About Fuctional MRI

Current Page
fMRI Reference

The Future Role of functional MRI
in Medical Applications

The recent discovery that magnetic resonance imaging can be used to map changes in brain hemodynamics that correspond to mental operations extends traditional anatomical imaging to include maps of human brain function. The ability to observe both the structures and also which structures participate in specific functions is due to a new technique called functional magnetic resonance imaging, fMRI, and provides high resolution, noninvasive reports of neural activity detected by a blood oxygen level dependent signal (Ogawa, et al, 1990 a and b, 1992, 1993; Belliveau, et al, 1990, 1991). This new ability to directly observe brain function opens an array of new opportunities to advance our understanding of brain organization, as well as a potential new standard for assessing neurological status and neurosurgical risk. The following briefly introduces the fundamental principles of fMRI, current applications at Columbia, and some potential future directions.

Functional MRI is based on the increase in blood flow to the local vasculature that accompanies neural activity in the brain. This results in a corresponding local reduction in deoxyhemoglobin because the increase in blood flow occurs without an increase of similar magnitude in oxygen extraction (Roy and Sherrington, 1890; Plum, Posner & Troy, 1968; Posner, Plum & Poznak, 1969; Fox and Raichle, 1985). Since deoxyhemoglobin is paramagnetic, it alters the T2* weighted magnetic resonance image signal (Ogawa, et al, 1990a and b, 1992, 1993; Belliveau, et al, 1990, 1991; Turner, et al, 1991; Tank, et al, 1992). Thus, deoxyhemoglobin is sometimes referred to as an endogenous contrast enhancing agent, and serves as the source of the signal for fMRI. Using an appropriate imaging sequence, human cortical functions can be observed without the use of exogenous contrast enhancing agents on a clinical strength (1.5 T) scanner (Bandettini, et al, 1992, 1993; Kwong, et al, 1992; and Turner, et al, 1993; Schneider, et al, 1993). Functional activity of the brain determined from the magnetic resonance signal has confirmed known anatomically distinct processing areas in the visual cortex (Belliveau, et al, 1991; Ogawa, et al, 1992; Blamire, et al, 1992; Schneider, et al, 1993; Hirsch, et al, 1995), the motor cortex (Kim, et al, 1993a; Kim, et al, 1993b), and Broca's area of speech and language-related activities (Hinke, et al, 1993, Kim, et al, 1995). Further, a rapidly emerging body of literature documents corresponding findings between fMRI and conventional electrophysiological techniques to localize specific functions of the human brain (Atlas, et al, 1996; Puce, et al, 1995; Burgess, 1995; Detre, et al, 1995; George, et al, 1995; Ives, et al, 1993). Consequently, the number of medical and research centers with fMRI capabilities and investigational programs continues to escalate.

The main advantages to fMRI as a technique to image brain activity related to a specific task or sensory process include 1) the signal does not require injections of radioactive isotopes, 2) the total scan time required can be very short, i.e., on the order of 1.5 to 2.0 min per run (depending on the paradigm), and 3) the in-plane resolution of the functional image is generally about 1.5 x 1.5 mm although resolutions less than 1 mm are possible. To put these advantages in perspective, functional images obtained by the earlier method of positron emission tomography, PET, require injections of radioactive isotopes, multiple acquisitions, and, therefore, extended imaging times. Further, the expected resolution of PET images is much larger than the usual fMRI pixel size. Additionally, PET usually requires that multiple individual brain images are combined in order to obtain a reliable signal. Consequently, information on a single patient is compromised and limited to a finite number of imaging sessions. Although these limitations may serve many neuroscience applications, they are not optimally suitable to assist in a neurosurgical or treatment plan for a specific individual.

The particular imaging methods and procedures vary from center-to-center because each group has independently developed the methods and analysis procedures required to acquire and process functional data. There is not yet a commercial package of software and standardized tasks for clinical use. A brief summary of our clinical techniques is described below, and serves to illustrate the general conditions.

Image Acquisition: Images are acquired using a T2* weighted gradient echo sequence: TE = 60 ms, TR = 3 secs, flip angle = 90 deg on a 1.5 T Magnetic Resonance Imaging System (General Electric) located in the Department of Radiology at Columbia. This system is equipped with echo planar options which provides very rapid image acquisitions. Slice thickness is usually set at 5 mm but can be as thin as 3 mm. Simultaneous images are acquired on as many as 16 contiguous slices (21 slices with longer acquisition times) oriented along any suitable plane. Each imaging series requires approximately 30 complete head volume acquisitions.

Image Processing: An image processing facility has been developed as a stand-alone system outside of the scanner system. This facility provides the computational capability required to reconstruct the large numbers of images and provides the statistical analyses that identify the anatomical regions that are active during specific tasks.

Task Procedure: Patients and subjectsare positioned in the scanner as for a conventional scan, and plane lines are set based on conventional imaging methods. During a typical functional imaging series, 30 images are acquired in a 90 sec run where the initial and last 10 images are baseline conditions and the middle 10 images (30 secs) are acquired during a task. For example, in the case of a typical task designed to identify eloquent brain tissue involved in hand and finger movement, the patient taps fingers and thumb during the activity epoch. The beginning and end of this activity period is cued by a visual or auditory signal and occurs at images 10 and 20, respectively. Language, sensory, visual, auditory and other targeted functions are imaged in a similar manner. A task-induced signal change is illustrated in Fig. 1 for a sensory task involving tactile stimulation (touching) the left hand. The abscissa represents a 30 image acquisition run during a 90 sec period. The initial 10 pre-stimulation (baseline) images are followed by 10 activation images (left hand stimulation) and 10 post-stimulation images. Each 90 sec imaging series (illustrated by the intensity levels for one voxel) is actually repeated twice although only one series is illustrated. In this example the left hand stimulation results in right hemisphere activity and presumably represents the post central sulcus sensory strip.

"Click to see bigger view"Figure 1. A 5 mm axial slice is shown with the voxels indicated (small boxes) that show significant signal changes with left hand tactile stimulation. The intensity trace indicates the changes for one voxel during one image acquisition sequence of 30 images where images 1-10 were acquired during baseline conditions, images 11-20 were acquired during the task, and images 21-30 were acquired during baseline-recovery. Statistical analysis of each voxel in the entire brain over the 60 total acquisitions identifies the voxels with activity-related signal changes.

Data analysis: Statistical analyses have been developed to identify areas of the brain activated by specific tasks and are based on a multistage comparison of stimulation and resting intensity levels as well as multiple replications (Hirsch, et al, 1994 a and b; Hirsch, et al, 1995). These successive stages of analysis are illustrated in Fig. 2 for left hand finger-thumb tapping. The schematic "homonculus" illustrates that the expected and observed locations of brain activity during this task appear coincident.

"Click to see bigger view"The fundamental unit of analysis is the single voxel as shown on Fig. 1 identified by its coordinates x, y, and z where x and y indicate the in-plane location and z indicates the slice. Significant voxels (as shown in Fig. 1) indicate regions of the brain activated by a specific task. Empirical probability distributions based on phantom images (CuSO4 solution filled sphere) confirm that the probability of a significant result by chance is less than 0.0001. Voxels that do not show a significant intensity change during the activity period are not colored. Rather, they are represented by the mean level which yields the anatomic detail for each slice of brain.

Preliminary investigations of human brain mapping with these procedures have yielded insights into the functional organization of various sensory, motor, and language systems (Hirsch, et al, 1994 a-c, 1995 a-c Kim, et al, 1995; Lee, et al, 1995) and our goal is to transfer and apply these techniques to benefit patient care. Some possible applications are discussed below.


Since neurosurgery relies on a precise delineation of the structural and functional aspects of brain, the role for fMRI in neurosurgical planning is potentially very significant. The need for individualized maps of brain function is enhanced when the presence of a tumor alters the expected location of a function, or when the location of the tumor is in an area with an uncertain function such as association cortices or language-related processes. An emerging group of investigators have reported fMRI results that are consistent with electrophysiology, PET, cortical stimulation, and magneto-encephalography and serve to document that fMRI does provide a source of precise functional and structural information for Neurosurgery (Burgess, 1995; George, et al, 1995; Simpson, et al, 1995, Puce, (1995a,b); Fried, et al, 1995; Peyron, 1995; Clifford, et al, 1995; Haglund, 1995; Detre, et al, 1995; and Ives, et al, 1993). Further, the potential role of fMRI in directing decisions about surgical and diagnostic procedures has also been demonstrated (Atlas, S.W., et al, 1996).

The following example illustrates the potential advantage of functional plus anatomical information for surgical treatment of brain tumors that are located near active areas of brain.

"Click to see bigger view"A single axial slice is shown in Fig. 3 for a 32 year old male patient before and after resection of a left frontal lobe GBM. Prior to surgery various language and speech tasks were employed to chart functional language areas in the vicinity of the tumor. The left slice shows an area active during extemporaneous speech just posterior and adjacent to the tumor. The right slice shows the same area of brain following the resection. The speech area remains just posterior to the resected tumor bed and the patient experienced no speech deficit following surgery. This example also illustrates a potential challenge for functional imaging. The choices of tasks must be selected based on prior knowledge of functions near the targeted areas. Therefore, a null finding may only mean that the optimal task was not applied.


Currently, a cohort of neurosurgical patients are participants in a protocol at Columbia to evaluate the potential applications of fMRI for neurosurgical planning. These patients receive a standard battery of tasks targeted to localize language, sensory, motor and visual areas both as candidates for surgery and as post-surgical patients. Functional imaging results are compared with all conventional mapping studies performed on each patient including the WADA test, intraoperative cortical stimulation, electrophysiological assessments, and neurologic assessments of surgical outcome. The objective of this on-going investigation is to determine the potential role of functional mapping for neurosurgical procedures.


The experience of chronic and persistent pain is a debilitating condition for which the role of cortical processing is not well understood. We have focused on the identification of cortical areas that are modified by the reduction of pain following pain therapy. This novel approach to investigate the cortical representation associated with relief of pain has originated from our pilot studies where patients with chronic sympathetically maintained pain affecting one extremity (post herpetic neuralgia) were studied by comparing brain responses to light touch applied to the "now-affected" limb and to the "painful" limb before and after treatment (Hewitt, et al, 1995). These studies indicate that the cortical representation of sympathetically maintained pain involves specific and identifiable cortical activity, as well as does the relief of that pain achieved by a peripheral nerve block procedure. Continuing investigations will extend these findings to other pain treatments to determine the extent to which this finding is generalizable to other pain relief mechanisms. These preliminary studies suggest a wide range of other approaches using fMRI to investigate cortical representations of specific pain types, and therefore, new specific therapy options.


The following example illustrates the potential of fMRI to yield new insights into physiological bases for disfunction. A 16 year old right handed female with a congenital malformation in the right posterior frontal lobe and a seizure disorder participated in a functional imaging study to identify functional sensorimotor areas. One of the patient's typical sensory seizures occured during one of the runs. This enabled us to localize fMRI signals associated with the onset of a spontaneous seizure, its progression, and the relationship to normally activated motor cortex. There was minimal movement artifact which was further reduced by alignment of the images prior to a voxel-by-voxel statistical analysis. The fMRI signals associated with the seizure were first observed in an area adjacent to the normal motor activity. MR signal amplitudes exceeded the normal functional activity by as much as a factor of 5. A sequential time analysis (6 sec intervals) revealed both local spreading of the onset focus and the emergence of subsequent foci first in the ipsilateral prefrontal areas, and the mesial surface areas. These were followed by activity in the homologous regions of the opposite hemisphere suggesting that the generalization followed a specific pattern of functional connectivity. Thus, eloquent motor activity and seizure activity were co-localized using fMRI, and the onset, progression pattern, time course, and relative MR amplitudes of the seizure event were observed (Hirsch, et al,1996). This case illustrates that fMRI may contribute to improved precision of seizure localization and understanding of seizure progression, and suggests a future direction for investigation.

Other neurological conditions currently under investigation using fMRI at Columbia include neglect syndromes, phantom pain, cerebellar dysfunction, and neural reorganization. Preliminary studies confirm that the pathways and processes involved in these neurological disorders and conditions can be observed for investigation by fMRI.


Due to the ability to image the entire 3-dimensional volume of brain, fMRI is capable of isolating many simultaneous and coordinated brain events. This "multi-level" view of brain activity can include "executive" functions and high level cognitive tasks simultaneously with the primary and secondary input such as vision and audition as well as cerebellar contributions. We are currently applying fMRI methods to identify brain structures uniquely involved with visual perceptions, language generation, comprehension of sequential information as in a movie, the execution of visually guided responses, and complex problem solving. These aspects of brain function have not previously been scrutinized with such precision, and represent some of the remaining frontiers in Neuroscience.


The goal of this presentation was to introduce the basics of fMRI and to suggest potential future applications in neuro-oncology. Based on our initial investigations, these future directions include neurosurgical planning and improved assessment of risk for individual patients, improved assessment and strategies for the treatment of chronic pain, improved seizure localization, and improved understanding of the physiology of neurological disorders. We look ahead to these and other emerging applications as the benefits of this technology become incorporated into current and future patient care.

Analysis Documentation

Reconstruction Documentation

Export Documentation

fMRI Documenation

SPM2 analysis manual

Spaces and spatial normalisation