To find the optimal calibration size for auto-calibrated estimation of sensitivity maps and estimate the relative signal-to-noise ratio (SNR), fully-sampled wave-encoded k-space and fully-sampled Cartesian k-space were acquired with a quality-assurance phantom on a GE MR750 3T scanner (GE Healthcare, Waukesha, WI) using an 8-channel cardiac coil and a 3D spoiled gradient echo (SPGR) acquisition sequence. The acquisition parameters for this sequence are shown in Table 1. The sensitivity maps for wave-encoding were estimated with different sizes of wave-encoded calibration k-space (8×8 to 96×96) using the proposed auto-calibration method. Since the phantom remains static during the entire scan, the sensitivity maps estimated from the Cartesian k-space were used as the reference. The normalized root-mean-squared error (nRMSE) between the sensitivity maps for wave-encoding and Cartesian acquisition was used as a metric for determining the optimal calibration size. Another set of simulated wave-encoded k-space generated from the Cartesian k-space was also tested in the same manner to exclude the influence of wave-PSF estimation error. The SNR of fully-sampled images was estimated using the pseudo random replicas approach (40 (link)).