HADAMARD-ENCODED BOLD FMRI FOR REDUCED SIGNAL DROPOUT

点击翻译按钮获取中文摘要

This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. Primary support for the subproject and the subproject's principal investigator may have been provided by other sources, including other NIH sources. The Total Cost listed for the subproject likely represents the estimated amount of Center infrastructure utilized by the subproject, not direct funding provided by the NCRR grant to the subproject or subproject staff. Introduction: BOLD fMRI suffers from signal dropout in frontal-orbital and lateral parietal/temporal regions from susceptibility differences between air and tissue induces intravoxel dephasing. By decreasing the slice thickness, dephasing is reduced and signal is regained, but at the expense of signal to noise ratio (SNR) in magnetically uniform regions of the brain (1). Here we introduce a novel solution: the use of Hadamard-encoding to simultaneously excite pairs of subslices that are subsequently combined incoherently using UNFOLD (2) to gain signal in dropout regions at no loss of SNR efficiency in uniform regions. Methods: Alternately applying sine- and cosine-modulated Hadamard pulses (3) in a dynamic acquisition, two sub-slices of half the desired slice thickness are excited in-phase and out-of-phase (Fig. 1). Assuming there is a phase shift of ¿ between subslices because of the susceptibility-induced gradients, the resulting complex-valued time series contains magnitude components , where ¿i are the magnitudes of the subslice signals, and the sign alternates with time frame t because of the alternating excitation. Squaring y(t) and taking its Fourier transform, one finds a component Y1 with spectrum centered at DC corresponding to the term and a second component Y2 centered at the Nyquist frequency corresponding to (Fig. 2). Applying an UNFOLD filter H(w) to remove the Nyquist component, inverse transforming and taking the square root yields a reconstructed timeseries , i.e. the square root of the sum of squares of the two subslices. The influence of the intravoxel dephasing component ¿ is thereby removed. The Hadamard method was implemented in a spiral-in/out pulse sequence (4). 60 2 mm thick subslices were acquired for 128 time frames. Timeseries corresponding to 4 mm slices were obtained from the magnitude reconstructed images as above using a two-point boxcar filter ( ). Functional data were obtained at 3T using a breath hold task to elicit activation in most of the brain (5). These scans were compared to a similar conventional method. To read about other projects ongoing at the Lucas Center, please visit http://rsl.stanford.edu/ (Lucas Annual Report and ISMRM 2011 Abstracts)