Method and apparatus for monitoring of a human or animal subject
A method and apparatus for monitoring a human or animal subject in a room using video imaging of the subject and analysis of the video image to detect and quantify movement of the subject and to derive an estimate of vital signs such as heart rate or breathing rate. The method includes techniques for de-correlating global intensity variations such as sunlight changes, compensating for noise, eliminating areas not of interest in the image, and quickly and automatically finding regions of interest for detecting subject movement and estimating vital signs. A logic machine is used for interpreting detected movement of the subject, and an artificial neural network is used to calculate a confidence measure for the vital signs estimates from signal quality indices. The confidence measure may be used with a normal density filter to output estimates of the vital signs.
1. A method of monitoring a human or animal subject comprising the steps of:
capturing a video image of the subject consisting of a time series of image frames each image frame comprising a pixel array of image intensity values; and
analysing the video image to determine automatically a vital sign estimate of one or more vital signs of the subject; wherein the step of determining automatically the vital sign estimate of one or more vital signs of the subject comprises:
analysing the video image to detect a plurality of signals each comprising the variation in the image intensity values with time at each of a respective plurality of positions in the video image,
determining a first plurality of signal quality indices of the signals and retaining only those signals whose signal quality indices are above a predetermined threshold,
analysing the retained signals in a multi-dimensional component analysis to obtain components thereof and retaining a predetermined number of the strongest components,
determining a second plurality of signal quality indices of the retained components,
selecting amongst the retained components on the basis of the second plurality of signal quality indices,
determining a confidence value and the frequency of each of the selected components, and
outputting the vital sign estimate based on said determined frequencies, wherein the determined frequency of each selected component contributes to the vital sign estimate with a weight determined by the confidence value of the selected component.
2. A method according to claim 1 wherein each of said plurality of signals comprises the variation in intensity at a plurality of pixels in a local neighbourhood in each of said image frames whose image intensity values are combined together to form one of said plurality of signals.
3. A method according to claim 1 wherein said positions are positions in the video image at which subject movement has been detected.
4. A method according to claim 1 further comprising the step of frequency filtering each of said plurality of signals to exclude those outside predetermined expected physiological range for said one or more vital signs.
5. A method according to claim 1 further comprising the step of calculating the logarithm of said image intensity values to form a logarithm image and said step of analysing the video image is performed on said logarithm image.
6. A method according to claim 1 wherein said first plurality of signal quality indices comprise measures of periodicity, for example one or more of peak consistency and amplitude consistency.
7. A method according to claim 1 wherein said step of multi-dimensional component analysis comprises decomposing the retained signals into their components, for example by one of principal component analysis or independent component analysis.
8. A method according to claim 1 wherein said second plurality of signal quality indices comprise measures of one or more of: periodicity, spatial distribution within the video image, uniformity and variability of the component, amount of movement associated with the component.
9. A method according to claim 1 further comprising the step of determining from said second plurality of signal quality indices a confidence value for each of said retained components and using said confidence value in said selecting step.
10. A method according to claim 9 wherein said selecting step comprises weighting said retained components by said confidence value and updating a prior estimate of said one or more vital signs by said weighted components.
11. A method according to claim 10 further comprising the step of down-weighting said prior estimate of said one or more vital signs by a predetermined amount before updating it with said weighted components.
12. A method according to claim 11 further comprising the step of detecting the amount of subject movement in the video image and varying said predetermined amount in dependence upon the detected amount of subject movement.
13. A method according to claim 11 wherein the amount of subject movement is detected by determining the amount of variation in image intensity with time at each of said respective plurality of positions in the video image.
14. A method according to claim 9 wherein said step of determining from said second plurality of signal quality indices a confidence value for each of said retained components is found using a machine learning technique.
15. A system for monitoring a human or animal subject in accordance with the method of claim 1 , comprising:
a video camera adapted to capture a video image of the subject;
a display;
a video image processing unit adapted to process the video image in accordance with the method of claim 1 .
16. A method of monitoring a human or animal subject comprising
capturing a video image of the subject consisting of a time series of image frames each image frame comprising a pixel array of image intensity values; and
analysing the video image to determine automatically one or more vital signs of the subject; wherein the determining automatically one or more vital signs of the subject comprises:
analysing the image to detect a plurality of signals each comprising the variation in image intensity with time at each of a respective plurality of positions in the image,
determining a first plurality of signal quality indices of the signals and retaining only those signals whose signal quality indices are above a predetermined threshold,
analysing the retained signals in a multi-dimensional component analysis to obtain components thereof and retaining a predetermined number of the strongest components,
determining a second plurality of signal quality indices of the retained components,
selecting amongst the retained components on the basis of the second plurality of signal quality indices,
determining the frequency of the selected components, and
outputting a vital sign estimate based on said determined frequencies; and
determining from said second plurality of signal quality indices a confidence value for each of said retained components and using said confidence value in said selecting,
wherein said selecting comprises weighting said retained components by said confidence value and updating a prior estimate of said one or more vital signs by said weighted components.
17. A method according to claim 16 wherein said selecting step comprises weighting said retained components by said confidence value and updating a prior estimate of said one or more vital signs by said weighted components.
18. A method according to claim 17 further comprising the step of down-weighting said prior estimate of said one or more vital signs by a predetermined amount before updating it with said weighted components.
19. A method according to claim 18 wherein at least one of:
the method further includes the step of detecting the amount of subject movement in the video image and varying said predetermined amount in dependence upon the detected amount of subject movement; and
the amount of subject movement is detected by determining the amount of variation in image intensity with time at each of said respective plurality of positions in the video image.
20. A method according to claim 16 wherein said step of determining from said second plurality of signal quality indices a confidence value for each of said retained components is found using a machine learning technique.