(Stroke. 1996;27:1548-1552.)
© 1996 American Heart Association, Inc.
Articles |
Multigated Doppler Ultrasound in the Detection of Emboli in a Flow Model and Embolic Signals in Patients
the Department of Neurology, King's College School of Medicine and Dentistry, London, UK.
Correspondence to Dr Hugh Markus, Department of Neurology, Institute of Psychiatry, De Crespigny Park, London, SE5 8AF, UK.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Background and Purpose The ability to detect asymptomatic circulating cerebral emboli may contribute to the management of patients with stroke, but its clinical usefulness will depend on effective systems for automatically detecting embolic signals (ES) and differentiating them from artifact. A new method involves the use of a multidepth probe that allows recording from both distal and proximal sample volumes along the same vessel. Theoretically, an embolus should appear sequentially, with a time delay, between the two channels, whereas an artifact should appear simultaneously in the two channels.
Methods We evaluated this method in an in vitro flow model and in patients. In an in vitro model, with a flow pattern mimicking intracerebral flow, 181 air bubbles and 193 thrombus emboli were compared with the signals resulting from 368 episodes of artifact; a sample volume of 5 mm and a channel separation of 10 mm were used. ES from two groups of patients—those with carotid artery stenosis (141 ES) and those with mechanical prosthetic cardiac valves (125 ES)—were studied and compared with 222 episodes of artifact produced in the same patients.
Results In the model the mean (SD) time delay was 17.32 (9.94) ms for air emboli and 17.78 (10.66) ms for thrombus emboli compared with -0.01 (0.39) ms for artifact (air and thrombus emboli versus artifact, P<.0001). A sensitivity of 100% and specificity of 100% were obtained when a cutoff of >2 ms was used for an embolus. The method allowed equally good detection of those air emboli that resulted in receiver overload and aliasing. In patients the mean (SD) time delay was 29.6 (28.2) ms for valve ES and 14.9 (15.42) for carotid ES compared with 0.00 (0.46) for artifact (carotid and valve ES versus artifact, P<.0001). Considering only those signals that were visible in both Doppler time domains resulted in a sensitivity for valve ES of 98.9% and for carotid ES of 94.0%, with a specificity of 99.0%. However, in one patient in the valve group some ES were visible only in the proximal channel, possibly because of passage of emboli down branch vessels between the two sample volumes. In addition, for the less intense carotid ES some signals were unclear or absent in one or both of the time domain signals at either depth, although visible in the post–fast Fourier transform spectra. Including those ES visible in only one channel reduced the sensitivity to 75.2% for valve ES and 92.6% for carotid ES.
Conclusions The multigated technique offers a new method of detecting ES and differentiating them from artifact and is the first reliable method for differentiating intense ES resulting in receiver overload from artifact. Occasionally its sensitivity is reduced when ES do not appear in the distal channel, probably because they pass down a side branch; this may be reduced by reducing gate separation. Some less intense carotid ES can be difficult to detect if the amplitude increase is small compared with the amplitude of the background Doppler signal.
Key Words: carotid artery diseases • cerebral embolism • heart valve prosthesis • ultrasonics
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Stroke is the third leading cause of death and a major cause of disability. Cerebral embolism is the underlying pathogenic mechanism in many cases of stroke. Emboli may arise from the heart, carotid plaques, aortic plaques, intracranial atherosclerotic stenoses, or from systemic venous thrombosis in the presence of a venous to arterial shunt. The ability to detect asymptomatic circulating cerebral emboli offers important potential advances in the localization of an actively embolizing source, the selection of high-risk patients for appropriate treatment, monitoring of the effectiveness of anticoagulant and antiplatelet therapy, and perioperative monitoring. Embolic signals (ES) appear as a unidirectional frequency-focused intensity increase, usually within the background spectral pattern and occurring at random within the cardiac cycle, accompanied by a characteristic harmonic sound. The technique has been demonstrated to be highly sensitive and specific in both in vitro1 and animal2 3 models. ES have been detected in patients with a variety of embolic sources, including carotid stenosis, atrial fibrillation, and prosthetic cardiac valves,4 5 6 and during and after operative procedures such as carotid endarterectomy7 and carotid angioplasty.8 One of the major factors hindering the wider clinical application of the technique is the time taken to analyze off-line recordings from patients. Although the optimal recording time has not yet been defined for each separate patient subgroup, recordings are normally made for a minimum of 30 minutes, with much longer periods being used by some groups. If embolic detection with transcranial Doppler ultrasonography is to become clinically useful, a reliable automated detection system is required. Such a technique should be both highly specific and sensitive for ES and be able to differentiate these from both patient and probe artifact. Previous investigators have used a computer algorithm to identify the characteristic bell-shaped relative intensity increase occurring with an ES and to differentiate this from the characteristic bidirectional intensity increase seen with an artifact. High sensitivity and specificity were achieved in an off-line system,9 but results with on-line systems have not yet been as good.10 An alternative approach is to train a neural network; such a system has been shown to identify ES on-line with a high specificity, but its sensitivity is only in the order of 70%.11 Unfortunately, artifact may also occasionally produce embolus-like signals and therefore cannot always be differentiated on the basis of intensity, duration, and directionality of the signal alone. Furthermore, in cases in which ES cause receiver overload, aliasing occurs and bidirectional signals may be produced; these cannot be differentiated from artifact on the basis of previous embolus-defining criteria.
A very promising method that involves the use of multigated Doppler ultrasound has recently been described by Aaslid. Since an embolus is in motion in the direction of blood flow within the vessel being studied, if recordings are made at two depths along the vessel length, there should be a time delay between the ES seen at the distal depth and that seen more proximally. In contrast, signals produced by external interference would be expected to be seen simultaneously in both channels. Results with such a system have been recently reported12 ; however, this study used the post–fast Fourier transform (FFT) spectra from which to calculate the time delay, and even better results would be expected when the time domain data with their much higher temporal resolution are used. Furthermore, the in vitro validation in this study12 used air emboli, and there has been no validation when solid or formed emboli, such as thrombus, are used, which result in less intense signals and have been more difficult to detect in previous automated systems.
In this study we initially evaluated a multigated Doppler ultrasound system in an in vitro flow model using both thrombus and air emboli, and we then applied it to a group of patients with potential embolic sources.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
In all studies the same transcranial pulsed Doppler machine was used (Pioneer 4040, EME Ltd) with a multidepth 2-MHz transducer. The Doppler signal was saved with software that allows the pre-FFT time domain Doppler signal, the post-FFT spectra, and the audio signal to be stored on the computer hard disk and replayed. The initial increase in amplitude at the time of arrival of an embolus or artifact in each of the two channels was measured from the time domain data, allowing a time resolution of 1 ms or greater.
In Vitro Studies
We constructed a flow circuit using polyethylene infusion tubing (4 mm ID, Codan Limited), driven by a programmable flow pump (University Developmental Cooperation Flow System) and filled with a proprietary blood analogue, a suspension of nylon filaments in machine oil that has been previously validated to have scatter properties similar to those of blood (Elf Atochem).13 A waveform similar to that normally obtained from the middle cerebral artery (MCA) with relative preservation of diastolic flow was obtained; mean systolic flow velocity was 40 cm/s (peak, 95 cm/s; diastolic, 30 cm/s). Air was removed from the circuit through a built-in reservoir tank. A length of the tubing was fixed in a water bath and insonated at an incident angle of 35°, such that there was a comparable waveform with equal gain settings at each depth. Channel 1 was set at a depth of 52 mm with the following parameters: sample size, 5 mm; power, 25% (100% power is equivalent to 675 mW/cm2); and gain, 8. Channel 2 was set at a depth of 42 mm at the same settings of sample size, power, and gain.
We introduced emboli of two types through a transparent sidearm device, allowing visualization of any accidentally introduced air: 181 air bubbles were introduced with a 1-mL syringe to produce bubbles of 0.01 mL in volume (diameter, 0.64 mm); 193 thrombus emboli were prepared from fresh human blood that had been allowed to clot, cut in cuboid shapes with a maximum dimension of 0.5, and then suspended in normal saline. In addition, 368 episodes of artifact were produced; 187 of these were produced by tapping the probe and 181 by tapping the tubing to mimic patient artifact.
Patient Studies
The aim of the patient studies was twofold: (1) to determine in what proportion of patients we could successfully insonate the MCA with satisfactory gate separation and (2) to determine the sensitivity and specificity of the method in detecting ES and differentiating them from artifact. ES in patients with mechanical prosthetic cardiac valves are more intense than those in patients with carotid stenosis14 and have been easier to detect by previous automated systems; therefore, we evaluated the method in both patient groups separately. In the carotid group we included ES recorded during the recovery phases after carotid endarterectomy and carotid angioplasty; these ES have an intensity similar to those recorded in symptomatic carotid stenosis.8
Insonation of 35 MCAs in 24 patients was attempted (16 men, 8 women; mean [SD, range] age, 66.5 years [10.39, 50 to 86]). Of these patients, 14 had a known potential source of emboli. Six patients with metallic prosthetic cardiac valves were monitored bilaterally for 20 minutes. Eight patients with symptomatic carotid stenosis (mean stenosis, 84.4%; range, 60% to 95%) were monitored ipsilateral to the stenosis for 1 hour. Four patients later underwent carotid endarterectomy, and 1 patient underwent carotid angioplasty. In these patients more extensive postoperative recordings were made, for a total of 300 minutes per patient. All these recordings were made following the procedure, either after catheter removal for percutaneous transluminal angioplasty or in the recovery room for endarterectomy, and no recordings were made during the procedures when ES could represent air bubbles. In the additional 10 subjects recruited from hospital inpatients, monitoring was continued only for the time it took to obtain a satisfactory signal; this was to determine whether adequate gate separation could be obtained.
Insonation was achieved by the transtemporal route, with the use of a 2-MHz probe held in place with an external fixation device. Once the MCA was identified, the depth in the two channels was adjusted, and the axial sample volume width was reduced while satisfactory visualization of the Doppler spectra within the two samples was maintained. A standard protocol was used in siting the position of the two sample volumes; we aimed for a sample volume of 5 mm and a distance between the center points of the sample volumes (ie, between the two depths) of 10 mm. Episodes of artifact were also recorded for off-line analysis as above. These were produced by requesting that the patient cough, speak, or swallow and by tapping or moving the probe, probe holder, or headgear.
Signal Analysis
For all ES, maximum relative intensity increase was determined. The intensity was taken from the intensity color-coded spectral display. The background intensity was calculated from a similar point in the preceding or following cardiac cycle. The pre-FFT time domain data from the two channels were used to determine whether the intensity increase was present in both channels; if so, the time delay in the onset of the intensity increase in the two channels was measured. Comparisons between groups were made with the use of Student's t test for unpaired data.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
In Vitro Studies
All 374 emboli were detected as high-intensity signals. Air emboli resulted in more intense signals than thrombus emboli (mean [SD, range] for air emboli, 31.5 dB [11.8, 11 to 39] versus 22.5 dB [4.8, 11 to 39] for thrombus emboli; P<.0001). Forty-three air emboli but no thrombus emboli resulted in receiver overload and a bidirectional signal, as previously reported; these were included in the analysis. All emboli were detected in both channels (Fig 1A and 1B
), in all cases first in the more proximal channel, with a mean (SD, range) time delay between the two channels of 17.56 ms (10.31, 3 to 53). The mean (SD, range) delay was 17.32 ms (9.94, 3 to 53) for air emboli and 17.78 ms (10.66, 4 to 50) for thrombus emboli. In contrast, artifact appeared simultaneously or near simultaneously in the two channels (Fig 1C), with a mean (SD, range) time delay of -0.01 ms (0.39, -4 to 2; P<.0001 versus emboli, t test).
|
Time delays for air and thrombus emboli and artifact are shown in Fig 2. When we specified a cutoff time delay of more than 2 ms to define a signal as an embolus, the method could detect emboli and differentiate them from artifact with a sensitivity of 100% and a specificity of 100%; this was similarly good when either air emboli alone or thrombus emboli alone were considered. These values include the 43 air emboli that resulted in receiver overload and aliasing; for these cases analyzed as a separate group, mean (SD, range) time delay between gates was 19.81 ms (9.61, 9 to 53).
|
There was a highly significant relationship between the velocity at which the ES intensity increase occurred and the time delay between the two channels: air emboli, r=-.77, P<.0001; thrombus, r=-.74, P<.0001.
Patient Studies
Successful insonation of the MCA was possible in 33 of 35 arteries (94.3%). In the two failed cases this was due to unilateral absence of an acoustic temporal window. In all MCAs that could be insonated, it was possible to record at two depths, with a mean (SD, range) sample volume of 4.96 mm (0.20, 4 to 5) and mean (SD, range) gate separation of 4.71 mm (1.85, 1 to 9).
Prosthetic Mechanical Cardiac Valves
We recorded 125 ES, with a mean (SD, range) relative intensity increase of 31.1 dB (7.7, 10 to 55). The mean (SD, range) time delay was 29.6 ms (28.2, 2 to 122). For 30 ES, all in the same patient, the ES were heard and detected in both the time domain data and the post-FFT spectral display in the proximal channel but were not audible or visible in the distal channel, suggesting that some emboli may have passed into a branching artery between the two sample volumes. All other ES were visible and audible in both channels. The mean time delay for ES was significantly longer than that for the 222 episodes of artifact, which had a mean (SD, range) time delay of 0.0 ms (0.46, -2 to 3; P<.0001). Time delays for the ES and artifact are shown in Fig 3. When we included only the ES detected in both channels and used a threshold of a time delay of 2 ms between the two depths as the defining criterion for ES, the sensitivity was 98.9% and specificity 99.0%; however, if the ES visible in only one channel were also included, the sensitivity fell to 75.2%. On the whole, the ES in patients with cardiac valves were clearly visible as a large amplitude increase in the time domain data, in contrast to some of the carotid ES (Fig 1D
).
|
Carotid Artery Stenosis
We recorded 141 ES, with a mean (SD, range) relative intensity increase of 19.3 dB (5.9, 5 to 33). No ES were visible and audible in the proximal channel but not in the distal channel, unlike the valve ES. However, two ES audible and visible in the post-FFT spectra in both channels were identifiable in the time domain data of only one channel; one ES was detected only in the proximal but not the distal time domain data, while one ES was clearly audible and visible in the proximal channel post-FFT spectral display but not visible in the time domain data at that depth. This appeared to be a reflection of the fact that the amplitude increase in the time domain data was small for many ES and frequently difficult to distinguish from the background Doppler signal; a typical example is shown in Fig 1E. The ratio of the maximum amplitude increase to the background Doppler signal amplitude in the time domain data was small for many carotid ES, and the mean value was significantly smaller than that for valve ES (5.1 [2.5] versus 13.2 [5.7]; P<.0001); individual values are shown in Fig 4. For ES identifiable in both channels, the mean (SD, range) time delay between the two channels was 14.9 ms (15.42, 0 to 90). The time delay was significantly longer than that of the 222 episodes of artifact (P<.0001). When the same 2-ms cutoff was used, a sensitivity of 94.0% and specificity of 99.0% were obtained when those signals identifiable in both channels were considered (Fig 3). The sensitivity fell to 92.6% when all ES, including those visible in only one channel, were considered.
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
In our in vitro model, the use of a multigated technique allowed the identification of ES and their discrimination from artifact with a very high sensitivity and specificity. Of particular note, it allowed detection of ES that resulted in receiver overload with a similar high sensitivity and specificity; this is the first method to be able to identify such ES, which cannot be identified from the frequency spectral data alone either visually or with the use of a neural network11 or computer algorithm.10 This application may be particularly useful when it is used during operative procedures, when both echogenic air emboli and artifact may be common. In addition, the multigate data may allow the unambiguous identification of low-velocity ES, which can sometimes be difficult to separate from artifact on frequency spectral data alone.
In all patients in whom an acoustic window could be obtained, it was technically possible to record using the multidepth probe with adequate gate separation. We identified the sensitivity and specificity of the multigated system using ES identified according to conventional criteria and compared them with known episodes of artifact. The results in patients were encouraging but not as impressive as the in vitro data for a number of reasons. This was primarily because in some patients the ES did not appear in both channels. In one patient in the valve group, 30 ES (approximately 40% of the signals in this patient) were detected proximally but were not audible or visible distally in either the time domain data or in the post-FFT spectral data. This is likely because between the two sample volumes a proportion of the emboli pass down a branch vessel. Our protocol aimed for a gate separation, between the centers of each sample volume, of 10 mm, and sometimes the distal gate was a fairly shallow depth of 42 mm; the problem may be reduced with the use of a deeper distal gate, thereby lowering the chance that this gate may be sited beyond the branching of the MCA. Further studies are required to examine whether reduced gate separation allows such good separation between emboli and artifact. On the other hand, gate separation that is too narrow will reduce the time interval between detection of emboli in the proximal and distal gates and will reduce the specificity in differentiating emboli from artifact. Apart from this difficulty in a single patient, the use of a multigated probe allowed detection of emboli and differentiation from artifact with a very high sensitivity, similar to that seen in the in vitro model. This is largely a reflection of the higher intensity of the more echogenic mechanical valve ES, as previously reported. This resulted in a large and clear amplitude increase in the time domain signal (Fig 1).
In contrast, for the less intense carotid ES, some signals were identifiable audibly and in the post-FFT spectral display but not in the time domain data. This occurred for low-intensity emboli, in which the power or intensity increase was small and lost in the background Doppler signal. An additional possible explanation is that the total cross-sectional area of the MCA is not fully covered by the sample volume at both depths. This will be less of a problem for more echogenic emboli, which have a larger effective sample volume.15 Separating the intensity increase at different frequencies by an FFT analysis allowed identification of the ES since the intensity increase of an ES is maximal at one frequency. Although only two carotid ES were not detectable in the time domain data, a much larger number resulted in an amplitude increase only slightly greater than that of the background Doppler signal. This is illustrated by the relatively low ratio of the maximum amplitude to background amplitude for many carotid ES in the time domain data (Figs 1E and 4). With the use of an off-line analysis in combination with the post-FFT spectral data, it is possible to determine the time delay for most carotid ES. However, it may prove difficult in automated on-line systems based on the time domain data alone to distinguish the small amplitude increase occurring with some carotid ES from amplitude fluctuations in the normal background Doppler signal. Further work is required to improve the sensitivity of the multigated technique for small-amplitude ES.
The predominant factor determining the time taken to travel between the two channels was the velocity at which the ES was maximal, which presumably reflects the speed at which the embolus is traveling. However, in the model there was a much greater range of time taken between detection in each channels than would be expected by simple mathematical principles. The theoretical distance traveled by an embolus as detected by the probe is 5 mm (ie, distance between the edges of the axial sample volumes). If we correct for angle of insonation, an embolus traveling at the mean velocity (40 cm/s) would be expected to take 15.3 ms to cross between the two axial sample volumes. Any embolus traveling at the peak velocity for the system (95 cm/s) would be expected to give a time delay of 6.4 ms between the two depths. It is noteworthy that 46 of the emboli we produced traversed the distance in 6 ms or less. This reflects the fact that rather than being cylindrical in shape, with a sharp cutoff of the ultrasound beam at each end of the sample, there is a gradual weakening of the beam at each end. It follows that the effective sample volume will be greater for more echogenic emboli such as air emboli.15 In patient studies some ES took much longer to travel between the two sample volumes, and this was particularly so for some valve ES. This may be due to their passage being slowed by turbulence, nonlaminar flow, and momentary adhesion to the vessel wall.
Our results demonstrate that a multigated approach can detect ES and differentiate them from artifact, but as with other methods, it has some inherent problems that need to be resolved before it is suitable for routine clinical use. Our results are similar to those reported previously12 but show a lower sensitivity for low-amplitude carotid ES. In this previous study, carotid ES, a minority of the ES, were not separated from those from patients with heart valve replacements or left ventricular assist devices, and no analysis of the intensity of the ES was made. Our results demonstrate that in patients with the more intense mechanical valve ES, high sensitivity and specificity are likely to be obtained; the only major difficulty appeared to be passage of emboli down a branching vessel between the two sample volumes, and this may be improved by reduced gate separation. In patients with carotid ES, the less intense ES may be unclear in one or both channels. For these ES, combining the multigated method with a method analyzing the post-FFT spectral data may improve sensitivity; use of the frequency spectra may provide greater resolution because the intensity increase associated with an ES is usually centered on a narrow frequency band.
|
Acknowledgments |
|---|
This study was supported by a British Heart Foundation Project grant. We thank Colin Deane for help with the flow model and Andy Healey for technical advice.
Received March 11, 1996; revision received May 13, 1996; accepted May 13, 1996.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
1. Markus HS, Brown MM. Differentiation between different pathological cerebral embolic materials using transcranial Doppler in an in vitro model. Stroke. 1993;24:1-5.
2.
Russell D, Madden KP, Clark WM, Sandset PM, Zivin JA. Detection of arterial emboli using Doppler ultrasound in rabbits. Stroke. 1991;22:253-258.
3. Markus HS, Loh A, Brown MM. Detection of circulating cerebral emboli using Doppler ultrasound in a sheep model. J Neurol Sci. 1994;122:117-124.[Medline] [Order article via Infotrieve]
4. Rams JJ, Davis AD, Lolley MD, Berger PM, Spencer MP. Detection of microemboli in patients with artificial heart valves using transcranial Doppler: preliminary observations. J Heart Valve Dis. 1993;2:37-41.[Medline] [Order article via Infotrieve]
5.
Tong DC, Albers GW. Transcranial Doppler–detected microemboli in patients with acute stroke. Stroke. 1995;26:1588-1592.
6.
Siebler M, Sitzer M, Steinmetz H. Detection of intracranial emboli in patients with symptomatic extracranial carotid artery disease. Stroke. 1992;23:1652-1654.
7.
Van Zuilen EV, Moll FL, Vermeulen FE, Mauser HW, van Gijn J, Ackerstaff RG. Detection of cerebral microemboli by means of transcranial Doppler monitoring before and after carotid endarterectomy. Stroke. 1995;26:210-213.
8. Markus HS, Clifton A, Buckenham T, Brown MM. Carotid angioplasty: detection of embolic signals during and after the procedure. Stroke. 1994;25:2403-2406.[Abstract]
9.
Markus HS, Loh A, Brown MM. Computerized detection of cerebral emboli and discrimination from artifact using Doppler ultrasound. Stroke. 1993;24:1667-1672.
10. Mess WH, Van Zuilen EV, Ackerstaff RGA. Comparison of three automatic embolus detection systems with a human expert in patients with symptomatic carotid artery stenosis. J Neuroimaging. 1995;5(suppl 2):S67. Abstract.
11.
Siebler M, Sitzer M, Rose G, Benfeldt D, Steinmetz H. Silent cerebral embolism caused by neurologically symptomatic high-grade carotid stenosis: event rates before and after carotid endarterectomy. Brain. 1993;116:1005-1015.
12.
Georgiadis D, Gocke J, Hill M, Konig M, Nabavi DG, Stogbauer F, Zunker P, Ringelstein EB. A novel technique for identification of Doppler microembolic signals based on the coincidence method. Stroke. 1996;27:683-686.
13. Oates CP. Towards an ideal analogue for Doppler ultrasound phantoms. Phys Med Biol. 1991;36:1433-1442.[Medline] [Order article via Infotrieve]
14.
Grosset DG, Georgiadis D, Kelman AW, Lees KR. Quantification of ultrasound emboli signals in patients with cardiac and carotid disease. Stroke. 1993;24:1922-1924.
15.
Smith JL, Evans DH, Fan L, Gaunt ME, London NJM, Bell PRF, Naylor AR. Interpretation of embolic phenomena during carotid endarterectomy. Stroke. 1995;26:2281-2284.
This article has been cited by other articles:
 |
|
 |
|
  Y. Kalairajah, A. J. Cossey, G. M. Verrall, G. Ludbrook, and A. J. Spriggins Are systemic emboli reduced in computer-assisted knee surgery?: A PROSPECTIVE, RANDOMISED, CLINICAL TRIAL J Bone Joint Surg Br, February 1, 2006; 88-B(2): 198 - 202. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 I. K. Moppett and R. P. Mahajan Transcranial Doppler ultrasonography in anaesthesia and intensive care Br. J. Anaesth., November 1, 2004; 93(5): 710 - 724. [Full Text] [PDF]
|
|
|
|
 |
|
 G. Devuyst, G.A. Darbellay, J.-M. Vesin, V. Kemeny, M. Ritter, D.W. Droste, C. Molina, J. Serena, R. Sztajzel, P. Ruchat, et al. Automatic Classification of HITS Into Artifacts or Solid or Gaseous Emboli by a Wavelet Representation Combined With Dual-Gate TCD Stroke, December 1, 2001; 32(12): 2803 - 2809. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Cullinane, G. Reid, R. Dittrich, Z. Kaposzta, R. Ackerstaff, V. Babikian, D. W. Droste, D. Grossett, M. Siebler, L. Valton, et al. Evaluation of New Online Automated Embolic Signal Detection Algorithm, Including Comparison With Panel of International Experts Stroke, June 1, 2000; 31(6): 1335 - 1341. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Z. Kaposzta, E. Young, P. M. W. Bath, and H. S. Markus Clinical Application of Asymptomatic Embolic Signal Detection in Acute Stroke : A Prospective Study Stroke, September 1, 1999; 30(9): 1814 - 1818. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. L. Taylor, M. A. Borger, R. D. Weisel, L. Fedorko, and C. M. Feindel Cerebral microemboli during cardiopulmonary bypass: increased emboli during perfusionist interventions Ann. Thorac. Surg., July 1, 1999; 68(1): 89 - 93. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
V. Kemeny, D. W. Droste, S. Hermes, D. G. Nabavi, G. Schulte-Altedorneburg, M. Siebler, and E. B. Ringelstein Automatic Embolus Detection by a Neural Network Stroke, April 1, 1999; 30(4): 807 - 810. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. Bernd Ringelstein, D. W. Droste, V. L. Babikian, D. H. Evans, D. G. Grosset, M. Kaps, H. S. Markus, D. Russell, and M. Siebler Consensus on Microembolus Detection by TCD Stroke, March 1, 1998; 29(3): 725 - 729. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Georgiadis, A. Wenzel, H. R. Zerkowski, S. Zierz, and A. Lindner Automated Intraoperative Detection of Doppler Microembolic Signals Using the Bigate Approach Stroke, January 1, 1998; 29(1): 137 - 139. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. S. Markus and J. Molloy Use of a Decibel Threshold in Detecting Doppler Embolic Signals Stroke, April 1, 1997; 28(4): 692 - 695. [Abstract] [Full Text]
|
|
|
|
|
|
D. W. Droste, G. Hagedorn, A. Notzold, H.-J. Siemens, H. H. Sievers, and M. Kaps Bigated Transcranial Doppler for the Detection of Clinically Silent Circulating Emboli in Normal Persons and Patients With Prosthetic Cardiac Valves Stroke, March 1, 1997; 28(3): 588 - 592. [Abstract] [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||