Signal processing and pattern recognition for nocturnal polysomnography sleep studies
Abstract/Contents
- Abstract
- In this dissertation, I show how signal estimation and classification techniques, combined with visual interaction and receiver operating characteristics (ROC) studies, a commonly used statistical analysis method, can be used to investigate polysomnography (PSG) based sleep studies (and measures) from large, diverse populations for genetic, medical, and clinically relevant purposes. I do this by considering four problems currently faced by the sleep research community and developing the signal processing, classification, optimization, and visualization measures needed for each. These problems include: (1) improving diagnostic criteria for narcolepsy using clinical and PSG measures; (2) selecting electroencephalogram (EEG) power spectral density phenotypes for genome wide association (GWAS) (3) dependably detecting and classifying periodic leg movements (PLM) in sleep; (4) measuring rapid eye movements in patients with post traumatic stress disorder and major depressive disorder. ROC theory was extended and a combinatorial, iteratively bounded search method presented and used to optimize diagnostic testing (both parameter cutpoints and configuration) in a tool we called softROC. The Stanford EEG Viewer (SEV), a MATLAB toolbox, is developed to graphically analyze individual sleep studies and automate analysis of collections of sleep studies. The SEV provided the framework necessary to develop and optimize a new PLM classification algorithm, which implements a novel two pass, variable threshold calculation base on the current noise floor calculation. This algorithm was validated using human scored PLM data from a healthy cohort and a cohort with known sleep disorders. Time locked analysis of PLM with respiratory events illuminated the prevalence of PLM apart from respiratory events. I provide empirical evidence for improving PLM measures that are currently being put into clinical practice. Lastly, several eye movement algorithms are evaluated and a new tracking approach is developed which uses a wavelet denoised signal estimate of movement using two ocular channels (horizontal and vertical). Eye movements are characterized by activity and position and examined progressively by sleep cycle and elapsed sleep in combat veterans across four consecutive sleep studies.
Description
| Type of resource | text |
|---|---|
| Form | electronic; electronic resource; remote |
| Extent | 1 online resource. |
| Publication date | 2013 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Associated with | Moore, Hyatt IV |
|---|---|
| Associated with | Stanford University, Department of Electrical Engineering. |
| Primary advisor | Mignot, Emmanuel |
| Thesis advisor | Mignot, Emmanuel |
| Thesis advisor | Shenoy, Krishna V. (Krishna Vaughn) |
| Thesis advisor | Widrow, Bernard, 1929- |
| Thesis advisor | Woodward, Steven (Steven H.) |
| Advisor | Shenoy, Krishna V. (Krishna Vaughn) |
| Advisor | Widrow, Bernard, 1929- |
| Advisor | Woodward, Steven (Steven H.) |
Subjects
| Genre | Theses |
|---|
Bibliographic information
| Statement of responsibility | Hyatt Moore, IV. |
|---|---|
| Note | Submitted to the Department of Electrical Engineering. |
| Thesis | Ph.D. Stanford University 2013. |
| Location | electronic resource |
Access conditions
- Copyright
- © 2013 by Hyatt Errol Moore
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...