Functional near-infrared spectroscopy (fNIRS) has become a viable approach for brain function investigation and is an interesting modality for brain-machine interfaces (BMIs) due to its portability and resistance to electromagnetic noise. In this work, a hemodynamic response based on fNIRS signals was utilized to classify the right and left ankle joint movements. To achieve this objective, 32 optodes (emitters and detectors) were used to measure the hemodynamic responses in the motor cortex area during the motor execution task of the ankle joint movements. Two-channel sets were formed one including the channels directly related to the movement task, and another including all of the proposed channels. The results of this study reveal that the scheme based only on the selected channels outperformed the scheme that uses all channels. The classification accuracies were 91.38 and 89.86 respectively. These results demonstrated that fNIRS signal classification can be enhanced by eliminating the redundant channels. © 2021 IEEE.