Non-Ionizing Imaging
Title: Sustained Imaging of Breath
Author: Ioannis Pavlidis
Abstract
In this talk we present a method of recovering the breath waveform through thermalimaging of the face. The passive and at a distance nature of sensing enables sustained monitoring at the desktop or at bed. However, the method is fraught with significant technical challenges. Specifically, in the talk we will describe algorithms to localize the measurement area (nostrils), neutralize head motion, and extract the breath signal. We will also present relevant experimental results and sensitivity analysis.
Short Bio
Ioannis Pavlidis is the Eckhard Pfeiffer Professor of Computer Science and Director of the Computational Physiology Lab at the University of Houston. His research is funded by multiple federal agencies including NSF, DOD, DHS, NIH, and corporate sources. He has written many journal articles and books on the topics of computational physiology, computational psychology, computer vision, and pattern recognition. His collaborative work with Dr. Levine on stress quantification appeared in Nature and Lancet, and received world-wide scientific and media attention. Dr. Pavlidis has also established several well-known IEEE Conferences and Workshops, among them the IEEE Advanced Video and Signal Based Surveillance (AVSS) Conference. He is a Fulbright fellow and senior member of IEEE.
Title: Contact-free Measurement of Blood Pressure – Is it feasible?
Author: Marc Garbey
Abstract
The ability to measure the blood pressure of a human subject with passive imaging has many potential applications, including psychophysiological evaluations and patient monitoring in Intensive Care Units (ICU).
A blood pressure indicator based on thermal imaging is based on the following facts: (1) Each systolic pressure wave brings in major arteries slightly larger fluxes of “hot” blood flow than diastolic pressure thanks to the compliance of the arterial wall. (2) This periodic source of heat in blood vessels proximal to the skin can affect the skin temperature. (3) The change of temperature distribution in the tissue comes from the convection effect of capillary flow, the diffusion effect, and tissue deformation driven by the arterial wall compliance to pressure variations. (4) A high quality thermal imaging sensor has thermal sensitivity close to one hundredth of a degree Celsius and time resolution better than 1/30 sec.
There are several cases where major arteries are close to the surface, including the temporal, carotid, and radial arteries. Although, theoretically feasible, there are several challenges to blood pressure determination through thermal imaging: (1) At first approximation, the radius of an isolated artery varies linearly with the tension of the wall, which in turn is proportional to the pressure applied to the wall. However, the artery is embedded in a heterogeneous tissue environment and the mechanical response of the artery wall depends on it. One can get only a rough approximation of the tissue composition. Also, an artery is always next to a vein that carries a counter-flow. This vein that has in general quasi-steady flow and almost no pressure variation is a heat sink that is difficult to assess. (2) The skin temperature atop the superficial vessel region may be partially driven heat mechanisms that are difficult to assess due to the depth of the artery and the amount of intervening fat. Many of these issues could be resolved through modeling and calibrating the direct numerical simulation using medical imaging data from human subjects instead of indirect measurements or a phantom data. (3) A significant challenge comes from accurate localization of the artery’s thermal imprint on thermal imagery and tissue motion tracking. (4) Assuming, all the previous problems have been solved, there is still a fair amount of noise removal that is necessary due to the nature of the sensor.
In this talk, we will concentrate on the modeling of a thermally-based pressure indicator and assume that vessel localization and tracking are perfect.
Short Bio
Marc Garbey is a Professor of Applied Mathematics and Computer Science at the University of Houston. He is also the Chair of the Computer Science Department at the University of Houston. Dr. Garbey received his Ph.D. from Ecole Centrale of Lyon in 1984 and his Habilitation from University of Lyon in 1989. His research focuses on modeling of physiological systems using Partial Differential Equations
|
Comments (0)
You don't have permission to comment on this page.