Mixedmode oscillations and coherence resonance widely exist in the biological nervous system and are closely related to certain physiological functions．In this paper，a improved FitzHughNagumo neuronal model that can generate mixedmode oscillations is used to construct an electrically coupled neuronal network model．In the deterministic network model，it is shown that through increasing the coupling strength，the neuronal firing reaches complete synchronization and the mixedmode oscillation turns into period1 spiking．After introducing Gaussian white noise to the network，it is shown that，when the coupling strength is within a large range，the coefficient variation of interspike intervals first increases，then decreases，and further increases if increasing the noise intensity，implying that transition from anticoherence resonance to coherence resonance occurs．The results not only expands complex stochastic dynamics of the neuronal networks，but also reveals potential functions of mixedmode oscillations．