A Real-Time Neural Decoder for Decoding Perceived Odor Identity Within a Single Sniff

Neural decoding is a pivotal technique in neuroscience for reconstructing sensory stimuli from neural responses. This study presents a real-time neural decoder hardware implemented using field-programmablegate array, designed to classify inhaled odorants with minimal latency. Tested in awake mice, the system effectively sorts neural spikes from the mitral/tufted cells in the olfactory bulb. It predicts the perceived odorant based on sniff-wise population activities with a processing latency of just 2.36 $\mu$s and achieves accuracy comparable to traditional offline neural decoding methods. This advancement holds significant potential for applications in real-time neural decoding, including the closed-loop neural control experiment, offering deeper insights into the neural mechanisms underlying sensory systems.