The narrow band hypothesis: an interesting approach for high-intensity transient signals (HITS) detection

We propose a new approach to detect microemboli automatically using the narrow band hypothesis. An initial database of 560 peripheral arterial Doppler high-intensity transient signals (HITS) was created to study microemboli and to define the normal limits to be used in our method. When a HITS occurs, our approach consists of modelling the Doppler signal using amplitude and frequency wave modulation. A threshold was defined experimentally using this database and then applied to 38 recordings from 12 patients. Using another database, six expert Doppler users reported 140, 176, 155, 161, 161 and 146 HITS, corresponding to a total of 197 different observed HITS. When an event was detected by 6, 5, 4, 3, 2 and 1 of the observers, the sensitivity of the automatic detection was 94.8%, 75.9%, 55.6%, 42.9%, 30% and 0%, respectively. The sensitivity of our automatic detection thus is highly associated with the likelihood (defined as the ratio of observers in agreement to the total number of observers) of an event: r = 0.99 for p < 0.0001. Although future research would result in improvement of the specificity, the narrow band hypothesis appears to be a promising technique for the detection of HITS.