Oscillating steady state imaging (OSSI) for fMRI commonly combines images across a fast oscillation to produce a steady time series suitable for fMRI analysis. The signal magnitude derived from L2-norm combination can be sensitive to frequency variations, which can exacerbate physiological noise, particularly respiration. Here, we use a 1D convolutional neural network (1DCNN) to directly estimate the underlying T2* BOLD response from both simulated and real OSSI fMRI signals. We demonstrated that our technique reduces B0 fluctuations and thermal noise as compared to the L2-norm combined OSSI fMRI signal.
This abstract and the presentation materials are available to members only; a login is required.