H Markus1, A Loh, M M Brown. 1. Division of Clinical Neuroscience, St George's Hospital Medical School, London, UK.
Abstract
BACKGROUND AND PURPOSE: Transcranial Doppler ultrasound can detect circulating cerebral emboli. Monitoring of patients with potential embolic sources may allow identification of high-risk patients who can then be selected for prophylactic treatment. However, practical patient monitoring will require automated programs that can detect emboli and differentiate them from artifact. METHODS: A new off-line algorithm for the detection of emboli, which detects the characteristic relative power increase occurring with an embolus, was evaluated in both an animal model and in patients. (1) In a sheep model, solid embolic materials (thrombus, platelet aggregates, and atheroma) were introduced into the proximal carotid artery while the distal carotid artery or a major branch was insonated. The signals resulting from 77 emboli (mean size, 1.77 mm) were studied and compared with the Doppler signals resulting from artifact. (2) In patients, 100 embolic signals occurring in three patients were analyzed and compared with signals associated with artifact in the same patients. RESULTS: (1) In the sheep model, emboli resulted in a short-duration, high-intensity signal, but intensity increase alone did not distinguish between emboli and artifact. In contrast, the algorithm discriminated embolus from artifact with a sensitivity of 98.7% and a specificity of 98.0%. (2) In patient studies, embolic signals were differentiated from artifact with a sensitivity of 97.2% and a specificity of 97.0% by the algorithm. CONCLUSIONS: Using such an algorithm, detection of cerebral emboli and discrimination from artifact are possible with a high sensitivity and specificity. Incorporation of such an algorithm into an on-line system should make prolonged patient monitoring practical.
BACKGROUND AND PURPOSE: Transcranial Doppler ultrasound can detect circulating cerebral emboli. Monitoring of patients with potential embolic sources may allow identification of high-risk patients who can then be selected for prophylactic treatment. However, practical patient monitoring will require automated programs that can detect emboli and differentiate them from artifact. METHODS: A new off-line algorithm for the detection of emboli, which detects the characteristic relative power increase occurring with an embolus, was evaluated in both an animal model and in patients. (1) In a sheep model, solid embolic materials (thrombus, platelet aggregates, and atheroma) were introduced into the proximal carotid artery while the distal carotid artery or a major branch was insonated. The signals resulting from 77 emboli (mean size, 1.77 mm) were studied and compared with the Doppler signals resulting from artifact. (2) In patients, 100 embolic signals occurring in three patients were analyzed and compared with signals associated with artifact in the same patients. RESULTS: (1) In the sheep model, emboli resulted in a short-duration, high-intensity signal, but intensity increase alone did not distinguish between emboli and artifact. In contrast, the algorithm discriminated embolus from artifact with a sensitivity of 98.7% and a specificity of 98.0%. (2) In patient studies, embolic signals were differentiated from artifact with a sensitivity of 97.2% and a specificity of 97.0% by the algorithm. CONCLUSIONS: Using such an algorithm, detection of cerebral emboli and discrimination from artifact are possible with a high sensitivity and specificity. Incorporation of such an algorithm into an on-line system should make prolonged patient monitoring practical.
Authors: Caroline Banahan; Zach Rogerson; Clément Rousseau; Kumar V Ramnarine; David H Evans; Emma M L Chung Journal: Ultrasound Med Biol Date: 2014-09-11 Impact factor: 2.998