On the use of autoregressive time series in the modelling of Doppler signals for diagnostic purposes.

Doppler instruments are widely used in the diagnosis of vascular diseases: however, the physician generally makes the diagnosis from personal experience, and by subjective evaluation of some quantities related to blood flow (e.g., maximum or mean frequency envelope, or zero-crossing density). In order to obtain more objective criteria, methods based on either spectrum analysis, or on Laplace or other orthogonal transforms have been proposed. An alternative approach to the diagnosis of vascular alterations is proposed, which relies on the modelling of flow signals by means of autoregressive time series. The parameters of the autoregressive model have been used for the detection of alterations. The basis of the method is discussed and some preliminary results are reported.