Recent progress has been made in bi-directional peripheral nerve interfaces, which is showing significant potential for restoring dexterity and providing tactile sensory feedback to upper-limb amputees. However, while most advancements have focused on expanding the range of sensations that intraneural intrafascicular electrodes can elicit in real-time experiments, the accuracy of motor intention decoding has not yet reached the level required for real-time application. In this paper, we present a spiking neural network computational model for pre-processing and decoding motor commands from Transverse Intrafasicular Mulichannel Electrodes (TIMEs), optimized to improve accuracy, reduce the computational complexity of the decoder and adhere to real-time constraints. The network has been designed to support a fully event-based neuromorphic processing architecture, suitable for implementation on compact, low-power hardware. We demonstrate the benefits of event-based computation, with results showing the successful decoding of four gestures from TIME signals of a transradial amputee. The accuracy of the classification increases from 63.03% ± 6.92% of a conventional SVM-based pipeline to 73.03% ± 2.18% of the event-based approach with a spiking neural network.