Lungengeräuschanalyse bei herzinsuffizienten Patienten - Eine Pilotstudie zur Detektion von Rasselgeräuschen mittels elektronischer Auskultation

Einleitung: Die klassische Auskultation des Thorax ist eine wichtige, nicht-invasive und leicht durchführbare Methode, die unmittelbare und manchmal lebensrettende Informationen über Struktur und Funktion von Herz und Lunge liefert. Moderne Methoden der Biosignalanalyse machen es heute darüber hinau...

Full description

Saved in:
Bibliographic Details
Main Author: Decker, Philip
Contributors: Koehler, Ulrich (Prof. Dr.) (Thesis advisor)
Format: Dissertation
Published: Philipps-Universität Marburg 2013
Innere Medizin
Online Access:PDF Full Text
Tags: Add Tag
No Tags, Be the first to tag this record!
Table of Contents: Introduction: Classic chest auscultation is an important, non-invasive and easy to use bedside technique. It provides immediate and sometimes even life-saving information about the structure and function of chest organs including the heart and lungs. Compared to a normal auscultation, current methods of biosignal analysis offer much more information regarding the detected sounds. Recently, recommended standards regarding terms and techniques for computerized respiratory sound analysis have been published. Since then, different working groups have developed procedures to analyze lung sounds which unite varying methods of modern signal analysis and allow a comprehensive evaluation of different breath and adventitious sounds. Although different methodological approaches have been published, there is still no system commercially available on the medical engineering market for the automatic detection of crackles. Methodology: In this pilot study, a computer-assisted method of electronic auscultation for the detection of crackles is presented. The recording system used in this work is a prototype that has been developed in Marburg. It has been especially designed for the analysis of lung sounds and meets the CORSA standard. In previous studies, the system has been mainly used for the detection of wheezing. Results: Compared to classic auscultation, the electronic approach revealed a sensitivity of 100 % (95 %-CI 85 % - 100 %) and a specificity of 88 % (95 %-CI 47 % - 100 %) regarding the ability to detect crackles. Additionally, reference data of 30 patients with congestive heart failure and 20 healthy subjects was generated. This data will be used to create a functional algorithm for the automatic detection of crackles through interdisciplinary collaborative work with biomedical engineers at the University of Applied Sciences in Gießen. Discussion: In the long-term, a diagnostic clinical trial should be carried out on a larger cohort to evaluate and validate the algorithm. In parallel, the recording system should undergo further enhancement. In this study, recommendations regarding capturing, preprocessing and digitization, as well as the further processing of the captured signals, are presented. They offer some incentives for further technical developments. There is a demand for suitable sensors that allow long-term recordings in a supine position. The present trial only dealt with the general detection of crackles. From a diagnostic point of view, the temporal resolution of single crackle-events as well as their exact quantification and a differentiated analysis of their spectral characteristics would be interesting. Further investigations need to be undertaken to clarify which procedures are suitable here. If further studies confirm, that lung sounds allow the diagnosis of congestive heart failure, it will be possible to use respiratory sound analysis as an additional, non-invasive diagnostic instrument. In addition to the areas of application described, the technique could be used for the monitoring of critically ill patients and the etiological clarification of paroxysmal nocturnal dyspnoea. Compared to classic auscultation techniques, computer-assisted, electronic respiratory sound analysis offers objective, reproducible, qualitative and quantitative analysis of sound signals.