Développement et applications de méthodes de mesure différentielle à haute résolution

Dupuis, Pascal
(2001)

Files

thes_main.pdf
  • Open Access
  • Adobe PDF
  • 2.61 MB

Details

Authors
  • Dupuis, PascalUCLouvain
    author
Supervisors
Eugène, Christian
Abstract
(en) During this research, we combined physical parameters measurement methods with numerical signal processing in order to extract indirectly physiological parameters signatures. The measures are characterized by the fact that they are differential and done at a high resolution. This permitted to acquire very low level signal variations caused by the physiological phenomenon of interest. Numerical processing permitted, starting from a very noisy raw signal, to separate the relevant information from what could be considered as noise and artefacts. Those measurement were done on systems combining, from an electrical point of view, low power consumption and low noise, in the perspective of developing ambulatory devices. Afterwards, data were acquired on a computer using a real-time environment. Two main applications have been studied. Although coming later chronologically, we will first mention combined evaluation of cardiac and respiratory rhythms, together with apnoea detection for sleep study and monitoring. The method induced low discomfort, as the sensor is a cuff used for blood pressure measurement, wrapped around the wrist and inflated at a low pressure. After the decomposition of the original signal in a number of frequency bands using wavelet filtering, we evaluated the resulting signals period, those signals representing respectively cardiac or respiratory rhythms. We showed there is a good agreement between the values provided by our numerical processing and reference values. We also established how to detect the signature of central and obstructive apnoeas. We also tried to characterize epidermis state and to detect its alterations through its heat transfer properties. This gave rise to intense activity in order to determine a heat propagation model. We established the parameters which seems the most correlated to epidermis state. The most important result is a generalization of the Prony method, permitting to clearly distinguish terms due to initial conditions from terms resulting from an applied command. Developed in a well defined framework, we still think our measurement methods and the numerical algorithms we worked out may also be applied to other engineering domains.
Affiliations

Citations

Dupuis, P. (2001). Développement et applications de méthodes de mesure différentielle à haute résolution. https://hdl.handle.net/2078.5/149476