Study Questions Technique For Mapping Brain

      SAN FRANCISCO - For nearly a decade, neuroscientists have used amazing tools such as functional magnetic resonance imaging (fMRI) to study what happens inside the brain.

      Some of the scientists nurse the hope of using fMRI and other tools to explain how that cantaloupe-sized jungle of nerve cells and neural wiring generates all mental activity.

      Reality may prove more complex, though.

      Recent research at the San Francisco and Davis campuses of the University of California suggests that in some situations, fMRI scanning is much less accurate than once hoped. Judging by experiments on macaque monkeys, the illuminated spots on the fMRI images sometimes lie a fair distance from where the relevant nerve cells (neurons) are actually firing. The research raises questions about efforts to use the scanning technique to pinpoint what a specific areas of the brain do.

      Some outside experts praise the research, conducted by scientists including Leah A. Krubitzer, an associate professor of psychology at UC-Davis. Krubitzer and her colleagues'' work, they say, helps to clarify the potential and limitations of fMRI as it is currently deployed. Her findings also suggest possible ways to improve the technique, they add.

      Not everyone in the fMRI research community welcomes Krubitzer''s findings.

      ``We had a hard time getting this paper published,'' she acknowledged.

      The paper eventually appeared in the prestigious Proceedings of the National Academy of Sciences, which published it in its Aug. 15 issue. Her co-authors are Elizabeth A. Disbrow, also of UC-Davis, who is listed as first author, and Daniel A. Slutsky of UC-Davis and Timothy P.L. Roberts of the department of radiology at UC-San Francisco.

      Krubitzer''s work ``alerts us to the fact that we have to be very careful. The main danger is the device is being hyped by people in psychology,'' said Christof Koch, a leading figure in fMRI research.

      ``It is a wonderful new tool,'' Koch added, ``and like any new tool, early on people are very enthusiastic about it. Some people . . . think it''s going to solve all problems. . . . In fact, we know it will solve some problems, but not all.''

      The Krubitzer paper is ``a very important contribution,'' said noted fMRI researcher and neuroscientist David Heeger of Stanford University. ``It''s very important to point this out, that (brain) signals can be mis-localized.''

      Asked if he is confident the problems can eventually be fixed, Heeger replied: ``I wouldn''t say confident, I would say hopeful. But even if we can''t localize (the signal) that precisely . . . we can still measure the level of (neural) activity averaged across several millimeters.''

      Some are more confident fMRI''s problems can be largely eliminated.

      ``The field is moving so rapidly that I think it''ll get there,'' said Mark D''Esposito, a neurscience professor at the University of California at Berkeley.

      Krubitzer and her colleagues'' findings offer particularly dramatic evidence of an old problem. As neuroscientists have long known, a problem with fMRI is that it doesn''t directly measure neural activity, or the electrical signaling between nerve cells.

      Rather, fMRI measures neural activity indirectly, specifically by tracking blood flow through the brain. The device does this by generating a magnetic field, which passes through the skull and into the brain. In the brain, the magnetic field causes iron atoms inside hemoglobin molecules, which transport oxygen through the blood, to emit radio pulses.

      An fMRI sensor detects and records those radio pulses. Then, through the use of complex mathematical software, the fMRI computer converts the pulses into images of blood flow within the brain. The fMRI images reveal second-by-second shifts in the intensity of blood flow.

      Until recently, scientists assumed that there was a direct correlation between the intensity of blood flow at one spot and the neural activity at the same locale, although there was no direct evidence of this.

      That isn''t always the case, though, as Krubitzer''s work shows.

      ``The (fMRI) technology has exploded in the last five years,'' Krubitzer said. While she is enthusiastic about the technology''s potential for mapping the brain, she cautions that ``so many people have gotten into it and gotten enthused (that) I think some (experimental results) are over-interpreted.''

      She and her colleagues conducted their experiments on two macaque monkeys. The creatures'' brains were opened, then a wire was repeatedly placed on different parts of the brain to measure neural activity. Later the results were compared with fMRI measurements of activity in the same brain regions.

      As it turned out, the match wasn''t very good. Almost half the time, the Krubitzer team reports in its article, the neural activity inferred indirectly by fMRI didn''t closely overlap the activity recorded directly by the brain electrodes.

      Before the experiment, Krubitzer was hoping to find a close match between electrode measurements and fMRI measurements of brain activity. ``We wanted this to work because (we) use fMRI,'' she said. ``But in only 55 percent (of the cases) did the data match. It kind of sucks.''

      In the experiments, she defined activity as not overlapping if the distance between the fMRI-measured and electrode-measured activities was more than a half-centimeter, that is, about one-fifth of an inch.

      The monkeys were anesthetized during the surgery. They would not have felt pain in any case, because the brain has no pain receptors, Krubitzer said. The monkeys were later euthanized.

      Despite Krubitzer''s downbeat findings, fMRI technology remains ``absolutely, positively tremendous,'' she said. That''s partly because fMRI allows researchers to study the human brain noninvasively, that is, without surgically opening it up.

      ``This isn''t the finding that we were hoping to find, for sure,'' she said. ``But I didn''t go home and put a gun to my head. . . . I don''t want to be the bad guy. We still use fMRI in our laboratory. I wouldn''t throw it out.''

     

c.2000 San Francisco Examiner