Laser Doppler flowmetry (LDF) is a noninvasive method to monitor skin perfu
sion and is very useful in studying reactive hyperemia signals. For this la
tter case, the determination of peak flow (pLDF) and time to peak flow (tpL
DF) is of great interest in discriminating between subjects with peripheral
arterial obliterative diseases and those who are healthy. However, the myo
genic mechanism provokes marked oscillations on all LDF signals. Therefore,
an accurate detection of the parameters is very difficult. The present stu
dy shows that wavelets are a powerful tool to overcome this drawback. Six d
ifferent processes using several wavelets are tested on 44 signals recorded
on 11 healthy volunteers. The results prove that thresholding coefficients
of a multilevel wavelet decomposition does not allow a valuable computatio
n of the parameters but that the reconstruction of the approximation branch
is an efficient method to accurately determine pLDF and tpLDF. Using this
latter method, mt:an results for a 3-min occlusion give pLDF = 46.80 a.u. a
nd tpLDF = 17.08 s. For a 2-min occlusion, pLDF and tpLDF are 39.19 a.u. an
d 11.63 s, respectively. For a 1-min occlusion, the results give pLDF = 36.
01 a.u., tpLDF = 8.48 s. Eventually, for a 30-s occlusion they give pLDF =
33.86 a.u. and tpLDF = 5.60 s. These results can now be compared with those
obtained on pathological subjects. (C) 2000 Academic Press.