In impedance cardiography (ICG), the detection of dZ/dt signal (ICG) characteristic points, especially the X point, is a crucial step for the calculation of hemodynamic parameters such as stroke volume (SV) and cardiac output (CO). Unfortunately, for beat-to-beat calculations, the accuracy of the detection is affected by the variability of the ICG complex subtypes. Thus, in this work, automated classification of ICG complexes is proposed to support the detection of ICG characteristic points and the extraction of hemodynamic parameters according to several existing subtypes. A novel pattern recognition artificial neural network (PRANN) approach was implemented, and a divide-and-conquer strategy was used to identify the five different waveforms of the ICG complex waveform with output nodes no greater than 3. The PRANN was trained, tested and validated using a dataset from four volunteers from a measurement of eight electrodes. Once the training was satisfactory, the deployed network was validated on two other datasets that were completely different from the training dataset. As an additional performance validation of the PRANN, each dataset included four volunteers for a total of eight volunteers. The results show an average accuracy of 96% in classifying ICG complex subtypes with only a decrease in the accuracy to 83 and 80% on the validation datasets. This work indicates that the PRANN is a promising method for automated classification of ICG subtypes, facilitating the investigation of the extraction of hemodynamic parameters from beat-to-beat dZ/dt complexes.

Impedance cardiography (ICG) was discovered nearly half a century ago, being proposed as noninvasive monitoring method for estimation of several hemodynamics parameter. Despite of nearly 5 decades of clinical research and use there is still certain controversy about its performance when estimating Left Ventricular Ejection Time (LVET). This work present a comparison between using the different ICG subtype waveform and the ensemble averaged (EA) method to calculate the LVET. The ICG has been recorded from four volunteers, and the LVET parameter has been calculated using the two approaches. The result shows that each volunteer have different percentage of subtypes, and the mean relative error between the two approaches for estimation of LVET varied between 0.62 to 2.9% with an average mean absolute percentage error of 18,02% ranging between 13.82 to 18.42%. © 2021, Springer Nature Switzerland AG.

The detection of the characteristic points of the complex of the impedance cardiography (ICG) is a crucial step for the calculation of hemodynamical parameters such as left ventricular ejection time, stroke volume and cardiac output. Extracting the characteristic points from the dZ/dt ICG signal is usually affected by the variability of the ICG complex and assembling average is the method of choice to smooth out such variability. To avoid the use of assembling average that might filter out information relevant for the hemodynamic assessment requires extracting the characteristics points from the different subtypes of the ICG complex. Thus, as a first step to automatize the extraction parameters, the aim of this work is to detect automatically the kind of dZ/dt complex present in the ICG signal. To do so artificial neural networks have been designed with two different configurations for pattern matching (PRANN) and tested to identify the 6 different ICG complex subtypes. One of the configurations implements a 6-classes classifier and the other implemented the divide and conquer approach classifying in two stages. The data sets used in the training, validation and testing process of the PRANNs includes a matrix of 1 s windows of the ICG complexes from the 60 s long recordings of dZ/dt signal for each of the 4 healthy male volunteers. A total of 240 s. As a result, the divide and conquer approach improve the overall classification obtained with the one stage approach on +26% reaching and average classification ration of 82%.