The goal of this paper is to implement a new minimum mean-squared error (MMSE) beamformer for coherent plane-wave compounding. In ultrasound imaging, an beamformer, in the MMSE approach, is usually derived as a multiplication of the minimum variance distortionless response (MVDR) output and a scalar, which is approximated by the coherent factor (CF). Beside the spatial smoothing adopted for estimating the data covariance matrix in MVDR beamforming, this approximation could be another factor that potentially reduces performance of the MMSE. In this paper, we extend the spatial coherent approach that we developed previously for MVDR beamforming, and show how this scalar can be estimated through an array of MVDR outputs. Advances of the proposed MMSE are evaluated on datasets available on PICMUS website. The imaging results show it offers improvements, in terms of spatial and contrast resolution, over the corresponding MVDR and a combination of MVDR beamforming and CF in the literature.