Silent sparse sampling for auditory fMRI

Introduction: The investigation of complex auditory processes has been difficult using fMRI due to the excessive image acquisition noise, typically 100-130 dB. The development of the sparse sampling paradigm was a major improvement in auditory research [1]. In this method, only a single volume is acquired with a long TR usually 12-16 seconds. The timing of the stimulus is jittered relative to the acquisition to sample different points of the hemodynamic response. This method is long and inefficient since only a single sample of the hemodynamic response is made per long TR. The development of the interleaved silent steady state imaging (ISSS) improved the efficiency of the acquisition at the expense of added noise during the stimulus period [2]. In this method, the EPI module was eliminated during the silent period which keeps the magnetization in steady state. The ISSS method does improve the efficiency by collecting a number of volumes per TR. However, the use of slice selection during the silent period has a large auditory component that interferes with subtle auditory experiments such as detection of voices in background noise or the detection of language related tones (Chinese or Hindi) compared to non-language tones. In this study, we propose a variation of the ISSS method that is a truly silent acquisition scheme to acquire subtle auditory fMRI data. Methods: Sequence development occurred on a 3T scanner (Siemens TIM Trio B15 software, Erlangen, Germany) based on the standard gradient echo EPI readout. Initial implementation of the ISSS method resulted in a significant level of noise during the silent period. This prompted the modification of the gradient switching during the silent period. The slice select gradient was modified such that it stayed on during the entire silent period. The TR of this design is short compared to the sparse sampling paradigm (figure 1). The RF pulses were played at the appropriate times to maintain the steady state magnetization. This reduced the gradient noise during the silent period. Actual sound pressure levels were measured using an SPL meter (Radio Shack). To demonstrate that the magnetization remains in steady state during silent and imaging periods, time series data from a phantom were collected using a standard EPI data collection and the proposed silent imaging method. The experiment used EPI imaging parameters of a TR=2s, TE=25ms, slice thickness of 3mm, flip angle of 80 degrees, BW of 2442 Hz/pix, FOV=220, and matrix of 64. This was used for the silent fMRI scanning as well as the traditional continuous imaging. The continuous imaging sequence was used as the reference for imaging stability and auditory measurements. The silent fMRI scan had 5 volumes (10s) of data collection preceded by 4 volumes (8s) of silence during which the constant slice select gradient and RF pulses were played. This is a typical sparse sampling auditory experiment setup. It allows enough time for the stimuli to generate a BOLD response that does not overlap with the response due to the scanner noise.