Predation is a common and important species interaction in many ecological systems. Mathematical models are important and very useful tools to understand and analyze the dynamic behavior of the predator-prey systems. Classical predator-prey models such as the Lotka-Volterra model and Leslie-Gower model can be used in a homogeneous environment. However, in general, the environment is heterogeneous and this can be represented using a set of discrete patches connected by migration. In the simplest situation, the two patches predator-prey model is used, which is composed of the local population process and the dispersal from patch to patch. Prey escaping predation in space or time is widely observed. Spatial or temporal refuges are well-known examples of this class of mechanism. Most theoretical studies have focused on how refuges add stability to the system with predator-prey interactions. The role of spatial and temporal refuges on species coexistence in communities with intraguild predation has also been investigated. These studies have two characteristics. First, most of the research is based on the Lotka-Volterra framework, only a few on the Leslie-Gower framework. Second, the traditional way to model this is to modify the functional response of predators and consider prey refuges, implicitly. The prey refuge can positively affect the growth of prey and negatively that of predators, because the decrease of predation success can lead to the reduction of prey mortality. On the other hand, the hiding behavior of prey could be either advantageous or detrimental for the involved populations. For example, the prey population in the refuges has a low or even negative growth rate because they are rarely offered feeding or mating opportunities. That is, there is a different population structure between the prey refuge and the normal habitat patch. Therefore, it is more reasonable to consider prey refuge explicitly. Based on the above consideration, this paper proposes a Leslie-Gower predator-prey model incorporating the effect of prey refuge explicitly in a two-patch environment. We assume that the prey dispersal is at a constant migration rate and is faster than the local predator-prey interaction, while the predator cannot disperse between patches. Taking advantage of the two different time scales, we use aggregation methods to obtain a reduced (aggregated) model governing the total prey and predator densities. The mathematical analysis of the aggregated model shows that there exists a unique and globally stable positive equilibrium under a certain condition. Simple mathematical analysis shows that increasing the amount of refuge can increase predator densities. As far as the prey species is concerned, under a certain condition, there exists a threshold, such that, for the prey refuge smaller than this threshold, increasing the amount of prey refuge can increase the prey densities, and if the prey refuge is larger than this threshold, increasing the amount of prey refuge can decrease the prey densities. Furthermore, prey and predator will become extinct when the effect of prey refuge is strong enough. In contrast to previous research, the effects of refuges used by the prey are different in spatially implicit and spatially explicit models. The discrepancy suggests that the spatial structure is important when refuges are included.