Magnetic detection electrical impedance tomography (MDEIT) is an imaging modality that aims to compute the cross-sectional distribution of the conductivity inside a volume. The current is injected into the volume by the surface electrodes and the resulting magnetic fields surrounding the object are detected by coils. The image resolution and contrast in MDEIT image reconstruction are affected by the parameters such as the numbers and locations of electrodes and measurements, and the finite-element mesh resolution. This paper addresses the numerical experiment applied to the singular value analysis (SVA) of the sensitivity matrix in the presence of noisy measurements, subsequently suggesting the optimal electrode and detector configurations for the whole imaging object region. For the region of interest (RoI), the combined SVA and redundancy reduction is used to obtain the optimum measurement arrangement. Finally, the optimum design is confirmed by examining the image reconstructions of the simulated data acquired with different measurement arrangements. The results indicate that properly increasing the number of current injections and the number of measurement circles, and locating preferentially the electrodes and detectors on the region nearest to the RoI produce more useful singular values and better reconstructed images. These results provide guidelines for the design of the MDEIT experimental system.