No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Phase Segmentation of Noisy Respiratory Sound Signals using Genetic Approach.
Abstract
In this paper, a new approach to automatically segment noisy respiratory sound signals is proposed. Segmentation is formulated as an optimization problem and the boundaries of the signal segments are detected using a genetic algorithm (GA). As the estimated number of segments present in a segmenting signal is initially obtained, a multi-population GA is employed to determine the locations of segment boundaries. The segmentation results are found through the generations of GA by introducing a new evaluation function, which is based on the sample entropy and a heterogeneity measure. Illustrative results for respiratory sound signals contaminated by loud heartbeats and other high level noises show that the proposed genetic segmentation method is quite accurate and threshold independent to find the noisy respiratory segments as well as the pause segments under different noisy conditions.
... The proposed method therefore should be independent of amplitude variation between these two respiratory phases and able to perform accurate annotation without knowing the structures of the respiratory cycles . The presented method is applied on the consecutive inspiration/expiration segments as obtained by using the segmentation method presented in [10] . Each pair of consecutive respiratory phase segments are first aligned using phase shift information. ...
... The sub-schemes together with the notations used in the block diagram are described in Section 2.2 and 2.3 respectively. Here, phase segmentation method which has been proposed in our previous paper [10] is used for the correct localization of the respiratory phase segments. ...
... As consecutive inspiration/expiration segments pairs are used as the input of the presented method, the effectiveness of the proposed annotating index depends on the correct boundary locations of segmented respiratory phases (inspiratory/expiratory phases). Although segmentation method with high accuracy in [10] is adopted here to obtain the inspiration/expiration segments pairs, any other suitable methods can also be used. Therefore, respiratory phase segment alignment scheme is introduced as signal conditioning in order to make the annotation index independent of respiratory phase segmentation accuracy. ...
This paper introduces a novel method to identify inspiratory and expiratory phases from single channel tracheal breath sound (TBS) of different types, by proposing a new anno-tating index name as "mixing index" (MI). An alignment scheme based on phase shift difference information has been firstly introduced to align the consecutive respiratory phase segments. MI is then proposed based on similarity mea-surements to annotate the respective inspiration/expiration in each aligned respiratory phase segment pair. By incorporat-ing the novel alignment scheme, the presented index over-comes the problem of phase cancellation which affects the cross-coherence of the input segment pairs. As MI is invari-ant to spectral content and amplitude dynamics, the proposed method maintains a good performance even in the presence of adventitious sounds. A high averaged accuracy of 97.4% for adventitious sounds and 100% for normal TBS have been thereby achieved. The proposed method has been a success-ful attempt to solve the clinical application challenge faced by the existing phase identification methods in terms of res-piratory dysfunctions.
... A relatively high estimation accuracy has been achieved in [7] but the predefined linear model applied does not support flow estimation for different types of TS other than normal TS. In this paper, an enhanced RR monitoring method for real TS recordings has been proposed by extending a recently developed GA based respiratory phase segmentation method [8]. The relationships between signal redundancy, disjointness and wide-sense stationarity have been investigated . ...
This paper addresses the problem of non-invasive respiratory rate (RR) monitoring using single channel tracheal sound (TS) recordings. We have recently developed a robust res-piratory phase segmentation method based on genetic algo-rithm (GA) which works well only for preprocessed clean TS. Therefore, an enhanced respiratory phase monitoring method is proposed in this paper by exploiting the signal re-dundancy to our existing method. In this context, appropriate overlapping windows have been applied to ensure sufficient redundancy of TS signals. The performance of the enhanced method is analyzed for different types of real TS and stan-dard preprocessed TS. The average accuracy of respiratory phase segmentation found for real TS is comparable to that of the standard preprocessed data by our proposed method.