Reliabilitätsuntersuchung zur standardisierten audio-visuellen Bewertung von Wheezing

Die computergestützte Analyse von Lungengeräuschen hat heute sowohl die weitere Erforschung ihrer Entstehungsmechanismen, als auch die Entwicklung einer automatischen Mustererkennung zur objektiven Beurteilung zum Ziel. Letzteres setzt jedoch die reproduzierbare und vergleichbare Beurteilbarkeit dur...

Ausführliche Beschreibung

Gespeichert in:
1. Verfasser: Benz, Julia Wilma Gunda
Beteiligte: Koehler, Ulrich (Prof. Dr.) (BetreuerIn (Doktorarbeit))
Format: Dissertation
Veröffentlicht: Philipps-Universität Marburg 2008
Innere Medizin
Online Zugang:PDF-Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
  • Computerized respiratory sound analysis is a non-invasive and objective method used to detect abnormal respiratory noises, with a great potential for monitoring of bronchial obstruction. For the development of advanced detection algorithms of pathological lung sounds homogeneous standards used for analysis and assessment are missing. Therefore, the goal of this study, using standardized analysis and conclusive methods, was to take a step forward in the development of universal standards for research and analysis. In this study 98 patients and healthy volunteers were recorded under standardized breathing (air-flow) using a prototype lung sound analyzer. From this collective, 24 patients with bronchial asthma (FEV1% MW+STD 83±21), 18 with COPD (FEV1% MW+STD 56±19) and 25 healthy volunteers (FEV1% MW+STD 106±17) were chosen. Men and women at different age (mean value: asthma 41±18; COPD 64±13; healthy volunteers 34±12) were included in the study. From the above collective 67 lung sound recordings were used for the “Testset”. All of the 42 recordings of the asthma and COPD patient-group had wheezing noise. The 67 lung sound recordings were presented in two rounds to five physicians from the department of Pneumology at the Marburg University Clinic, who possessed comparable clinical experience. They had to rate these recordings from a simultaneous audio and visual presentation on the computer using a 1-7 rating-scale for evaluating the severity. The rating contained the assessment of the continuous adventitious sound of “wheezing” using the aspects “wheezing rate” and “clinical relevance”. Prior to the ratings, the five physicians had been shown a “Lernset” (standardized training session), containing comparable data. In the case of interrater reliability, there could be found a significant correlation (p<0,01) between the five raters for both aspects (wheezing rate: r=0,868 to r=0,952; clinical relevance: r=0,882 to r=0,924). The assessment of the raters did not deviate by more than one scale point in 95,5% of the cases for the aspect wheezing rate and 89,7% for the clinical relevance (share of the total number=67). The validity, assessed by the re-test, also showed a significant correlation (p<0,01) for each rater and both aspects (wheezing rate: r=0,857 to r=0,976; clinical relevance: r=0,879 to r=0,962). In 96,7% of the cases for the wheezing rate and 94,9% for the clinical relevance (share of the total number=67) did the assessment of the raters not deviate by more than one scale point. This result shows a continuous and valid rating ability for each rater during repeated presentation. Evaluating the connection between the two aspects wheezing rate and clinical relevance, there could also be found a significant correlation (p<0,05) (r=0,8307 to r=0,9532). This displays a subjective connection between the two aspects, perceived by the raters. The 25 lung sound recordings of the healthy volunteers had been identified in total by each rater in both rounds, therefore the presence and absence of wheezing could be detected correctly. In conclusion, this study shows that clinical experienced physicians are able to rate the continuous adventitious sound of wheezing by using an audio-visual presented lung sound recording with high agreement and validity. The method of standardized assessment of lung sound recordings was proven to be reproducible. Using a 1-7 rating-scale can be therefore recommended for further studies as a standardized technique.