Functional Brain Imaging Using a Blood Oxygenation Sensitive Steady State Karla L. Miller, 1 * Brian A. Hargreaves, 1 Jongho Lee, 1 David Ress, 2 R. Christopher deCharms, 3 and John M. Pauly 1 Blood oxygenation level dependent (BOLD) functional MRI (fMRI) is an important method for functional neuroimaging that is sensitive to changes in blood oxygenation related to brain activation. While BOLD imaging has good spatial coverage and resolution relative to other neuroimaging methods (such as positron emission tomography (PET)), it has significant limita- tions relative to other MRI techniques, including poor spatial resolution, low signal levels, limited contrast, and image arti- facts. These limitations derive from the coupling of BOLD func- tional contrast to sources of image degradation. This work presents an alternative method for fMRI that may over-come these limitations by establishing a blood oxygenation sensitive steady-state (BOSS) that inverts the signal from deoxygenated blood relative to the water signal. BOSS fMRI allows the imag- ing parameters to be optimized independently of the functional contrast, resulting in fewer image artifacts and higher signal- to-noise ratio (SNR). In addition, BOSS fMRI has greater func- tional contrast than BOLD. BOSS fMRI requires careful shim- ming and multiple acquisitions to obtain a precise alignment of the magnetization to the SSFP frequency response. Magn Reson Med 50:675– 683, 2003. © 2003 Wiley-Liss, Inc. Key words: BOLD; oxygenation; functional MRI; neuroimaging In the 10 years since its inception, blood oxygenation level dependent (BOLD) functional MRI (fMRI) (1–3) has be- come a dominant tool for functional neuroimaging. BOLD fMRI has the unique ability to map activity at its source without the use of invasive procedures or tracers. BOLD fMRI also achieves higher resolution than other methods for global mapping. A number of localized physiological changes accom- pany neural activity, including increases in blood flow, blood volume, and oxygen metabolism. BOLD imaging senses changes in blood oxygenation by exploiting the paramagnetism of deoxyhemoglobin. During activation the cerebral blood flow increases more than oxygen consump- tion (4), resulting in a dilution of the deoxyhemoglobin concentration. Because deoxyhemoglobin experiences a resonance frequency shift relative to water (5), the pres- ence of deoxyhemoglobin introduces a spread in reso- nance frequency that causes the signal to dephase more quickly (6). In BOLD fMRI, data acquisition is delayed following RF excitation (typically by 30 – 60 ms) to allow signal dephasing to occur, resulting in signal levels that depend on the local concentration of deoxyhemoglobin (1). Since blood has a lower concentration of deoxyhemo- globin during activation, the signal during activation is larger than the signal at rest (7). This signal change is the source of the BOLD contrast. BOLD fMRI has a number of important limitations that result from the coupling of the BOLD contrast mechanism to image artifacts and signal loss. First, significant signal is lost due to transverse relaxation during the long echo time (TE), resulting in a low signal-to-noise ratio (SNR) (8). Second, sources of off-resonance in the image other than deoxyhemoglobin also cause significant dephasing, result- ing in image distortion (9,10) and signal dropout (11). Third, the long delay between excitation and acquisition necessitates the use of long readouts to maintain temporal resolution. Off-resonance and T * 2 decay over the course of these long readouts cause warping and loss of resolution (12). While these artifacts are always present in BOLD imaging, they are particularly pronounced near susceptibility bound- aries, such as the sinus cavities in the head. The severity of artifacts in these regions effectively precludes the use of BOLD fMRI in areas adjacent to the sinuses. This work introduces a new method for fMRI using balanced steady-state free precession (SSFP) imaging (13,14) that may overcome these issues. The balanced- SSFP signal is highly sensitive to resonance frequency (13,15) and can be used to directly detect the deoxyhemo- globin frequency shift associated with brain activity (16). The sequence presented here uses the phase profile of SSFP (17) to invert signal from the deoxyhemoglobin fre- quency shift relative to water, establishing a blood oxygen- ation sensitive steady-state (BOSS) signal. This method is intrinsically sensitive to the deoxyhemoglobin frequency shift, and decouples the functional contrast mechanism from sources of artifact. The major difficulty with the BOSS approach is that the resonance frequency of oxygen- ated blood must be accurately aligned with the balanced- SSFP frequency response. In practice, the accuracy re- quired to cover important regions of interest (ROIs), such as the visual cortex, can be obtained by combining careful linear shimming with acquisition at a small number of frequencies. In addition to a reduction of image artifacts and signal dropout, this method offers higher SNR and greater functional contrast than traditional BOLD fMRI. THEORY Balanced SSFP consists of a rapidly repeating series of RF pulses and fully refocused imaging gradients, as shown in Fig. 1. For short repetition times, (TRs), the magnetization 1 Department of Electrical Engineering, Stanford University, Stanford, Califor- nia. 2 Department of Radiology, Stanford University, Stanford, California. 3 Department of Psychology, Stanford University, Stanford, California. Grant sponsor: NIH; Grant number: 1P41 RR09784. *Correspondence to: Karla L. Miller, Packard Electrical Engineering Building, Room 210, Stanford University, Stanford, CA 94305-9510. E-mail: bison@stanford.edu Received 14 April 2003; revised 9 July 2003; accepted 9 July 2003. DOI 10.1002/mrm.10602 Published online in Wiley InterScience (www.interscience.wiley.com). Magnetic Resonance in Medicine 50:675– 683 (2003) © 2003 Wiley-Liss, Inc. 675