L. Bunegin et al., DETECTION AND VOLUME ESTIMATION OF EMBOLIC AIR IN THE MIDDLE CEREBRAL-ARTERY USING TRANSCRANIAL DOPPLER SONOGRAPHY, Stroke, 25(3), 1994, pp. 593-600
Background and Purpose Cerebral embolism has been implicated in the de
velopment of cognitive and neurological deficits following bypass surg
ery. This study proposes methodology for estimating cerebral air embol
us volume using transcranial Doppler sonography. Methods Transcranial
Doppler audio signals of air bubbles in the middle cerebral artery obt
ained from in vivo experiments were subjected to a fast-Fourier transf
orm analysis. Audio segments when no air was present as well as artifa
ct resulting from electrocautery and sensor movement were also subject
ed to fast-Fourier transform analysis. Spectra were compared, and freq
uency and power differences were noted and used for development of aud
io band-pass filters for isolation of frequencies associated with air
emboli. In a bench model of the middle cerebral artery circulation, re
petitive injections of various air volumes between 0.5 and 500 mu L we
re made. Transcranial Doppler audio output was band-pass filtered, acq
uired digitally, then subjected to a fast-Fourier transform power spec
trum analysis and power spectrum integration. A linear least-squares c
orrelation was performed on the data. Results Fast-Fourier transform a
nalysis of audio segments indicated that frequencies between 250 and 5
00 Hz are consistently dominant in the spectrum when air emboli are pr
esent. Background frequencies appear to be below 240 Hz, and artifact
resulting from sensor movement and electrocautery appears to be below
300 Hz. Data from the middle cerebral artery model filtered through a
307- to 450-Hz band-pass filter yielded a linear relation between embo
li volume and the integrated value of the power spectrum near 40 mu L.
Detection of emboli less than 0.5 mu L was inconsistent, and embolus
volumes greater than 40 mu L were indistinguishable from one another.
Conclusions The preliminary technique described in this study may repr
esent a starting point from which automated detection and volume estim
ation of cerebral emboli might be approached.