This paper presents a brain-computer interface (BCI) designed for patients affected with Amyotrophic lateral sclerosis (ALS) or with other diseases with Bulbar-Spinal Muscular Atrophy that affect the patient's interaction with the environment. By using an EEG portable (Emotiv Epoc) we acquired neural signals which were processed and used to make a choice in the initial prototype for healthy subjects. Considering this, we worked with EEG's artifacts with the purpose of training and validating a neural network on seven different subjects. The accuracy of this technique is about 81.58% when the interface's validation is finished with the same subjects. When the validation has finished with different subjects its accuracy is 73.68%. We can conclude that the Emotiv headset is a good alternative for the development of technological devices, which purpose is to reduce the communication gap between a diseased patient and his environment.