Related Experiment Video
Updated: May 24, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.6K
Chin EMG Scalogram-Based Deep CNN for OSA Screening.
Summary
A deep convolutional neural network effectively screens for Obstructive Sleep Apnea (OSA) events by analyzing breathing patterns. This AI model distinguishes between OSA events and normal breathing, aiding in faster diagnosis and treatment.
Area of Science:
- Neurology
- Sleep Medicine
- Artificial Intelligence
Background:
- Obstructive Sleep Apnea (OSA) is a prevalent sleep disorder.
- It is characterized by breathing cessations due to upper airway muscle relaxation.
- Changes in Chin Electromyography (EMG), airflow, and oxygen saturation signal these events.
Purpose of the Study:
- To develop and evaluate a deep convolutional neural network (DCNN) for screening Obstructive Sleep Apnea (OSA) events.
- To differentiate between OSA events and normal breathing in OSA subjects.
- To analyze motor neuron firing patterns during OSA and non-OSA events.
Main Methods:
- Utilized data from 5 OSA subjects from the American Center for Psychiatry and Neurology (ACPN) database.
- Developed a deep convolutional neural network (DCNN) architecture.
- Investigated motor neuron firing patterns using EMG signals.
Main Results:
- Achieved a validation accuracy of 80% and a testing accuracy of 75% for OSA event screening.
- Observed significantly lower motor neuron firing patterns during OSA events, indicating muscle relaxation.
- Demonstrated high motor neuron activity during non-OSA events.
Conclusions:
- The proposed DCNN system effectively discriminates between OSA and non-OSA events.
- This facilitates prompt diagnosis and treatment for Obstructive Sleep Apnea patients.
- Analysis of motor neuron activity provides insights into the physiological mechanisms of OSA.
