Contact impedance has an important effect on micro electrical impedance tomography (EIT) sensors compared to conventional macro sensors. In the present work, a complex contact impedance effect ratio ξ is defined to quantitatively evaluate the effect of the contact impedance on the accuracy of the reconstructed images by micro EIT. Quality of the reconstructed image under various ξ is estimated by the phantom simulation to find the optimum algorithm. The generalized vector sampled pattern matching (GVSPM) method reveals the best image quality and the best tolerance to ξ. Moreover, the images of yeast cells sedimentary distribution in a multilayered microchannel are reconstructed by the GVSPM method under various mean magnitudes of contact impedance effect ratio |ξ|. The result shows that the best image quality that has the smallest voltage error UE = 0.581 is achieved with measurement frequency f = 1 MHz and mean magnitude |ξ| = 26. In addition, the reconstructed images of cells distribution become improper while f < 10 kHz and mean value of |ξ| > 2400.