Signal processing for brain temperature change detection by noninvasive measurement: An independent component analysis approach

Abstract

SUMMARYThe distribution of temperature inside the brain is an important piece of information for investigating brain activity. Up to now, various kinds of noninvasive measurement techniques have been proposed. Among them, there is a technique that treats the human head as a dielectric material and then makes use of the temperature dependence of this material. One of its major problems is that accurate numerical measurements of the brain temperature might not be possible because the entire brain is surrounded by the skull, which has a relatively low permittivity. It has been reported that when blood flow stops due to brain death, the brain temperature changes in response to external temperature variations, and the response of the brain and skull are thought to differ. This paper focuses on this point in particular, and attempts to separate the brain and skull signals based on an independent component analysis (ICA) approach. In order to mimic this situation, a full human head model is constructed that incorporates the two components from the brain and skull, and the differences in the impedance per unit length in response to external variations in temperature are treated as the original signals. Two different electrode configurations that have different brain and skull component ratios are provided, and the total potential difference between the electrodes when they are connected to a constant current power source is measured.