Electricity: measuring and testing – Particle precession resonance – Using a nuclear resonance spectrometer system
Patent
1991-11-21
1993-07-20
Tokar, Michael J.
Electricity: measuring and testing
Particle precession resonance
Using a nuclear resonance spectrometer system
324306, G01R 3320
Patent
active
052297160
DESCRIPTION:
BRIEF SUMMARY
The invention relates to the real-time adaptive spectral analysis of unstable signals having complex values representing a physical phenomenon.
The invention can be used whenever it is necessary to analyse in real time such signals which can be represented by a self-regressing parametric model, by assuming that the coefficients of the model remain steady over short time intervals. It applies more particularly to all the fields of medical imagery, for example using NMR, and more generally whenever spectral analysis must be made on organic signals, for example in electrocardiography and electro encephalography. It finds however a particularly important application in the spectral analysis of the signals furnished by ultrasonic Doppler velocimeters for studying the blood flow in the heart cavities or the vessels.
At the present time, parametric spectral analysis is relatively little used in Doppler velocimetry and in medical ultrasonic echography. It remains at the laboratory stage. It consists in modelling the signals using a parametric model, generally of the adjusted mean self-regressing type, whose parameters are adjusted by optimization using algorithms of the least square type. This approach has the advantage of working on short signal sections but it involves using low order models and has consequently limited frequency resolution, which is particularly troublesome for the spectral analysis of very unsteady signals. Moreover, the known methods do not process the demodulated signals with complex values as such, but process each of the components separately, respectively in phase and in quadrature.
Commercial apparatuses use non parametric spectral analysis methods of Fourier analysis type so as to be able to use them with rapid Fourier transformation algorithms. But these methods require working on long sections of the signals so as to obtain a suitable estimate of the desired spectral representation, which is incompatible with the nature of the unsteady states met with in the above mentioned fields.
The invention aims more particularly at providing a spectral analysis process and device answering better than those known heretofore the requirements of practice, particularly in that they improve the frequency resolution, make possible a more accurate analysis of the signals and restrict the variability of the spectra obtained for representing flows in closely related conditions.
To this end, the invention proposes more particularly an adaptive spectral analysis process by modelling the phenomenon represented by the signal characterized in that successive sections of the sampled signal are modelled, using a high order self-regressing process, close or equal to the maximum authorized by the duration of the sections and the sampling frequency, by making an estimate of the parameter characterizing the model by rapid Kalman filtering using an algorithm which is simplified in the full limit authorized by the local steady state of the model in the section and in that the spectral power density is determined from the estimated parameters; to avoid risks of instability related to the choice of the high order, the estimate of the parameters of the self-regressing model are regularized by using a constraint which consists in minimizing a criterion of the form designating the sampled function modelling the signal via the parameters a to be determined and y the function represented by the samples of the signal and where .OMEGA.(a) is a regularizing function taking into account the a priori knowledge of the nature of the function or obtained by optimization.
Through using a high order model, very good frequency definition is obtained. The process makes it possible to obtain parameters directly for calculating the spectral power density of the signal in the analysis window. It is in particular possible to use a model order equal to the number of sampling points in the Doppler signal, which gives an excellent frequency definition, whereas the present-day spectral analysis methods using a self-regressing model and criter
REFERENCES:
patent: 4391149 (1981-05-01), Herzl
patent: 4585992 (1986-04-01), Maudsley et al.
patent: 4789832 (1988-12-01), Nagayama
patent: 5019978 (1991-05-01), Howard, Jr. et al.
Ultrasonics, vol. 24, No. 4, Jul. 1986, Butterworth & Co., Ltd., (Guilford, GB)-R. P. Williams: "On the relationship between velocity distribution and power spectrum of scattered radiation in doppler ultrasound measurements on moving suspensions", pp. 197-200.
Journal of Magnetic Resonance, vol. 61, 1985, Academic Press Inc., H. Barkhuijsen et al.: "Retrieval of frequencies, amplitudes, damping factors, and phases from time-domain signals using a linear least-squares procedure", pp. 465-481.
Arcile Claude
Demoment Guy
Herment Alain
Houacine Amrane
Mouttapa Indira
Institut National de la Sante et de la Recherche Medicale
Tokar Michael J.
LandOfFree
Process and device for real-time spectral analysis of complex un does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Process and device for real-time spectral analysis of complex un, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Process and device for real-time spectral analysis of complex un will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-1763583