fMRI: Out of the Lab and Into the Clinic

Image from GE's BRAINWAVE softwareResearch into functional magnetic resonance imaging (fMRI) as a non-invasive method of understanding the complex relationship between the structure and function of the brain began in the 1990s. Current research programs in tandem with advances in techniques and increased magnet field strength of MR systems hold promise for development of robust clinical applications.

While some clinical departments use fMRI for pre-surgical brain mapping to guide neurosurgical procedures, the approval of three new CPT billing codes that became effective on the first day of 2007 promises to propel use of these scans into a wider array of clinical applications.   Meanwhile, major research centers continue to develop new methods designed to improve specificity in data produced by these specialized scans.

Inside the research

Truman R. Brown, director of MR research and professor of radiology at the Columbia University Medical Center in New York explains that the primary thrust of their “bread and butter” studies focuses on a wide variety of psychological experiments to determine which portions of the brain are activated when the subject performs specific tasks.
Functional MRI experiments usually involve looking at single event trials. The subject is shown a stimulus or asked to perform a task, and then the response is measured via analysis of hemodynamic shifts in the brain through analysis of BOLD (Blood Oxygen Level Dependent) signals. In some instances, response time is longer, and sometimes shorter.

A new direction in their research efforts involves using simultaneous measurements from EEG (electroencephalograms) and fMRI. This combination enables measuring neurotemporal events in a millisecond time scale, with spatial resolution provided by fMRI images. “We are able to tell from the EEG whether there is a difference in the electrical signals from the brain when the response is longer. If that is the case, then we can see differences in the fMRI to try to understand where those variations come from, and we are very excited about the possibilities of doing this,” Brown concludes.

He anticipates that while currently these combined scans provide additional basic understanding of the brain’s function, his clinical colleagues are hopeful that this combined technique may prove valuable in studying psychiatric illnesses.

This center has been involved in studying Alzheimer’s disease and other forms of dementia, although they have not begun using the combined EEG/fMRI approach at this point. Preliminary results suggest that there may be different blood flow distributions in individuals with Alzheimer’s as opposed to those who do not exhibit symptoms.

The department of biomedical engineering at Columbia installed a Philips Medical Systems Achieva 3.0 Tesla MR system in July 2006. Brown explains that with the higher field strength, the signal is bigger, and they are able to push the signal to noise limits more effectively, in less time.

One of Brown’s research interests that benefits from the use of a higher field strength magnet is the ability to study metabolites rather than only studying water protons. “In principle, you can observe the metabolites that are present in much lower concentrations than water, like 1000 times less. As we get to higher and higher field strengths, our ability to measure neurotransmitters in the brain — our ability to see what is happening and changing on a minute by minute basis in the brain — I believe will give us greater insights, and this is an area where higher field strength will be invaluable.”

Finally, this group has been using Arterial Spin Labeling (ASL) in their research by applying radiofrequency fields with the gradients. They utilize spins in blood flow up the carotids and magnetically label them by inverting the magnetization, which then slowly reverts to its initial state. By making the signal negative and then positive, they can determine how much blood flowed into the tissue. This provides similar information to those studies where contrast agents are employed, but does not require such an injection.

Brown concludes, “I mention ASL, which has the potential to be better on the 3T because of the relaxation times on T1, so that inverted blood, say in the carotid, will ‘remember’ that for a longer period of time.”

Gary H. Glover, PhD, professor of radiology and director of radiological sciences at Stanford University School of Medicine in California describes his experience in cognitive neuroscience as focusing on fMRI technology development rather than clinical applications. However, this department has been involved in using fMRI for pre-surgical planning studies since 1995, using systems they developed themselves.

Currently, they use Signa Excite 1.5 Tesla and Signa HDx 3.0T MR scanners from GE Healthcare, and they have just installed a GE 7.0 T prototype for research projects. In addition, the medical center uses the GE BRAINWAVE fMRI software packages on a couple of their GE scanners that are used solely for clinical applications.

One of the distinctions between clinical fMRI and cognitive neuroscience is that in the latter 20 to 30 subjects are scanned and then analysis of group data is performed, Glover explains. In the clinical setting they must be careful about what they call activation signals because there is no probability map. On the other hand, Glover notes that the types of tasks they have subjects perform are quite robust, which provides the information a clinician might need about precise areas where motor function or language function is processed.

In pre-surgical planning, fMRI can provide reference points about where functionality resides in the brain, despite any distortion caused by a tumor. “You set your threshold higher, so that fewer voxels appear and you can be more confident that those are the voxels that are truly activated by that task.” When the neurosurgeon is trying to determine the margin of a tumor in relation to specific functions of the brain, more information is better than less. Glover concludes that issues of where to set thresholds in fMRI are not completely solved at this point, but research continues.

“There is a huge advantage of a 3T scanner over a 1.5T, roughly a factor of 2,” Glover says. “It will give you more confidence when you think a section has been activated. If you spend the same scan time, you’ll have higher confidence.”

He anticipates that in the future, there is promise for development of a therapeutic use of fMRI to enhance individual’s ability to volitionally control pain remediation. As a member of an interdisciplinary team that included several Stanford faculty members and Christopher deCharms, PhD of Omneuron in Menlo Park, Calif., they designed a research project to determine whether subjects could learn to control a specific region of the brain involved in pain perception known as the rostral anterior cingulated cortex (rACC). This group produced software that provides near real-time analysis of fMRI imaging data, which permits subjects to observe their own brain activity as they are being scanned. Subjects were given suggestions about how they might change their brain activity in the rACC. While the resulting ability of subjects being able to learn volitional control over their pain appears quite promising, much work remains to perfect and document the technique.

Planning brain surgery

Neeraj Chepuri, MD, radiology department chair of Abbott Northwestern Hospital in Minneapolis describes his clinical practice where they have leveraged past research into fMRI with the Siemens Magnetom Trio 3T system to enhance their pre-surgical planning capabilities for neurosurgery. Over the past year, he estimates that they have performed between 40 and 50 BOLD fMRI studies to provide a roadmap for neurosurgeons to employ.

By combining software from Siemens that permits overlay of fMRI activation maps over typical MR images, neurosurgeons can plan their approach to reduce damage to normal brain tissue as much as possible. “I’ve taken some shareware software to provide surface-rendered images. We can provide the neurosurgeon with an image of the brain surface as if he or she took a film of the brain plus the activation map.”

For example, if hand control functionality resides in the portion of the brain to the left of the tumor, the neurosurgeon will use an operative approach from the right side of the tumor rather than from the midline. “This allows them to tailor their approach to the tumor, and allows them to create a smaller hole in the skull which leads to better patient care and easier recuperation,” Chepuri explains. In addition, there is less chance of damage to hand motor function because the surgical approach did not disrupt neurologic pathways.

In his department, he has coupled sophisticated software from Siemens with his home grown solution that uses an off-the-shelf LCD projector and computer with PowerPoint to record patient actions, such as finger tapping. This approach provides specific data that is of value to clinicians and the MR technologist to communicate with the patient while he or she is in the scanner.
Chepuri and his colleagues are using the Magnetom Trio 3T with Tim (Total Imaging Matrix) technology, a short bore magnet with 32 receiver channels designed to increase resolution.

CPT codes improve outlook

The addition of three new CPT codes for fMRI that took effect in January should help to propel the use and development of fMRI further.  There is one code for neurology and two for radiology that pertain to fMRI reimbursement.  Prior to adoption of these codes, some centers would add functional aspects of studies onto regular MRI scans that were required for diagnostic workups, while not receiving reimbursement for the fMRI portion of the study. In other instances, patients who requested brain mapping either had to pay for the scan themselves or be entered into a research protocol that provided access to an fMRI exam.

Experts have suggested that once these CPT reimbursement codes are in play, the number of clinical studies should escalate, and new applications will be developed.

Stanford’s Glover describes other emerging potential uses for fMRI.  Some groups have begun to explore the use of fMRI as a means of modulating specific brain function in conditions such as addiction, depression and neuropsychiatric disorders. And a non-clinical application currently under study pertains to how or if a subject is affected by specific advertisements. Whether or not this will ultimately develop clinical applications remains to be seen.

Other experts have suggested that fMRI may prove useful in assessment of awareness and consciousness, evaluation of comatose patients, brain mapping of seizures in tandem with EEG studies, and potentially in evaluation and diagnosis of dyslexia and autism.


Functional MRI has begun the migration from the research lab to radiology departments and other clinical settings with promising applications that benefit patient care. Higher strength magnets improve image resolution as they reduce the time required for imaging studies to be completed. With the adoption of three new CPT codes, many experts anticipate increased utilization for fMRI in the future both in current applications such as pre-surgical mapping to improve patient outcome and in new fields.