P16071: MRI-Compatible EEG Cap
/public/

Subsystem Build & Test

Table of Contents

Team Vision for Subsystem Level Build & Test Phase

Subsystem Level Build & Test Directory

Team Vision for Week 5:

During the subsystem-level build and test phase, the team planned to build, program, and verify the functionality of the microprocessor and ADC. Additionally, we planned to test the mechanically spliced, bias-free modulator to verify the performance improvements expected after professional fiber splicing. Finally, we intended to finish the PCB design, finalize and order the laser components, and complete MRI training with the eventual intent of performing materials testing.
Unfortunately, the inability of the bias-free modulator to perform as anticipated severely delayed our project timelines, as some of the team had to instead focus efforts on troubleshooting and brainstorming alternative system solutions. While the microprocessor was verified to be functional, the team was also unable to pass a signal through the ADC, so troubleshooting efforts are currently underway.

Test Results

SPI-UART Conversion and ADC Function Testing

The SPI-UART Conversion and ADC Function Testing Document can be found here.

Purpose:

To demonstrate the functionality of the SPI-UART conversion with the use of the MSP430 microcontroller as well as the accuracy of the analog to digital converter.

Procedure:

A 2Vp-p sine wave with a 2V offset had been applied to the first channel of the ADC. The Micro-controller had been placed closing to the ADC to ensure a stable SPI communication without error. The software used within the micro-controller can be viewed through the following flow chart:
Micro-Controller Software Flowchart

Micro-Controller Software Flowchart

Results and Conclusion:

There has not been a recorded output of the ADC given the current setup. However, it has been determined that the issue relays within the connections of the ADC rather than the output of the microcontroller. As of currently, it can be said for certain that the UART section of the communication transfer is functional. It can also be said that the master SPI output has been detected. The finding the solution to the issue remains in progress.

Ag/AgCl Electrode Testing

The Ag/AgCl Electrode Testing Document can be viewed here.

Purpose:

To evaluate the use of MRI-compatible Ag/AgCl electrodes, clips, and wires in recording EEG signals.

Procedure:

Alpha waves were first collected with the ADInstruments kit and the standard MRI-incompatible electrodes and leads. The ground electrode was placed on the left temple, positive on the right temple, and negative behind the right ear. The electrodes and wires were then replaced with the MRI-compatible Ag/AgCl setup. Wires were connected to the ADInstruments hardware by wrapping them around the EEG plugs in the BioAmp.

Results:

While the mean SNR was slightly lower with the new Ag/AgCl electrodes compared to the standard gold cup electrodes, this difference was not statistically significant.
Ag/AgCl vs Gold cup Electrode

Ag/AgCl vs Gold cup Electrode

Conclusion:

The MRI-compatible Ag/AgCl electrodes will perform comparably to the gold-cup electrodes when used in the EEG system.

Bias-Free Modulator Testing

The Bias-Free Modulator Testing Document can be viewed here.

Purpose:

To compare the performance of a bias-free modulator to that established previously in a biased modulator.

Procedure:

The bias-free modulator was introduced into the system, and the phase plates were adjusted as needed to obtain the proper light polarization. A series of laser powers and wavelengths were assessed for signal quality. A new modulator was ordered, fusion spliced, and introduced into the system as well to verify that the first modulator was functioning properly.

Results:

The minimum resolvable voltage was found to be 500 uV, which was significantly worse than the 12.5 uV recorded by the biased modulator previously.
Figure 1: First modulator noise followed by the 500 uV signal

Figure 1: First modulator noise followed by the 500 uV signal

Figure 2: Second modulator noise followed by a 1 mV signal

Figure 2: Second modulator noise followed by a 1 mV signal

Conclusion:

The bias-free modulator would be entirely unsuitable for the collection of EEG, as even the strongest EEG signals would not be resolved by this new system.

Modulator Bias Testing

The Modulator Bias Testing Document can be viewed here.

Background:

The JDSU OC-192 biased electro-optic modulator has exhibited a bias drift over time (see below). This drift is a known phenomenon in the literature and is attributed to charge buildup in the crystal. Bias control circuits are frequently used to counteract this effect; however, additional circuitry is highly undesirable for our application.
Bias drift was not studied extensively by our group during testing, as we were operating under the assumption that we would ultimately switch to a bias-free modulator. Now that we are considering a biased system, it was necessary to study this effect more in detail, as bias has a large impact on the resolution of microvolt-range signals.
Bias drift of the unfiltered data (bottom). The drift is filtered out with the application of a bandpass filter in the range of our frequency of interest (top)

Bias drift of the unfiltered data (bottom). The drift is filtered out with the application of a bandpass filter in the range of our frequency of interest (top)

Purpose:

The purpose of this experiment was to explore potential methods of counteracting this charge buildup. This is necessary since, while filtering counteracts the effect for data analysis purposes, the drift leads to changes in modulator performance over short time scales and, left uncontrolled for several minutes, can overload our amplifier.

In order to characterize the effects of bias drift, the output from the modulator was connected directly to a USB detector. When no bias is applied to the modulator, the following behavior is observed:

1mW Raw Laser

1mW Raw Laser

While the power levels here oscillate by roughly 5 uW, the noise is well distributed around a center point and no trends in performance are apparent. In an effort to reduce this noise, an optic isolator was added to the system:

1mW Isolated Laser

1mW Isolated Laser

Although the total power going through the system is reduced by a factor of 1.38, the noise amplitude was reduced by a factor of 25. This is a promising result and suggests that an optical isolator may be able to help provide higher resolution in our final system. More important to the immediate experiment, there is still no general trend in the signal level over time. This is in stark contrast to system performance when a bias voltage is applied to the system:

2kHz/2V Bias

2kHz/2V Bias

Adding bias to the modulator reduces the total amount of power that reaches the sensor (since the bias causes some of the signal to destructively interfere). Thus, while the absolute value of the power change in this case is less than that of the previous noise oscillations, the relative change over time is severe. This propensity to drift under the influence of bias voltage was born out over many trials, showing little sign of slowing, and never being observed to stop, even after tens of minutes of observation (much too long to wait given the cost of MRI time). As demonstrated by the figure below, it is plausible that the drift may slow over time, but the magnitude of the change is still too great for our system.

4mW Laser 6.88 Bias After 20

4mW Laser 6.88 Bias After 20

In an effort to combat this drift, we attempted to use a square wave signal generator to periodically charge and discharge the modulator. After testing a variety of settings, it was determined that a 0.1Hz square wave with a 9V amplitude and a +1V DC offset is able to keep the operating voltage relatively constant during the high-voltage phases, as shown below:

0.1Hz Square Wave 9V AC/1V DC

0.1Hz Square Wave 9V AC/1V DC

While there is still a large deal of decay at the top of each of the square waves, this can be filtered out after the fact and should not run the risk of overloading the amp. Unfortunately, these dramatic changes in modulator bias appear to lower its performance temporarily, hampering our ability to collect high-resolution signals. It is also not clear whether such a solution could be implemented in the final design, where the oscillatory input would need to be provided by some form of energy harvester within the MRI environment. As of this review, the search for a better solution to the bias drift problem is still underway.

Modulator Options Table

The Modulator Options Table can be viewed here.

Modulator Options Table

Modulator Options Table

PCB Schematic and Board Screenshots

General Schematic

General Schematic

Ground Schematic

Ground Schematic

Power Schematic

Power Schematic

Risk Assessment

The Risk Assessment can be viewed here.

Problem Tracking

The Problem Tracker can be viewed here.

Test Plan

The Test Plan can be viewed here.

Bill of Materials (BOM)

The updated Bill of Materials can be viewed here.

Plans for next phase

Team Vision for Phase 8:

During the week 8 demo, the team plans to, at minimum, demonstrate functionality of the assembled PCB, microprocessor, and ADC in the single-modulator system. Ideally, the team will also demonstrate the use of a three electro-optic modulator system in acquiring alpha waves; however, this may be beyond the scope of this next phase given the timeline delays. The team also plans to discuss the results of the MRI materials compatibility testing.

The MSD II schedule can be viewed here.

MSDII Schedule

MSDII Schedule

The 3 Week individual projections can be viewed here.


Home | Planning & Execution | Imagine RIT

Problem Definition | Systems Design | Subsystem Design | Preliminary Detailed Design | Detailed Design

Build & Test Prep | Subsystem Build & Test | Integrated System Build & Test | Integrated System Build & Test with Customer Demo | Customer Handoff & Final Project Documentation