In Parkinson’s disease (PD), the person has problems in the coordination of movements, as these tend to be slow at the start and execution (bradykinesia), tremor even when the person is at rest and muscle stiffness. This disease is due to a loss of pigmented substantia nigra (SNR) and other cell cores. One of the medical solutions to the disease is the surgical procedure, where it is needed to accurately locate some brain areas, to excite, injure or implanting stem cells, or generally locate targets for treating neurological disorders and thus possibly alleviating symptoms of PD. This requires the intervention of specialists formed by neurophysiologists and neurosurgeons. The identification of brain signals from microelectrode recording (MER) , is a key process in deep brain stimulation (DBS English) among patients with Parkinson’s disease (PD). This paper presents an approach for optimal representation of MER signals is presented by the method of Frames , obtaining coefficients that minimize the Euclidean norm of order 2. From the optimized coefficients of an extraction signal features dictionaries Wavelet Packet combining (WP) and the cosine is made. The purpose is to identify with high accuracy a brain structure called the subthalamic nucleus (STN) , since it is the most common target structure is where the greatest therapeutic results are achieved DBS . For a comparison frame , signals are also characterized using the discreteWavelet transform (DWT) with different stem functions. The proposed methodology is validated in a real database , for which simple supervised learning machines such as Classifier K -Nearest Neighbors (K -NN 1 and 3), the Linear Bayesian Classifier (LDC) and used Quadratic (QDC). Classification results obtained with the proposed method significantly improves the performance achieved with the DWT , making a positive identification of superior 97;6%STN . These results show that the method of Frames accurately describes the dynamic behavior of the MER signals and allow a more discriminating characterization between classes, which is essential to achieve satisfactory therapeutic results in patients.