DETERMINING A RESPIRATORY PATTERN FROM A VIDEO OF A SUBJECT

摘要

What is disclosed is a system and method for determining a subject's respiratory pattern from a video of that subject. One embodiment involves receiving a video comprising N≧2 time-sequential image frames of a region of interest (ROI) of a subject where a signal corresponding to the subject's respiratory function can be registered by at least one imaging channel of a video imaging device. The ROI comprises P pixels. Time-series signals of duration N are generated from pixels isolated in the ROI. Features are extracted from the time-series signals and formed into P-number of M-dimensional vectors. The feature vectors are clustered into K clusters. The time-series signals corresponding to pixels represented by the feature vectors in each cluster are averaged along a temporal direction to obtain a representative signal for each cluster. One of the clusters is selected. A respiratory pattern is determined for the subject based on the representative signal.

主权项

1. A method for determining a subject's respiratory pattern from a video, the method comprising:
receiving a video of a subject, said video comprising N≧2 image frames of a region of interest of said subject where a signal corresponding to the subject's respiratory function can be registered by at least one imaging channel of a video imaging device used to capture said video, said region of interest comprising P pixels, where P≧2; generating a plurality of time-series signals S1, . . . , SP each of duration N whose samples are values of pixels in said region of interest in said image frames; extracting, for each of said time-series signals, a set of features and forming a P-number of M-dimensional feature vectors, where M≧2, each feature vector individually quantifying an overall temporal orientation of a respective time-series signal; clustering said feature vectors into K≧2 clusters according to their temporal alignment; averaging, in a temporal direction, all time-series signals corresponding to pixels represented by said feature vectors in each of said clusters to obtain a representative signal for each cluster; selecting one of said clusters; and determining a respiratory pattern for said subject based on said selected cluster's representative signal.