Purpose: To visualize changes in microcirculation of irritated skin through Eulerian amplification of high-resolution monochrome videos correlated to known detected temperature differences by a thermal camera as a control.
Methods: MATLAB code was written to implement Eulerian amplification, a technique developed by the M.I.T. Media Lab, to exaggerate changes in skin color to make them more discernable in video sequences. In an investigator self-monitoring feasibility trial, a monochrome machine vision camera (model Edmund Optics EO-2323) with detector resolution of 1920 x 1200 pixels and frame rate of 180 fps was used to collect high-resolution video of a region of interest (ROI) undergoing circulatory changes induced by a cold compress. Heart rate was extracted from subtle color changes, then the video was amplified by a factor of 100 with a bandwidth of 0.1 Hz encompassing the heart rate. A thermal camera (model FLIR A325sc) with detector resolution of 320 x 240 pixels was used to collect static thermal images to monitor temperature changes over time to act as a standard against which the visualization of changes in blood flow by the monochrome camera could be correlated.
Results: Following removal of a cold compress, blood flow increased to the region of interest as a result of vasodilation and was noted by the increase in amplitude of non-visible skin color changes in the magnified video as time increased. These increases in amplitude were correlated to increases in temperature observed by the thermal camera.
Conclusion: The thermal camera temperature differences were mapped to blood flow changes and demonstrated the monochrome camera’s ability to detect non-visible changes in microcirculation based on amplification of subtle skin color changes. This technique was evaluated for application in radiotherapy to assess skin changes throughout treatment and detect erythema before clinical manifestation in patients at risk for skin reactions.
Blood Flow, Breast, Radiation Effects
TH- Response Assessment: Radiomics/texture/feature-based response assessment