Enabling computer decisions based on EEG input

Culpepper, Benjamin J.; Keller, Robert M.

Home/Archive/NASA NTRS

NASA NTRS · 20050154891

Enabling computer decisions based on EEG input

Culpepper, Benjamin J.·Keller, Robert M.

View original source ↗Export metadata ↓

Abstract

Multilayer neural networks were successfully trained to classify segments of 12-channel electroencephalogram (EEG) data into one of five classes corresponding to five cognitive tasks performed by a subject. Independent component analysis (ICA) was used to segregate obvious artifact EEG components from other sources, and a frequency-band representation was used to represent the sources computed by ICA. Examples of results include an 85% accuracy rate on differentiation between two tasks, using a segment of EEG only 0.05 s long and a 95% accuracy rate using a 0.5-s-long segment.

Explore related subjects

Validation Studies Evaluation Studies Electroencephalography/methods User-Computer Interface Artificial Intelligence Cognition/physiology Brain/physiology Algorithms Neural Networks (Computer)Male Reproducibility of Results Humans Principal Component Analysis Adolescent Evoked Potentials/physiology Sensitivity and Specificity

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Culpepper, Benjamin J., Keller, Robert M.. 2003-12-01. Enabling computer decisions based on EEG input. https://ntrs.nasa.gov/citations/20050154891

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related reports

NASA NTRS2005

Modeling regulation of cardiac KATP and L-type Ca2+ currents by ATP, ADP, and Mg2+

Changes in cytosolic free Mg(2+) and adenosine nucleotide phosphates affect cardiac excitability and contractility. To investigate how modulation by Mg(2+), ATP, and ADP of K(ATP) and L-type Ca(2+) channels influences excitation-contraction coupling, we incorporated equations for intracellular ATP and MgADP regulation of the K(ATP) current and MgATP regulation of the L-type Ca(2+) current in an ionic-metabolic model of the canine ventricular myocyte. The new model: 1), quantitatively reproduces a dose-response relationship for the effects of changes in ATP on K(ATP) current, 2), simulates effects of ADP in modulating ATP sensitivity of K(ATP) channel, 3), predicts activation of Ca(2+) current during rapid increase in MgATP, and 4), demonstrates that decreased ATP/ADP ratio with normal total Mg(2+) or increased free Mg(2+) with normal ATP and ADP activate K(ATP) current, shorten action potential, and alter ionic currents and intracellular Ca(2+) signals. The model predictions are in agreement with experimental data measured under normal and a variety of pathological conditions.

Validation Studies↗

NASA NTRS2004

Evaluation of the Paratrend Multi-Analyte Sensor for Potential Utilization in Long-Duration Automated Cell Culture Monitoring

BACKGROUND: Compact and automated sensors are desired for assessing the health of cell cultures in biotechnology experiments. While several single-analyte sensors exist to measure culture health, a multi-analyte sensor would simplify the cell culture system. One such multi-analyte sensor, the Paratrend 7 manufactured by Diametrics Medical, consists of three optical fibers for measuring pH, dissolved carbon dioxide (pCO(2)), dissolved oxygen (pO(2)), and a thermocouple to measure temperature. The sensor bundle was designed for intra-vascular measurements in clinical settings, and can be used in bioreactors operated both on the ground and in NASA's Space Shuttle and International Space Station (ISS) experiments. METHODS: A Paratrend 7 sensor was placed at the outlet of a bioreactor inoculated with BHK-21 (baby hamster kidney) cells. The pH, pCO(2), pO(2), and temperature data were transferred continuously to an external computer. Cell culture medium, manually extracted from the bioreactor through a sampling port, was also assayed using a bench top blood gas analyzer (BGA). RESULTS: Two Paratrend 7 sensors were used over a single cell culture experiment (64 days). When compared to the manually obtained BGA samples, the sensor had good agreement for pH, pCO(2), and pO(2) with bias (and precision) 0.005(0.024), 8.0 mmHg (4.4 mmHg), and 11 mmHg (17 mmHg), respectively for the first two sensors. A third Paratrend sensor (operated for 141 days) had similar agreement (0.02+/-0.15 for pH, -4+/-8 mm Hg for pCO(2), and 24+/-18 mmHg for pO(2)). CONCLUSION: The resulting biases and precisions are com- parable to Paratrend sensor clinical results. Although the pO(2) differences may be acceptable for clinically relevant measurement ranges, the O(2) sensor in this bundle may not be reliable enough for the ranges of pO(2) in these cell culture studies without periodic calibration.

Validation Studies↗

NASA NTRS2004

Goal-directed ultrasound in the detection of long-bone fractures

BACKGROUND: New portable ultrasound (US) systems are capable of detecting fractures in the remote setting. However, the accuracy of ultrasound by physicians with minimal ultrasound training is unknown. METHODS: After one hour of standardized training, physicians with minimal US experience clinically evaluated patients presenting with pain and trauma to the upper arm or leg. The investigators then performed a long-bone US evaluation, recording their impression of fracture presence or absence. Results of the examination were compared with routine plain or computer aided radiography (CT). RESULTS: 58 patients were examined. The sensitivity and specificity of US were 92.9% and 83.3%, and of the physical examination were 78.6% and 90.0%, respectively. US provided improved sensitivity with less specificity compared with physical examination in the detection of fractures in long bones. CONCLUSION: Ultrasound scans by minimally trained clinicians may be used to rule out a long-bone fracture in patients with a medium to low probability of fracture.

Validation Studies↗