Research Article
Candidate Estimators for Aorta Diameter Estimation Using Monostatic Radar
@INPROCEEDINGS{10.1145/2221924.2221949, author={Lars Solberg and Ilangko Balasingham and Svein Hamran}, title={Candidate Estimators for Aorta Diameter Estimation Using Monostatic Radar}, proceedings={5th International ICST Conference on Body Area Networks}, publisher={ACM}, proceedings_a={BODYNETS}, year={2012}, month={6}, keywords={Ultra wideband radar Medical Radar Aorta diameter estimation}, doi={10.1145/2221924.2221949} }- Lars Solberg
Ilangko Balasingham
Svein Hamran
Year: 2012
Candidate Estimators for Aorta Diameter Estimation Using Monostatic Radar
BODYNETS
ACM
DOI: 10.1145/2221924.2221949
Abstract
This work is aimed at using a radar for the estimation of central blood pressure in a non-invasive manner. For this objective, we address in this article the issue of aorta diam- eter estimation.
The problem studied consist in estimating the radius of a slowly time-varying cylinder, the aorta, in a dispersive, lossy tissue with frequency-dependent attenuation. It is assumed blood pressure may be calculated based on the radius as a function of time.
Signals are generated using an FDTD electromagnetic wave simulator with a simplified geometry. We have considered three candidate estimators: an impulse radar based on a matched filter, an FMCW radar and a least squares prob- lem using a channel model inspired by the ideal cylinder response. All estimators use a propagation model to com- pensate for propagation effects.
To achieve a relative precision of 0.1 mm, a ratio between emitter energy and receiver noise spectral density above 110 dB to 130 dB, depending on estimator, was found sufficient. Furthermore, precision for all estimators was best at lower frequencies in the range we investigated, between 1 GHz and 3 GHz. Although linearity was generally good, due to very small delay variations over the expected 1.1±0.22 mm peak-to-peak variation of aorta radius, non-linear effects is difficult to avoid, especially at the lowest frequencies. Offsets were also present depending on estimator.