 Functional near-infrared spectroscopy, FNIRS, has become increasingly popular for studying brain activity. However, current methods for signal quality control are not standardized and can be heavily reliant on manual processes. This paper proposes a deep learning approach to automatically identify poor quality signals within a dataset of FNIRS recordings. The proposed method was tested against existing manual thresholding techniques and outperformed them in terms of accuracy. Additionally, it was shown to be more robust than existing methods when applied to real data. These findings suggest that deep learning could provide a more reliable and efficient way to detect poor quality signals in FNIRS data. This article was authored by Andrea Bezago, Michelle Nao, Giulio Gabrielli, and others.