sleep monitoring
Sleep is something that every person needs. Without sleep a normal days task seem never ending. Your body suffers and you suffer even people who come in contact with you suffer too. Without sleep you can't function normally. Your moods change changing your personality, changing how you perceive the world. A sleep disorder is a condition that prevents a person from getting restorative sleep causing daytime sleepiness and dysfunction. There are over 80 different types of sleep disorders but there are only four that are the most common. These disorders are Insomnia, Sleep Apnea, Restless Leg Syndrome, and Narcolepsy. Two immediate consequences of bad sleeping are fatigue (see link1, link2) and pain [1].
[1] http://www.ncbi.nlm.nih.gov/pubmed/21873265
[1] http://www.ncbi.nlm.nih.gov/pubmed/21873265
|
sleep posture reconstruction |
There is strong evidence showing the correlation between the sleeping posture and the appearance of back pain [2]
Detecting the sleeping posture we could prevent pain but it can be a hard task. Loading a person with numerous wearable sensors would be expensive and almost impossible to deploy at home. Non wearable sensors are a better option, but they would need to be able sense through the blankets. We work with thermal imaging to achieve see if we can see though the covers.
For example, figure 1 depicts the thermal image of a person sleeping covered with a blanket. Here, the silhouette can be reconstructed. This is possible thanks to the transmission of the person's heat through the cover. We are working on automatic methods to extract the skeleton from such a silhouette.
[2] http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1119282/
Detecting the sleeping posture we could prevent pain but it can be a hard task. Loading a person with numerous wearable sensors would be expensive and almost impossible to deploy at home. Non wearable sensors are a better option, but they would need to be able sense through the blankets. We work with thermal imaging to achieve see if we can see though the covers.
For example, figure 1 depicts the thermal image of a person sleeping covered with a blanket. Here, the silhouette can be reconstructed. This is possible thanks to the transmission of the person's heat through the cover. We are working on automatic methods to extract the skeleton from such a silhouette.
[2] http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1119282/
|
respiratory rate |
We are developing new techniques to diagnose and triage common respiratory conditions. In particular we will extract respiratory rate, chest excursion, and time trends from thermal imaging videos. The videos 1 and 2, were taken with a thermal camera and show how respiratory frequency could be extracted. The footages were recorded in home settings and total darkness.
1. This video shows a baby. On the left, the original footage. On the right the result of applying a de-noising and eulerian magnification [1] filters. Now, recovering the numerical breathing rate is straight forward by tracking the accumulation of heat in the whole magnified video.
[1] http://people.csail.mit.edu/mrub/vidmag/ |
2. This video shows an adult sleeping. On the left, the original footage. On the right the result of applying a de-noising and eulerian magnification filters.
|
|
sleeping cycle |
Untitled from GuidoPusiol on Vimeo. |
We aim at reconstructing the sleeping cycle through the analysis of postures. The video (left) shows a person sleeping during 6 hours. We have developed an algorithm to summarize a whole night into a meaningful graph. The graph of figure 2 shows clusters of the person's posture changes over time.
|
{1} Medical Infrared Imagine, Principles and Practices Mary Diakides , Joseph D . Bronzino, and Donald R. Peterson CRC Press 2012