MIT is poised to make a big difference in medical monitoring coming up with a method to use low quality webcams to measure pulse rate, respiration and blood oxygen levels. Previously only available through the use of invasive commercial pulse
oximetry or electrocardiogram sensors, this breakthrough method uses variations in transmitted or reflected light sources from image capture to measure cardiovascular pulse waves without a dedicated source of light (red or infrared light sources). Attempts in the past have been stymied by motion artifacts or signal noise that drown out the relevant frequencies to be measure physiologically.
The MIT breakthrough uses an open source image processing library called Open Computer Vision released through the BSD licensing (so can freely be used by non-commercial and commercial sources). This library provides the basis to stabilize captured images and to capture and account for image movement and artifacts. The captured images are separated into red, green and blue channels and then Fourier analysis is used (breaking a large function into simpler sums of trigonometric functions). These function points are assumed to be linear (pulse rates jump but the differentiation between plotted points are somewhat the same) and are measured on a Bland Altman plot (the primary application of the Bland-Altman plot is to compare two clinical measurements that each provide some errors in their measure) to come up with a mean measurement.
The experiments were done using a simple MacBook webcam and was extended to include the measurement of multiple faces in the same field of vision of the camera.
Here is a link to the full MIT study
