Figure 2:

Self-gating (SG) signal extraction and processing pipeline. First, the interleaved self-gating readout lines through the center of the k_y-k_z k-space plane are extracted and reorganized into Casorati matrix, M. Two parallel filtering operations are performed along the temporal dimension (rows) of M, followed by principal component analysis (PCA) to extract cardiac and respiratory motion surrogate signals. Detected peaks (red triangles) are shown on the cardiac signal and used for cardiac binning. The weighting function overlaid on the respiratory amplitude histogram is centered at end-expiration and incorporated into the reconstruction process. Parameters used for this example are as follows: TR = 4.0 ms, SG sampling frequency = 9TR, acquisition time = 5 min, readout direction = superior-to-inferior (SI), τ_C,low/τ_C,high = 0.5/3 Hz, τ_R,low/τ_R,high = 0/0.5 Hz, and respiratory efficiency = 50%. Only 30 seconds of the motion signals are shown for brevity.