The U-net is a deep-learning network model that has been used to solve a\nnumber of inverse problems. In this work, the concatenation of two-element\nU-nets, termed the W-net, operating in k-space (K) and image (I) domains, were\nevaluated for multi-channel magnetic resonance (MR) image reconstruction. The\ntwo element network combinations were evaluated for the four possible\nimage-k-space domain configurations: a) W-net II, b) W-net KK, c) W-net IK, and\nd) W-net KI were evaluated. Selected promising four element networks (WW-nets)\nwere also examined. Two configurations of each network were compared: 1) Each\ncoil channel processed independently, and 2) all channels processed\nsimultaneously. One hundred and eleven volumetric, T1-weighted, 12-channel coil\nk-space datasets were used in the experiments. Normalized root mean squared\nerror, peak signal to noise ratio, visual information fidelity and visual\ninspection were used to assess the reconstructed images against the fully\nsampled reference images. Our results indicated that networks that operate\nsolely in the image domain are better suited when processing individual\nchannels of multi-channel data independently. Dual domain methods are more\nadvantageous when simultaneously reconstructing all channels of multi-channel\ndata. Also, the appropriate cascade of U-nets compared favorably (p < 0.01) to\nthe previously published, state-of-the-art Deep Cascade model in in three out\nof four experiments.\n