Applying video magnification techniques to the visualization of blood flow

Abstract

In this thesis, we investigate the use of video magnification for the visualization and assessment of blood flow. We address the challenge of low signal-to-noise ratios in video magnification by modeling the problem and developing an algorithm for measuring the SNR in the context of video magnification. We demonstrate that the algorithm can be used to estimate the SNR of a real video and predict the SNR in the magnified video. We use several techniques based on video magnification to visualize the blood flow in a healthy hand and a hand with an occluded artery, and show that these visualizations highlight differences between the hands that might be indicative of important physiological differences.