(Sunday, 7/12/2020) [Eastern Time (GMT-4)]
Room: AAPM ePoster Library
Purpose: To detect nonvisible erythema as blood flows into and out of the face through the use of Eulerian video amplification and extract physiological data to quantify changes in blood flow over time for application in predicting skin toxicity.
Methods: High-resolution videos of the face were recorded for 5 seconds prior to and after stimulating erythema and intensifying blood flow by applying heat. MATLAB code was written to implement a technique developed by the M.I.T. imaging lab to amplify colors present in video sequences and to extract quantitative information regarding blood flow and color patterns. The average pixel value within a region of interest was tracked over time and descriptive statistics were recorded. The amplitude of the periodic color changes in RGB and grayscale intensity (I) values [0 to 255] throughout the frames and the average center value of the envelope encompassing the periodic signal were compared with independent samples t-tests.
Results: The software provided a platform to successfully analyze periodic nonvisible color variations in skin corresponding to blood flow based on the feasibility study. While imperceptible to the human eye, changes in blood flow were amplified and detectable. The center value and amplitude of the amplified signals for heated skin were 238.0 and 3.6 (R), 184.1 and 4.3 (G), and 196.2 and 3.8 (I), and non-heated skin were 251.9 and 0.7 (R), 196.3 and 1.3 (G), and 207.6 and 0.7 (I), respectively and the difference was found to be statistically significant (p <0.05).
Conclusion: Image processing techniques involving Eulerian amplification used in this feasibility study have applications in radiotherapy for detecting radiation-induced microvascular changes in the skin prior to visual changes being apparent. This technique uses video as input and returns amplified information regarding erythema and blood flow changes that can allow for earlier prediction of toxicity.
Keywords
Blood Flow, Breast, Radiation Therapy
Taxonomy
IM/TH- Informatics: Informatics in Imaging (general)
Contact Email
