Breathing Pattern Monitoring by Using Remote Sensors

The ability to continuously and unobtrusively monitor and classify breathing patterns can be very valuable for automated health assessments because respiration is tightly coupled to many physiological processes. Pathophysiological changes in these processes often manifest in altered breathing patter...

Full description

Saved in:
Bibliographic Details
Published inSensors (Basel, Switzerland) Vol. 22; no. 22; p. 8854
Main Authors Kunczik, Janosch, Hubbermann, Kerstin, Mösch, Lucas, Follmann, Andreas, Czaplik, Michael, Barbosa Pereira, Carina
Format Journal Article
LanguageEnglish
Published Switzerland MDPI AG 16.11.2022
MDPI
Subjects
Abdomen
Acidosis
Algorithms
Brain research
Breathing
Cameras
Classification
contactless
dataset
Datasets
Humans
Hyperventilation
Hypoventilation
infrared thermography
Medical research
Metabolism
Monitoring
Monitoring, Physiologic - methods
Physiological aspects
Remote sensors
Respiration
Respiratory Rate - physiology
RGB video
Sensors
signal
Software
Spinal cord
Support Vector Machine
Support vector machines
Thorax
infrared thermography
contactless
classification
extraction
RGB video
dataset
signal
Online AccessGet full text

Cover

More Information
Summary:The ability to continuously and unobtrusively monitor and classify breathing patterns can be very valuable for automated health assessments because respiration is tightly coupled to many physiological processes. Pathophysiological changes in these processes often manifest in altered breathing patterns and can thus be immediately detected. In order to develop a breathing pattern monitoring system, a study was conducted in which volunteer subjects were asked to breathe according to a predefined breathing protocol containing multiple breathing patterns while being recorded with color and thermal cameras. The recordings were used to develop and compare several respiratory signal extraction algorithms. An algorithm for the robust extraction of multiple respiratory features was developed and evaluated, capable of differentiating a wide range of respiratory patterns. These features were used to train a one vs. one multiclass support vector machine, which can distinguish between breathing patterns with an accuracy of 95.79 %. The recorded dataset was published to enable further improvement of contactless breathing pattern classification, especially for complex breathing patterns.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ISSN:1424-8220
1424-8220
DOI:10.3390/s22228854