In this study, an image reconstruction algorithm for magnetic resonance electrical impedance tomography (MREIT) is proposed to achieve maximum benefit of optimum current injection patterns. By doing so, considerable reduction in probing current amplitude could be possible. In the proposed algorithm, field of view (FOV) is divided into a number of segments. Image of each segment is reconstructed separately, based on measurements obtained using the best (optimum) current patterns, which maximize distinguishability for the same segment. Images reconstructed individually for all segments are then merged to form an image of the entire FOV. The proposed regional image reconstruction (RIR) algorithm is evaluated with simulated measurements obtained from a conductivity phantom having Shepp-Logan head phantom geometry. Smaller reconstruction errors and perceptively better images are obtained by using RIR instead of conventional reconstruction (CR). Improvement is more significant for small inhomogeneities which are away from the outer surface. When SNR is 13 dB, conductivity error for small inhomogeneities reconstructed by RIR is almost half of the errors of CR.