NOISE ELIMINATION BY THE SAVITZKY-GOLAY FILTER WITH NON-INVASIVE MONITORING OF HUMAN HEALTH

Abstract

The article highlights the details of the construction of the algorithm that helps to filter out the noise and artifacts of the musculoskeletal activity in non-invasive monitoring of the human cardiovascular system. Since the work of the cardiovascular system is an important indicator of the health of the body, monitoring its parameters is one of the main tasks for such devices. For the task two main methods are used – photo plethysmography and rheography. The article takes rheography as a basis for non-invasive monitoring. However, one of the main problems of this technology is a high sensitivity to noise. The main source of this noise is muscular activity. The peculiarity of these distortions is that they are much larger in amplitude than the information signal. This, in turn, distorts the shape of the main signal. The article presents the details of the construction of the algorithm, which helps filter out the noise and artifacts of the motor activity of the muscles and partially restore the signal. The algorithm is based on the Savitzky-Golay filter. However, the peculiarity of this filter is that it cannot track sharp drops in the level. Therefore, you first need to differentiate the signal, remove all peaks that exceed a certain level determined empirically, and then the signal is back integrated. After this operation, we retain the shape of the main signal, but we smooth the component that is responsible for muscle activity. Examples are given using this algorithm, which measures the cardiovascular system for changes in the impedance of soft tissues, signal graphs before and after processing, as well as a block diagram that implements the proposed filtering algorithm.