Àá½Ã¸¸ ±â´Ù·Á ÁÖ¼¼¿ä. ·ÎµùÁßÀÔ´Ï´Ù.
KMID : 1144120110010040220
Biomedical Engineering Letters
2011 Volume.1 No. 4 p.220 ~ p.225
Development and preliminary evaluation of video analysis for detecting Gross movement during sleep in children
Okada Shima

Ohno Yuko
Shimizu Sachiko
Kato-Nishimura Kumi
Mohri Ikuko
Taniike Masako
Makikawa Masaaki
Abstract
Purpose: Obtaining an adequate amount of sleep is a necessity for normal childhood development. Gross body movements (GMs) during sleep in normal children vary by sleep stage and developmental state, and abnormalities in GMs characterize specific developmental pathophysiologies. Therefore, GM monitoring in children is of great importance when tracking normal development. While videosomnography is a widely used technique for monitoring GMs, it is qualitative, and a quantitative method for assessing GMs is required. We developed a novel, simple noncontact method based on video analysis for the detection of GMs during sleep.

Methods: Our method used image difference processing of videos to detect GMs. We conducted a preliminary evaluation of our technique by assessing the relationships between sleep stages determined by polysomnography and GMs assessed by the video images of 14 young children (3?7 years old).

Results: Our experiments suggest that this method may be used to detect GMs during sleep in children. The rates of GMs during sleep in children significantly differed among wake, light sleep, and slow wave sleep stages. However, rapid eye movement sleep was difficult to distinguish, probably due to age-related changes in neurological development.

Conclusions: Our novel method could detect GMs during sleep by difference processing of video images, and we demonstrated the possible utility of this technique for assessing sleep in children. Our proposed technique requires further evaluation, but we suggest that GM detection via video images may be more useful than the presently available qualitative methods.
KEYWORD
Sleep, Body movement, Video analysis, Young children
FullTexts / Linksout information
 
Listed journal information