Influencia del oleaje sobre la temperatura corporal del mitílido

The algorithm has been used to control a number of electrophysiological recording setups. After describing the general experimental setup, the particular setups are discussed.

neural signal is shown in Figure 6.1. The microdrive is placed over the brain region of interest and its electrodes are lowered to the depth of the targeted neurons. The electrical signals from the electrodes are typically routed first to a headstage. The purpose of the headstage is to lower the impedance of the signal pathway from the high impedance (∼1 MΩ) of the electrode, making the pathway less susceptible to interference from electromagnetic noise sources such as the building’s 60 Hz power system [30]. Next, the signal is high-pass filtered to dampen 60 Hz and local field potential (LFP) activity [34] in order to accentuate the spikes.

The filtered signal is sampled (at 20 kHz) by an analog-to-digital (A/D) card, for example, the PCI-MIO-16-E-4 multifunction data acquisition (DAQ) card by National Instruments (Austin, TX). The input signal to the DAQ card is the difference between the signal potential and a reference potential. The reference point (i.e., ground) of the signal varies across systems. Typically, the guide tube, which surrounds the electrodes and is used to protect them during insertion through the dura, is used as a reference. It is a low impedance pathway with a large uninsulated area (compared to the electrode tip) that is theoretically at the brain’s average “resting potential,” and so differences in potential between the guide tube and the electrode tip should be due to local activity near the electrode tip. Measuring this difference should eliminate the common influence of electromagnetic noise on the signal and reference pathways (common-mode rejection [30]). In practice, noise may be inserted between the electrode signal and the guide tube reference by large currents such as those used to control the microdrive. Given the complex geometries of the current pathways, using multiple reference points might be necessary to reduce noise. Thus, in common experimental practice, the optimal referencing to minimize noise must be found through trial and error.

The system shown in Figure 6.1 could be considered a feedback control system for positioning the electrodes. The plant includes the electrodes, microdrive and neural tissue. The controller is the software described above, implementing the algorithm of the previous chapters. The filtered and sampled voltage signal and the electrode depths are the sensor inputs to the controller. The output (actuating signal) of this controller is an electrode movement command for each electrode being controlled. These commands are typically sent to a motor control unit which translates the movement commands into the proper voltages to move the electrodes, closing the feedback loop.

6.2.2

FHC Single-Electrode Microdrive

The control system was first used to control a single-electrode hydraulic microdrive by FHC, Inc. (Bowdoin, ME), as it is a standard microdrive in use by electrophysiologists. The system I used is

Figure 6.1. System diagram for the closed-loop electrode positioning system. The spe- cific microdrive shown is the Movable Array Testbed described in Section 6.3, but could be replaced with any other microdrive described in the text.

typically controlled by a hand-held remote control (Figure 6.2A). The remote control sends voltage pulses to a motor control unit, which in turn advances or retracts the hydraulic drive mechanism. In order to use the autonomous algorithm to control the microdrive, it must somehow be controlled by a computer instead of the remote control. To achieve this, the voltage pulses emitted by the remote control are mimicked by a Digital I/O card (National Instruments) connected to the motor control unit in place of the remote control.

The algorithm also requires the current depth of the electrode as an input. This position cannot be obtained simply by integrating the movement commands sent to the motor control unit for several reasons. First, the electrode depth must be initialized when the algorithm starts up. Also, commands to move the electrode beyond the end of its stroke will not be executed, and this must be known by the algorithm. There must be some feedback from the motor control unit about the current position of the electrode. As there is no depth output signal built into the FHC motor control unit other than the visual LED display, the voltage signals from the LED depth display were routed to the Digital I/O card on the computer running the algorithm. A MATLAB function was written which decodes these voltage signals into the digits displayed on the LEDs, thus giving the depth of the electrode. The complete interface with the FHC motor control unit is shown in Figure 6.2B.

Figure 6.2. System diagram for the FHC microdrive. A) The system as it is used under manual control. The remote control is used to send movement commands to the motor control unit, which in turn moves the hydraulic actuator in the microdrive. B) The system modified for closed-loop computer control. Inside the motor control unit, wires carrying the voltages to light the LEDs on the depth display are diverted to the digital I/O card on the computer. The computer decodes these voltages to determine the current electrode depth. The same digital I/O card is connected in place of the remote control and voltage pulses are sent to mimic the remote control to advance or retract the electrode. These two signal pathways (depth reading and motor control) are hardware hacks to make the manually controlled FHC microdrive computer controlled.

6.2.3

Thomas and NAN Multielectrode Microdrives

Two commercial multielectrode microdrives have also been controlled using the closed-loop system, the NAN electrode drive (NAN Instruments, Israel) and the Thomas Mini-Matrix System (Thomas Recording, Germany). They can hold sixteen and five electrodes, respectively, and, to date, a maxi- mum of five electrodes have been controlled autonomously at a time. Both systems were designed to be manually controlled using a computer interface, and so interfacing with the autonomous control software did not require hardware modifications as with the FHC system. For these microdrives, MATLAB functions were written to send movement commands through the computer’s serial port.