Synchronizing Respiratory Sounds and EIT

WELMO aims to develop and validate a new generation of low-cost and low-power miniaturized sensors, integrated in a comfortable vest, enabling the effective and accurate monitoring of the respiratory function, through the simultaneous collection of sound and electrical impedance tomography (EIT) signals. Higher inference capacity to diagnose and monitor respiratory diseases can be achieved with the simultaneous acquisition of multiple bio-signals.

To develop and train the automatic detection and monitoring algorithms that will be deployed in the WELMO vest, large amounts of data are required (containing both respiratory sound and EIT data). To date, there is no integrated system available that allows to simultaneously record respiratory sound and EIT. Thus, during data collection two separate equipments will be used to record the bio-signals (AKG C417L Microphone and 3M Litmmann 3200 Electronic Stethoscope for respiratory sound recording, and Goe-MF II EIT device for EIT signal recording). However, with this approach the collected data is not synchronous, thereby, both signals need to be aligned in post-acquisition, since data synchronism is mandatory to effectively combine the two bio-signals.

With the purpose of synchronizing respiratory sound and EIT data, we have developed an architecture based on the use of an auxiliary signal. The system works by generating an auxiliary sound signal (a pure sinusoidal tone at a specific frequency), which is then split to a loudspeaker and to the EIT device, using an audio splitter and a 3.5mm to BNC adapter (see Figure 1). This division will allow the auxiliary signal to be simultaneously detected in both systems responsible for the respiratory sound and EIT recording, accordingly.

Figure 1: Synchronization architecture scheme

The synchronization process is presented in Figure 2. The proposed solution works by using an initial interval (10 seconds in the example presented in Figure 2) of the acquisition to record the auxiliary signal, both in the lung sound and EIT recordings. Then, the difference between the end of the auxiliary signal and the end of the initial synchronization interval is determined (t1 in Figure 2). Using t1, itis then possible to align the sound recording with the EIT signal, as seen in Figure 2. The result of the alignment process is represented in Figure 3.

Figure 2: Lung sounds and EIT synchronization example
Figure 3: Synchronized interval

Using this solution to obtain synchronous data, the WELMO consortium will have a valuable and large database that will serve as an indispensable tool for researchers to develop their algorithms. Ultimately, it provides an opportunity to develop groundbreaking algorithms to monitor and assess respiratory function and contributes to the success of the project.

Rui Pedro Paiva

University of Coimbra