Functional Electrical Stimulation (FES) provides a promising technology for rehabilitation of upper-limb motor functions following stroke. It enables activation of individual muscles to assist restoration of impaired muscle synergies toward normal patterns. However, there lacks a systematic approach to optimize the FES stimulation patterns delivered to patients with stroke. Our preliminary work demonstrated that it is feasible to use muscle synergy patterns to guide the generation of FES patterns. Here, we present the methodology of customizing synergy-based FES using parameterized formulae with three strategies: weight-sensitive, variability-sensitive, and duration-sensitive. Each of them is comprised of two parameter sets, which represent different directions of parameter search. Two patients with ischemic stroke were recruited to participate in the preliminary test of these strategies. Preliminary results indicate that all strategies could increase the peak velocity in reaching movements, but only the "variability-sensitive" strategy restrained unwanted shoulder excursions. This pilot study demonstrates the feasibility to explore in the parameter space the directions, along which the clinical benefit of synergy-based FES can be tracked and continuously optimized.