EGG DWPack: System for Multi-Channel Electrogastrographic Signals Recording and Analysis

The stomach is electrified and has its own pacemaker which generates electrical events (slow waves) with a frequency of about 3 cycles per minute (cpm) [14, 15, 31, 44, 48]. Gastric motility is regulated by gastric myoelectrical activity that consists of gastric slow waves and spike or second potentials [8, 14]. The gastric slow waves determine the propagation and maximum frequency of gastric contractions [7]. The cutaneous EGG includes only slow waves and disturbance potentials, and usually does not include spikes. This phenomenon is caused by the human abdominal wall, which behaves like a low-pass filter and attenuates higher harmonics of signals [41, 48]. Nowadays, abdominal surface electrogastrography is the one of methods of stomach examination that is used for diagnosing patients with chronic intractable nausea, vomiting and gastroparesis [6, 7, 12, 22, 26‐28, 36, 46, 48]. The typical recording process usually takes about 2 hours and consists of a 30-minute pre-prandial part (before the standardized meal), a meal part (when the examined subject eats a standardized meal), and two or three 30-minute postprandial parts. The standard meal depends on the examining center. The typical range of frequency for the EGG signal is from 0.5 to 9.0 cpm (0.008–0.15 Hz) and its amplitude is about 50–400 μV [5, 14, 29]. The EGG signal recording procedure is described in detail by Jieyun Yin et all in [48] and by Levanon et all [23]. The analysis of the EGG signal can not be performed by visually examining the electrogastrogram. This is because the EGG signal is the sum of all gastric slow waves presented in the stomach and in addition the recorded signal consists of not only the pure EGG but can include signals from other organs (e.g. respiration signals, colon signal, electrocardiographic signal (ECG), movement artifacts and other disturbances) [36, 48]. The EGG analysis is based on spectral analysis and is performed by computer applications. The spectral analysis of EGG signals is used to determine the frequency distribution of an EGG data fragment (segment). The typical EGG examination is divided into mentioned above parts (often called periods). Each period is usually divided into segments containing 60 to 240 seconds of data. The power spectrum density (PSD) is calculated for each segment. The overall power spectrum density (OPSD) is calculated as an average PSD of all segments for the whole examination and for each period. Based on the PSD and OPSD the dominant frequencies and dominant powers are calculated for the each segment and for the whole examination, and for the each period (overall dominant frequencies (ODF) and overall dominant powers (ODP)). The dominant frequency is defined as a value of frequency for the highest peak (dominant power) of the PSD in the range of 0.5–9.0 cpm [14, 29, 37, 48]. The clinically established parameters are: fasting-fed power ratio, percentage of normal gastric slow waves, percentage of gastric dysrhythmias, percentage of power distribution, and percentage of slow waves coupling [14, 29, 32, 37]. These parameters are calculated based on the DF, DP, ODF and the ODP values. A detailed description of these parameters is included in the next part of that article and can be found in the literature [14, 29, 37].

In this paper, the author introduces the EGG DWPack software (ver.2.0), which is an easy to use EGG analysis tool including a wide variety of analysis options. The tool is a MATLAB based software package.

This section introduces the method of processing the EGG signal and analyzed parameters which are commonly used in clinics and included in EGG DWPack software. The computation and descriptions of the parameters are mainly based on the information given in Polygram NET^TM reference manual [29], the article written by Parkman H.P. et all; “Electrogastrography: a Document Prepared by The Gastric Section of The American Motility Society Clinical GI Motility Testing Task Force” [37], and given in literature [5, 14, 48].

The part of the EGG DWPack system for analysis the EGG examination is a MATLAB based software. The application is used for multi-channel (maximum 4 channels) EGG data conditioned with the wide range of sampling frequency and various resolutions of analog to digital (A/D) conversion. The software has been developed using MATLAB Version 7.11.0.584 (R2010b) 64-bit (win64) and also has been tested using MATLAB (R2013b).

For validating EGG DWPack, the author used synthetic and real EGG signals. The synthetic 4-channel EGG data was constructed on the assumption that the slow wave can be represented by the sine with added noise (SNR (signal to noise ratio) ∈{0,3,6,10} dB). The characteristic frequency can be set in the range of 0.5 to 9.0 cpm (0.008–0.15 Hz), and the amplitude is about 50–400 μV. The test signals were 90 minutes in length, were sampled with frequency equal to 200 Hz and discretized with 12 bits A/D resolution. The parameters of used synthetic signals and basic results of OSA analysis can be seen in Table 5.

Except that four 4-channel synthetic signals were used to test the EGG DWPack, the validating process was performed in fourteen real human EGG data (twelve archival (Year 2009) and two new). The archival records were performed for four healthy, young subjects (all female, mean age 25.75 yr). The healthy subjects had no history of gastrointestinal diseases and were free of gastrointestinal symptoms. Body mass index (BMI) ranged from 19.8 to 21.1 kg/m². The EGG data were recording with water, yogurt, and a solid meal (pancake) stimulation, respectively. Every volunteer provided a written consent to participate in that study. The research project was approved by the Bioethics Committee of the Silesian Medical University. The archival data were recorded simultaneously with the EGG Medtronic System (Polygraf ID/Polygram Net) [29]. It allows the parameters of the EGG signal analysis to be compared. The archival signals were sampled with frequency equal to 200 Hz and discretized with 12 bits A/D resolution. The new EGG data were sampled with frequency equal to 250 Hz and discretized with 24 bits A/D resolution. The mean values of the overall dominant frequency for fast, and postprandial stage were found to be 2.96±0.21 cpm (cycles per minute), and 3.05±0.33 cpm, respectively. The mean values of the normogastria index (NI) for fast, and postprandial stage were found to be 91.9±11.7 %, and 82.4±13.52 %, respectively. The mean values of slow wave coupling (SWC) for fast stage between channels A1–A2, and A1–A4 were found to be 73.59±13.43 %, and 77.54±18.98 %, respectively, whereas for postprandial stage, the mean values of SWC between channels A1–A2, and A1–A4 were found to be 66.98±18.30 %, and 48.55±13.50 %, respectively. The values established in the validation process are consistent with typical human physiological values [14, 26]. Table 6 consists of the comparison of basic EGG parameters produced by EGG DWPack and Medtronic system for pre-prandial stage. The results are similar, and the slight differences can be caused by e.g. recoding process.

For validating HRV module, 1-hour length of the slp59 record of the MIT-BIH Polysomnographic Database [11] was used to construct the EGG and RR data. The computed results was compared against those obtained using gHRV software [43] and HRVAS software[42]. The default settings in gHRV were changed to the default settings of EGG DWPack:

I compared only six parameters: pNN50, RMSSD, VLF, LF, HF and LF/HF ratio. The results of HRV analysis are very similar. They are presented in Table 7. The parameters were calculated as mean of each segment (window) (for EGG DWPAck and gHRV). The most different parameters are absolute powers of VLF, LF and HF bands. It may be caused by the method of spectrum estimation or artifacts correction method. I also compared values of VLF, LF, HF and LF/HF ratio for 4-hour length of the slp59 record to results presented in [43]. This comparison is included in Table 8. The results of validation have proved that EGG DWPack software produces reliable outcomes.

In the presented paper, the author introduced the EGG DWPack software package (ver.2.0) for EGG signal analysis. The software package is easy to use through its user-friendly graphical interface. Due to the lack of standards of the EGG data, a new data format is proposed. Although only one simple file format is supported by the presented software, the software can be used for the wide range of EGG records. The conversion from any format to the proposed format should not be difficult. The software consists of various type of methods for the EGG signal analysis. The results of analysis are presented in graphic and text form. The user can set the parameters of the EGG analysis to their preferences (e.g. the range of frequency of normogastria rhythm, the length of EGG segment, the length of overlap). It is also possible to view each EGG segment and observe its spectrum. The software allows the basic parameters of the HRV signal (for EGG signal sampling with high frequency) to be analyzed as well. It allows the EGG and HRV to be simultaneously examined. The software provides the tools for preprocessing the EGG records: filtering, artifacts and noise removal, settings the time of the meal part, settings range of the analyzed signal. The software also includes the script for generating the test synthetic EGG data. The software allows calculating some parameters of the EGG examination. These parameters are calculated according to the definitions described in the literature [14, 29, 37]. It is possible to extend the range of computed EGG parameters because the majority of them are based on the spectra and values of the DF and the MPSD (DP). The presented version of the software is ready for use by the researchers and enthusiast rather than clinical users. For clinicians, the software requires further validation and creating the compiled version of proposed software, adding a module for generating and printing reports. But of course it is possible to use the software by an advanced clinical user (who know MATLAB and have basic knowledge in digital signal processing). The settings of the software options can slightly influence the obtained results (e.g. method and range of filtering, the method of computing spectra, and the length of segment, the overlap or the frequency range of normogastria). The user should be aware of these nuances.

For validating EGG DWPack, the synthetic and real EGG signals were used. The EGG software validation is a difficult task, because EGG records are not commonly and freely available (e.g. PhysioBank archives of PhysioNet (http://www.physionet.org) does not contain EGG data). The comparison to other software is also difficult due to the fact that any free available software for processing standard EGG signal is not known to author of this publication. The other limited available software is GEEMS [47]. That software is dedicated for analyzing and visualizing high-resolution (multi-electrode) recordings in spatiotemporal details. The described software will be probably the first free of charge and free available software for standard EGG data analysis. The majority parts of EGG DWPack software were validated and tested by the author over several years of dealing with the EGG signal processing [18, 19, 21, 39, 40].