Red neuronal convolucional aplicado a la estimulación de motricidad fina con un videojuego en un paciente con hemiparesia izquierda.

Abstract

The objective of this project is to develop a video game for the recognition of the movement of the hand in a patient with left hemiparesis and thus perform control actions in order to rehabilitate the person. Algorithms of a convolutional neural network with deep learning were implemented. The network was trained with 600 images of open hands and 600 images of closed hands of a child with left hemiparesis. This allowed the classification of possible solutions to perform control actions and link to a video game on the Unity platform. The results show that there is a 99% probability for sending commands for the video game to work when the left hand is open or closed. In this way, when the patient with hemiparesis needs to open and close his left hand, he executes control actions and provokes his mobility and rehabilitation. At the end of the tests, the child felt tired by the fact that the system encouraged him to move his arm. Generally, he did not move that limb that much. It can be concluded that the system works correctly, fulfilling its objective.