Modeling and estimation for a wearable size sensor to monitor lower leg swelling

This paper presents a novel wearable leg size sensor for early detection of acute decompensated heart failure. The size sensor is based on the measurement of the magnetic field of an inductor and its variation with size. It has a simple and inexpensive construction involving two 3-axis magnetic sensors, an inductor and a microprocessor. In order to enable robust size estimation during rough real world use of the sensor, three estimation algorithms are developed based on models of the magnetic field variation over a cylindrical surface. The magnetic field models developed include those for a dipole and for a uniformly magnetized cylinder. The estimation algorithms used include a linear regression equation, a nonlinear regression equation and an extended Kalman filter. Experimental in vitro tests show that the size sensor performs accurately, with sub-millimeter estimation errors. The uniformly magnetized cylinder model yields the best performance that is robust to bias and misalignment errors. The developed size sensor can also be used for a number of other biomedical applications as well.