resonance imaging.
The method may prove more accurate than traditional machines, they told the Radiological Society of North America.
The conventional polygraph lie detector looks for body changes linked with lying such as sweating and changes in blood pressure, heart rate and breathing.
But Dr Scott Faro and his team say the accuracy is limited because people who are telling the truth can show similar changes merely as a result of being anxious about being tested.
Furthermore, those adept at lying can learn how to cheat the polygraph test.
They asked six of 11 volunteers to fire a toy gun and then lie about what they had done. The other five were asked to tell the truth about what had happened.
Each of the volunteers was then scanned with fMRI while being asked questions by the scientists.
A polygraph test was also carried out for comparison.
In all cases the polygraph and the fMRI accurately distinguished between the volunteers who were telling the truth and those who were lying.
On the brain scans, different areas of the brain were active when the person was lying than when they were telling the truth.
Also, more areas of the brain were activated when the person was trying to deceive the questioner.
Although it is too early to tell whether confident liars could cheat the fMRI test, Dr Faro is hopeful it could be a more accurate way of spotting deception.
Professor Richard Wiseman, from the Psychology Department at the University of Hertfordshire and who has carried out research into lie detection, said: "I'm sure it would be better than the polygraph.
"The problem with the polygraph is it's a measure of how anxious somebody is.
"Lots of people become anxious when they are attached to the polygraph anyway and good liars are not anxious when they lie.
"With fMRI you are looking at the brain's activity and lying is cognitively quite hard.
"You are having to think what is plausible, what does the person know, what can they go and check on, and so on.
"So, in terms of brain activity, the indicators are likely to be more reliable."
He said the only shortfall was how practical it was to use fMRI routinely because it requires the patient to remain relatively still inside a large, expensive tube-like machine which performs the scanning.
"It's not the sort of thing every police station has in the back, but in the future, potentially in high profile cases, it might be something people want to look at," he said.
Story from BBC NEWS:
Published: 2004/11/30 01:46:14 GMT © BBC 2013
The technique, called "brain fingerprinting", has already been tested by the FBI and has now become part of the key evidence to overturn the murder conviction of Jimmy Ray Slaughter who is facing execution in Oklahoma.
Brain Fingerprinting, developed by Dr Larry Farwell, chief scientist and founder of Brain Fingerprinting Laboratories, is a method of reading the brain's involuntary electrical activity in response to a subject being shown certain images relating to a crime.
Unlike the polygraph or lie detector to which it is often compared, the accuracy of this technology lies in its ability to pick up the electrical signal, known as a p300 wave, before the suspect has time to affect the output.
"It is highly scientific, brain fingerprinting doesn't have anything to do with the emotions, whether a person is sweating or not; it simply detects scientifically if that information is stored in the brain," says Dr Farwell.
"It doesn't depend upon the subjective interpretation of the person conducting the test. The computer monitors the information and comes up with information present or information absent."
A few days ago Dr Farwell ran the test on Jimmy Ray Slaughter at the maximum security state prison in Oklahoma.
A jury convicted Slaughter of shooting, stabbing and mutilating his former girlfriend, Melody Wuertz, and of shooting to death their eleven-month old-daughter, Jessica.
The crimes for which he is sentenced to death took place in a house that he is very familiar with. The results were revealing.
"Jimmy Ray Slaughter did not know where in the house the murder took place; he didn't know where the mother's body was lying or what was on her clothing at the time of death - a salient fact in the case," says Dr Farwell.
During the test, the suspect wears a headband equipped with sensors to measure activity in response to recognition of an image relating to the crime - for example, a murder weapon or possibly a code word in the case of a spy.
Brain Fingerprinting has profound implications for the criminal justice system.
Any decision relies on more than just the outcome of a forensic test such as brain fingerprinting. However, in the light of these findings, the case for appeal hopes that Slaughter will either be granted a pardon, clemency or a retrial.
Critics of brain fingerprinting believe it needs far more refinement before its use becomes widespread and cases are won and lost on its evidence.
Needless to say, Dr Farwell disagrees.
"What I can say definitively from a scientific standpoint, is that Jimmy Ray Slaughter's brain does not contain a record of some of the most salient details about the murder for which he's been convicted and sentenced to death," says Dr Farwell.
Story from BBC NEWS:
Published: 2004/02/17 10:47:43 GMT
© BBC 2013