Skip to main content

Comparison of Automated Artifact Rejection Methods for ICU EEG Delirium Detection

Primary Investigators:
Shawniqua Williams Roberson, M.Eng., M.D.
Brief Description of Project:
Automated signal analysis of electroencephalography (EEG) is a highly sought after method of diagnosis and detection of delirium in intensive care unit (ICU) patients. However typical EEG recordings acquired in real-world settings are frought with movement and muscle artifact that grossly undermine the signal-to-noise ratio. Despite decades of research, the most reliable artifact detection methods require visual inspection, which is time consuming and resource intensive. Yet several new automated detection algorithms have recently been published. The goals of this project are (1) to build a GUI-based platform by which ICU EEG recordings can be automatically preprocessed and run through one of two novel artifact rejection algorithm, and (2) to compare the two artifact rejection algorithms in terms of their impact on delirium detection.

Desired Qualifications:
must have taken CS 2201 or CS 2204 or CS 3251 or PSY 4219
experience with MATLAB
familiarity with ACCRE
Nature of Supervision:
Remote implementation with all meetings by zoom
Weekly team meetings, weekly one-hour 1:1 with professor, daily 15-30 minute checkins
A Brief Research Plan (period is for 10 weeks):
Weeks 1-2: orientation, literature review, walk through current implementation process with senior student, 
Week 3: develop requirements
Weeks 4-5: platform design, proof of concept 
Weeks 6-7: coding and implementation
Weeks 8-9: test and documentation 
Week 10: presentation/training
Number of Open Slots: 1
Contact Information:
Name: Melissa Polson
Department: Biomedical Engineering