The robotic scrub nurse is an important technology to assist and minimize the physical dependence of the scrub nurse in the surgical procedure through Surgical Instrument Signaling (SIS). The aim of this study is to determine if automatic feature extraction based on One-Dimensional Convolutional Neural Networks (1DCNN) can be applied to Surface Electromyography (sEMG) in the SIS context. The employed sEMG database comprises 14 SIS hand gesture contexts and ten subjects. Two 1DCNN topologies are analyzed. The K-fold cross-validation approach was employed for individual classification and cross-subject classification. In the individual classification, topology 1 reached higher accuracies than topology 2. The minimum and maximum accuracies obtained for topology 1 were 79% and 92%, respectively. For topology 2, the minimum and maximum accuracies were 49% and 83%, respectively. In the cross-subject scenario, the mean accuracy for topology 1 was 89%, and for topology 2, it was 78%. In all the cases, topology 1 presented higher results. Through the analyses, 1DCNN reached results close to those found in the literature, and we can suggest the use of 1DCNN (specifically topology 1) for the classification of SIS gestures based on the sEMG database.