6. DERIVATION OF AN OBJECTIVE QUANTITATIVE METHOD FOR CLINICAL EGG APPLICATIONS
Many authors (82-89, 92-94, 97, 103, 111) attempted to use electrogastrography for clinical assessment of gastric motility disorders. Unfortunately, only qualitative (and therefore subjective) methods of evaluation of the EGG were suggested. In this study an attempt is made to derive an objective quantitative method of evaluation of gastric electrical activity including EGG. In order to establish the differences between normal and abnormal EGG patterns a study of normal volunteers was performed. The discussion in the previous chapters imply several important suggestions for the direction of the study:
(a) the frequency of the EGG signals reliably represents gastric electrical frequency and at the present time it is the only parameter that can be used for clinical evaluation of EGG;
(b) time-frequency plots obtained with or without overlap are a convenient way to represent dynamics of the dominant spectral component of EGG signals during the experiment;
(c) gastric electrical activity in normal subjects is assumed to be regular and coupled, therefore frequencies recognized from different EGG channels should be ideally the same, but since the recordings are performed in vivo some minor deviations are expected;
(d) overlap of the time intervals used in calculation of time-frequency plots can affect the stability of these plots.
This study was performed on 4 patients with antral serosal electrodes implanted using the technique described before (42), 15 healthy volunteers (6 male and 9 female) without any history of gastrointestinal complaints for one hour in the fasting state and one consecutive hour postprandially with a short (15-20 min.) interruption for feeding. A standard 500 Kcal meal was given to the subjects for postprandial studies.
Internal short distance bipolar (SDB) gastric electrical signals were recorded for 1 hour in the fasting state and 1 extra hour postprandially. The patients were tolerating normal diet and their gastric electrical activity was qualified as normal after a careful visual inspection by two independent evaluators. Gastric electrical activity (GEA) waves were counted in 4.27 min. intervals without any overlap, and with 75% overlap between the intervals. The mean frequency (MF) value was obtained for each interval.
To record the EGG a row of five standard neonatal EKG monitoring electrodes numbered from 1 (most proximal) to 5 (most distal) was placed along the abdominal projection of the stomach axis on each subject. The distance between the centers of two electrodes in the row was 2.5 cm. Eight bipolar EGG channels were recorded using the same electrode combinations for all subjects. Table 6.1. shows the electrode combinations used. The electrode surface was kept constant.
Table 6.1. Electrode combinations used for different EGG channels. Electrode 1 was the most proximal, Electrode 5 - the most distal.
The settings of the equipment were the same as in section 5 (experiments with cutaneous electrodes). The signals were acquired with a 10 Hz sampling frequency. In off-line studies however, the sampling frequency was reduced to 2 Hz after digital filtering with sampling filters in the frequency band 0.02- 0.1 Hz, i.e. only the first one or two EGG frequency components were retained for further processing. Separate time-frequency plots were obtained in the fasting and postprandial states from each subject. The plots were built from successive spectra obtained after 512-point FHT of 4.27 min. intervals of the recorded EGG signals. Plots obtained from different EGG channels were superimposed on the computer screen using different colors. The impact of 75% overlap on the time-frequency plots was investigated.
To evaluate quantitatively MF values obtained from the patients with implanted electrodes, and EGG recorded from the volunteers, statistical evaluation of the points that built up each time-frequency plot was performed including calculation of the mean value, variance and the standard deviation as well as computation of the probability density function of the frequencies present in the plots (see section 4).
It was assumed that the standard deviation was indicative of the level of stability of the time-frequency plot in a given GEA channel. Instability of the time-frequency plot could be caused by two factors:
(a) irregularities in gastric electrical frequency;
(b) external factors (noise, motion artifacts, etc).
Keeping in mind that the probability density function (pdf) shows the distribution of the points that represent different GEA frequencies in the time-frequency plot, it is obvious that the maximum of this function for a given GEA channel (implanted or cutaneous) would be at the frequency that is most often present in this channel during the whole test. If the pdf maxima derived from the time-frequency plots of different GEA channels coincide, the same frequency dominates in these channels. Lack of coincidence would indicate an absence of the same dominating frequency in different GEA channels.
6.3.1. STATISTICAL EVALUATION OF TIME-FREQUENCY PLOTS.
Mean value, variance and standard deviation (called here basic averages) of the points that built up the time-frequency plots of different GEA (regardless whether they were obtained with implanted or cutaneous electrodes) channels were calculated to evaluate the stability of the dominant spectral component.
Standard deviations of the MF values calculated from the three GEA channels recorded from the patients with implanted electrodes varied from 0 to 0.450 cycles-per-minute (cpm), and their probability density functions were bell-shaped, with coinciding maxima. The GEA frequency ranged from 2.5 to 3.75 cpm.
The overall results from all volunteers showed that at least in three out of eight EGG channels the standard deviation of the dominant frequency component as assessed from the time-frequency plots was less then 0.450 cpm. These channels were qualified as stable. The frequency range as assessed by the mean frequency in the stable channels was in the range 2.5 - 3.75 cpm. Typical time-frequency plots obtained from a healthy volunteer and its statistical evaluation are shown on Fig. 6.1 and Table 6.2 respectively.
Figure 6.1. Typical time-frequency plots obtained for each EGG channel recorded from a healthy volunteer.
Table 6.2. Basic averages obtained from the time-frequency plot on Fig.6.1. Only one EGG channel showed standard deviation higher then 0.450 cpm.
Maxima of the probability density functions (pdf) obtained from the time frequency plots of the stable EGG channels coincided at the frequency of the dominant spectral component and the pdfs were bell-shaped (Fig. 6.2).
Figure 6.2. Probability density functions of the points that built up the time-frequency plots of Fig. 6.1. Maxima coincided at 3.047 cpm, but some other frequencies were also present.
6.3.2. IMPACT OF THE OVERLAP.
Introducing a 75% overlap of the time domain intervals when building time-frequency plots improved their stability (Fig.6.3), reduced the standard deviation (Table 6.3) and narrowed the probability density functions (Fig.6.4) in 8 out of 15 cases. In 3 out of 15 cases the overlap had a negative effect on the stability of the time-frequency plots, worsened the standard deviations and widened the probability density functions. In all these cases, however, at least three EGG channels exhibited standard deviations less then 0.450 cpm and the maxima of their probability density functions coincided. Four out of 15 recordings were not influenced by the overlap at all, i.e. the stability of the time-frequency plots obtained without any overlap remained the same after the overlap was introduced.
Figure 6.3. Time-frequency plot obtained with 75% overlap from the same set of EGG data used to obtain the time-frequency plots shown on Fig. 6.1.
Table 6.3. Basic averages of the time-frequency plots shown on Fig. 6.3. Channels 3-8 showed excellent stability.
Figure 6.4. Probability density functions of the time-frequency plots obtained with 75% overlap were narrower than these shown on Fig. 6.2. Coincidence of their maxima was more evident and the pdfs became narrower.
6.3.3. CRITERION FOR STABILITY.
The above results indicated that a criterion for stability of the dominant frequency component as assessed from the time-frequency plots of different EGG channels could be established as follows: The standard deviation of points that build up the time-frequency plots obtained with or without 75% overlap (whichever technique provided minimal standard deviation) should be less than 0.450 cpm in at least 3 out of 8 EGG channels recorded with standard disposable neonatal EKG electrodes positioned in a row collinear with the abdominal projection of the stomach axis. An important question is how to assess stability when for different reasons fewer or more than 8 EGG channels are recorded. Of course, the first recommendation would be to use the methodology described above and keep the number of channels at 8. When this is impossible the ratio between the number of stable channels and the total number of channels should be as close to 3/8 as possible, keeping in mind that reducing the total number of channels worsens the precision of the suggested method. Another relevant issue is whether it matters which three channels exhibit stable recordings. The model explicitly showed that cutaneous recordings are subjected to a strong integration. That is why, only minor differences between the EGG channels are to be expected, so it would be very difficult, if not impossible to relate some of the channels to distinct areas of the stomach. Therefore, all electrode combinations used at the present time should be regarded as equally informative.
6.3.4. CRITERION FOR COUPLING.
The assumption that EGG signals recorded with standard EKG electrodes are quite integrated, which was confirmed by the computer modeling in section 2, implies that only minor differences (if any) should be expected in the EGG frequencies recorded from different channels. Of course, in the normal patient one should expect a complete coincidence of the maxima of the probability density functions obtained from the time-frequency plots. This was the case in this study. The criterion for electrical coupling was established as follows: The maxima of the probability density functions of the points that build up the time-frequency plots of all stable EGG channels (as determined by the criterion for stability) should coincide at the same frequency (between 2.5 and 3.75 cpm). The suggested criteria are not independent - poor stability of the time-frequency plots can jeopardize the coincidence of the maxima of the probability density functions, although relatively good coincidence was observed even when the standard deviations of some time-frequency plots were in the range of 0.6 - 0.7 cpm.
In this section two quantitative criteria for normality were introduced. They were extracted from the results obtained from 4 patients with implanted electrodes and normal GEA, and 15 healthy volunteers with cutaneous electrodes only. The quantitative criteria for evaluation of EGG were extracted from the internal GEA recorded from the patients.
The first criterion determines the stability of the dominant spectral component of gastric electrical frequency as recognized from different EGG channels, i.e. it assesses cutaneously gastric irregularities or/and changes in gastric electrical frequency. Unfortunately, random motion artifacts, respiration, or any other external disturbances can mislead the investigator if they are consistent, with high amplitude and long duration. This is especially dangerous when patients with unexplained nausea and vomiting and anorexic patients are studied, because motion artifacts are usually related to their condition and are more or less inevitable. In these cases the recordings could be repeated several times with more careful preparation or they could be split in shorter recording intervals, which could be combined during the evaluation process.
The second criterion assesses the electrical coupling between different EGG channels. Despite te fact that EGG signals recorded with standard EKG electrodes appear to be strongly integrated, eventual uncoupling should manifest itself with differences in the frequency of dominant spectral components obtained from different EKG channels, i.e. the maxima of the probability density functions extracted from the time-frequency plots would not coincide in the normal frequency range of 2.5 - 3.75 cpm. This criterion can be considered dependent on the criterion for stability, because greater standard deviations inevitably indicate wobbling of the dominant spectral component and this would make the probability density function wider. Despite that, if the pdf maxima obtained from the time-frequency plots of different EGG channels coincide, the requirements of this criterion are met. Eventual elimination of some of the external factors for instability in EGG recordings (mainly motion and respiration artifacts) would allow an additional requirement to be established for the pdf maxima to be greater than a certain percentage value.
7. CLINICAL EGG STUDY.
In the previous section normal EGG was defined through the development of two objective quantitative criteria. In this section the usefulness of these criteria in differentiating normal from abnormal GEA in the clinical situation is assessed. Several groups of patients were studied:
(a) patients with implanted and cutaneous electrodes;
(b) patients with partial gastric resection (gastrectomy);
(c) ileus (postoperative) group of patients;
(d) patients with idiopathic constipation;
(e) patients with other gastrointesinal complaints.
Studies in the first group of patients allowed a direct comparison between internal and cutaneous signals. Indirect testing of the method was done in the second, third and fourth group of patients. Gastric electrical abnormalities are most likely to be present in patients with partial gastrectomy and in patients during the first postoperative days (ileus group). Constipated patients would be expected to have normal GEA.
7.2. POSSIBLE PATTERNS OF GASTRIC ELECTRICAL ABNORMALITIES RECOGNIZED BY THE EGG.
The two criteria introduced in section 6 combined with the established normal range of EGG frequency as recorded from volunteers indicate that six possible patterns of gastric electrical abnormalities might be recognized with the suggested method:
1. Irregular gastric electrical activity with normal electrical coupling of different parts of the stomach. The standard deviations of more than 6 out of 8 standard EGG channels should be higher than 0.450 cpm. The maxima of the probability density functions of at least 3 out of 8 channels, however, must coincide in the normal frequency range of 2.5 - 3.75 cpm.
2. Irregular gastric electrical activity combined with electrical uncoupling of different parts of the stomach. The standard deviations fall out of the normal range, as described #1, and the maxima of the probability density functions of more than 6 out of 8 standard EGG channels do not coincide in the normal EGG frequency range.
3. Regular gastric electrical activity in the normal frequency range, but abnormal electrical coupling of different parts of the stomach. Gastric electrical frequency is in the normal range and remains stable (standard deviations less than 0.450 cpm) in at least 3 out of 8 EGG channels, but the maxima of the probability density functions of at least 6 out of 8 standard EGG channels do not coincide in the normal frequency range.
4. Regular gastric electrical activity outside the normal frequency range combined with normal electrical coupling. Gastric electrical frequency is outside the normal frequency range, but is stable in at least 3 out of 8 channels. The maxima of the probability density functions of at least 3 channels coincide, but outside the normal frequency range.
5. Irregular gastric electrical activity combined with normal electrical coupling, but outside the normal frequency range. Gastric electrical frequency is not stable in more than 6 out of 8 channels. The maxima of the probability density functions coincide, but outside the normal frequency range.
6. Regular gastric electrical activity outside the normal frequency range and abnormal electrical coupling. Gastric electrical frequency is outside the normal frequency range, but is stable in at least 3 out of 8 channels. The maxima of the probability density functions of at least 6 out of 8 channels do not coincide.
It should be made clear, that in certain occasions the registered frequency irregularities from EGG might be due to external factors, not actual gastric electrical frequency irregularities. These factors include sudden but consistent motion artifacts and poor signal-to-noise ratio in patients with very high body mass index (calculated as a ratio between the height and the weight). In studies, where there was a suspicion of excessive motion artifacts, the tests were repeated. Elimination of the intervals with motion artifacts, or even interruption of the recording during these intervals (especially during periods of nausea and vomiting) could help also. When recording from obese patients repetition of the test would not help, because the high resistance of the abdominal fat is a factor which is difficult to eliminate. More careful preparation of the skin and higher gains of the amplifier can help in these situations.
7.3. SUBJECTS AND SETTINGS.
The study was performed on two postoperative patients (one male and one female) with suspected gastric motility disorders (6 one-hour recordings during the fasting state and 4 one-hour recordings during the postprandial state for five consecutive postoperative days), with implanted short distance bipolar (SDB), long distance bipolar (LDB) and cutaneous electrodes overlying the implanted pairs and 30 other patients with a variety of gastrointestinal complaints (1 hour in fasting state and one consecutive hour postprandialy) with cutaneous electrodes only. A standard 500 Kcal meal was given to the subjects for postprandial studies. To record the EGG a row of five standard neonatal EKG monitoring electrodes was placed along the abdominal projection of the stomach axis on each subject. The distance between the centers of two electrodes in the row was 2.5 cm. Eight bipolar channels cutaneous EGG were recorded using the same electrode combinations as in the study of the 15 healthy volunteers. The electrode surface was kept constant.
The female patient with implanted electrodes was the 42-year-old woman whose medical history was discussed in section 5.
The second patient with implanted electrodes was a 52-year-old male patient with a past history of pyloroplasty and vagotomy for peptic ulcer disease in 1965, who underwent fundoplication and implantation of an electrode set similar to the one used on the female subject. The standard configuration of the EGG electrodes overlying the internal pairs was used. The duration of the tests was the same as in the study of the female patient.
The other patients were classified with respect to their major complaint or clinical condition into several groups (Table 7.1). The settings of the equipment were the same as in the previous chapter. Body mass indexes, weight, pulse, maximal and minimal chest circumferences as well as the abdominal circumference were measured prior to the recordings.
Table 7.1. Number of patients with identical or similar complaints.
7.4.1. COMPARISON BETWEEN INTERNAL AND CUTANEOUS RECORDINGS IN THE TWO POSTOPERATIVE PATIENTS.
The female patient exhibited normal gastric electrical activity in records obtained with internal electrodes both in terms of frequency regularity and electrical coupling. When processed with the technique described in section 4, the EGG recordings also displayed normal gastric electrical activity. Typical time-frequency plots of all channels, basic averages extracted from them and probability density functions are shown on Fig.7.1, Table 7.2 and Fig. 7.2 respectively. In order to be processed with the suggested method, SDB channels were filtered with a very narrow bandpass digital filter (0.03 - 0.08 Hz) because the first harmonic (which represents the period of the signal) of the signal was with very low power. Only this harmonic was left for further processing.
Figure 7.1. Typical time-frequency plots recorded from the female patient.
Table 7.2. Basic averages of the time-frequency plots from Fig. 7.1. Only one channel was out of the range of stability.
Figure 7.2. Probability density functions obtained from the time-frequency plot on Fig.7.1. All maxima coincided in the normal frequency range.
Only in one out of 8 recordings from the male subject did the implanted SDB electrode recordings show good coupling and few frequency irregularities. In this particular recording the EGG processed with the suggested technique showed good stability of the frequency in 2 out of 5 channels. Taking into consideration that the established criterion for stability was valid for 3 out of 8 EGG channels the above result does qualify as normal. In the other seven recordings the implanted electrodes displayed abnormal GEA. Simultaneous EGG recordings were also abnormal (Pattern 2) in all seven recordings. The two criteria for normality were not met (Fig.7.3, Table 7.3, Fig.7.4).
Figure 7.3. Typical time-frequency plots recorded from the male patient.
Table 7.3. Basic averages of the time-frequency plots on Fig.7.3. Standard deviations of all EGG channels were out of the normal range.
Figure 7.4. Probability density functions obtained from the time-frequency plots on Fig.7.3. The maxima did not coincide.
7.4.2. PATIENTS WITH PARTIAL GASTRECTOMY.
All four patients with partial gastrectomy (3 female and 1 male) exhibited abnormal EGG. One of the patients had a normal EGG in the fasting state, while the postprandial recording was abnormal. In one other patient the opposite was noted. Two of the patients (one of them postprandially only) displayed unstable EGG frequency and different maxima of the probability density functions (abnormal pattern 2). One patient had regular EGG frequency and good electrical coupling both in fasting and postprandial states, but the mean value of the frequency was outside the normal range of 2.5 - 3.75 cpm seen in volunteers. This was defined earlier as an abnormal pattern 4. In one other fasting patient the EGG frequency was regular but very low (below 2.5 cpm), while the electrical coupling was abnormal, which was described as abnormal pattern 6. After feeding, the mean EGG frequency of this patient entered the normal range probably due to the posprandial increment of gastric frequency seen in some subjects. However, the pdf maxima coincided, which indicated improved electrical coupling and thus qualified the whole postprandial recording as normal.
7.4.3. ILEUS PATIENTS.
The three patients from this group underwent cutaneous recordings of GEA during the first 48 hours after a laparotomy. In all subjects the recordings clearly demonstrated abnormal EGG (pattern 2) both in the fasting and the postprandial states. In two of patients the tests were repeated to make sure that the results were not influenced by external noise factors. Because of the postoperative condition of the patients, the standard meal for the postprandial studies was replaced with liquids.
7.4.4. PATIENTS WITH IDIOPATHIC CONSTIPATION. In the fasting state, all patients (1 male and 6 female) had normal EGGs. The male patient showed an abnormal EGG (pattern 2) after feeding, i.e. his EGG frequency was not stable in more than five channels, and the pdf maxima of the EGG channels did not coincide.
7.4.5. PATIENTS WITH OTHER GASTROINTESTINAL COMPLAINTS.
This group of patients was divided into four subgroups (see Table 7.1), each of which will be considered separately. Patients with unexplained epigastric pain. In this subgroup of 8 subjects, 6 were normal. One patient had abnormal EGG (pattern 2) only during the fasting state. One other patient displayed abnormal EGG (pattern 4; stable EGG frequency outside the normal frequency range and good electrical coupling) both in the fasting and postprandial states. Patients with unexplained epigastric pain and distention. Two out of four patients in this subgroup exhibited abnormal pattern 2 (irregular EGG frequency and electrical uncoupling). In one of them this was noted in fasting state only, while the postprandial recording was normal. The other two patients showed normal EGG pattern. Patients with simple gastro-esophageal reflux. One out of 2 patients displayed an abnormal EGG (pattern 6; stable EGG frequency outside the normal range and abnormal electrical coupling) in fasting state only. The postprandial recording from this patient, as well as the recordings from the other patient (both in fasting state and after feeding) were normal. Patients with unexplained nausea and vomiting. The only patient with this complaint showed an abnormal EGG (pattern 2) both in the fasting state and after feeding. Patients with post obesity surgery. This patient exhibited normal EGG both in fasting and posprandial states. The test was performed several months after the surgery.
7.5. STUDY OF THE ABNORMAL EGG PATTERNS.
7.5.1. ABNORMAL ELECTRICAL ACTIVITY - PATTERN 2.
Table 7.4 shows the number of patients with similar complaints who had abnormal gastric electrical activity with the second defined pattern (irregular gastric electrical activity and abnormal electrical coupling). Altogether 12 out of 30 patients (40 % from the total number), exhibited this abnormal EGG pattern. In 3 of these cases the tests were repeated on a different day because of suspicion that external motion artifacts influenced the results. In all three cases, however, the new tests confirmed the findings.
Table 7.4. Quantitative comparison between abnormal (Pattern 2) and total number of patients with identical or similar complaints.
Typical time-frequency plots, basic averages and probability density functions obtained from a patient which with abnormal gastric electrical activity (Pattern 2) are shown on Fig. 7.5, Table 7.5 and Fig. 7.6 respectively. Time frequency plots were unstable. Standard deviations calculated from these plots were greater than 0.450 cpm in at least 6 out of 8 EGG channels. The maxima of the probability density functions did not coincide.
Figure 7.5. Unstable time-frequency plots recorded from a patient with abnormal gastric electrical activity (Pattern 2).
Table 7.5. Basic averages obtained from a patient with abnormal gastric electrical activity (Pattern 2). Only 2 EGG channels were within the normal frequency range.
Figure 7.6. Probability density functions calculated from the time-frequency plots of different EGG channels. The maxima did not coincide in the normal frequency range.
7.5.2. ABNORMAL ELECTRICAL ACTIVITY - PATTERN 4.
Table 7.6 shows the number of patients with similar complaints who exhibited abnormal gastric electrical activity with the fourth pattern - regular gastric electrical activity outside the normal frequency range combined with good electrical coupling of different parts of the stomach. This pattern was observed in 2 out of 30 patients (6.7 % from the total number). Typical time-frequency plots, basic averages and probability density functions are shown on Fig. 7.7, Table 7.7 and Fig. 7.8 respectively. Standard deviations calculated from the time-frequency plots in at least 3 out of 8 EGG channels were less than 0.450 cpm, but the maxima of the probability density functions coincided at frequency outside the normal frequency range.
Table 7.6. Quantitative comparison between abnormal (Pattern 4) and total number of patients with identical or similar complaints. Complaint Total Abnormal Abnormal Abnormal Patients Patients Patients Patients (Pattern 4, fasting) (Pattern 4, (Pattern 4, fed) Total) Previous Partial 4 1 0 1 Gastrectomy Epigastric Pain 1 1 1 1
Figure 7.7. Time-frequency plots recorded from patient with abnormal GEA (Pattern 4).
Table 7.7. Basic averages obtained from a patient with abnormal gastric electrical activity (Pattern 4).
Figure 7.8. Probability density functions calculated from the time-frequency plot on Fig. 7.9. Note that the maxima coincided, but outside the normal frequency range.
7.5.3. ABNORMAL ELECTRICAL ACTIVITY - PATTERN 6.
Table 7.8 shows the number of patients with similar complaints who exhibited abnormal gastric electrical activity with the pattern 6 - regular gastric electrical activity outside the normal frequency range, but electrical uncoupling of different parts of the stomach. The pattern was observed in 2 out of 30 patients (6.7 % from the total number). Typical time-frequency plots, basic averages and probability density functions are shown on Fig. 7.9, Table 7.9 and Fig. 7.10 respectively. Stable frequency outside the normal range was registered, but the maxima of the probability density functions did not coincide in more than 2 EGG channels.
Table 7.8. Quantitative comparison between abnormal (Pattern 6) and total number of patients with identical or similar complaints.
Figure 7.9. Time-frequency plots recorded from patient with abnormal GEA (Pattern 6).
Table 7.9. Basic averages obtained from a patient with abnormal gastric electrical activity (Pattern 6). More than 3 EGG channels exhibited stability, but outside the normal frequency range.
Figure 7.11. Probability density functions calculated from the time-frequency plots on Fig. 7.11. The maxima did not coincide well.
7.5.4. ABNORMAL ELECTRICAL ACTIVITY - OTHER PATTERNS.
Other abnormal patterns were not seen in any of the studied patients, although they were defined theoretically. Irregular gastric electrical activity but normal electrical coupling either inside or outside the normal frequency range (patterns 1 and 5) probably could not be recorded because of the possible inter relation between gastric electrical irregularities and uncoupling.
In this study a method for extraction of valuable diagnostic information from cutaneous gastric electrical signals using current amplification and electrode techniques has been developed and tested. The objective was to transform electrogastrography (EGG) from a research tool into a useful clinical measurement based on an objective quantitative method. The main advantage of the proposed method for evaluation of EGG is its potential to become more reliable with future technological development. More powerful amplification methods or better digital signal processing procedures, for example, could make the suggested criterion for stability even more precise by decreasing the standard deviation required to qualify an EGG channel as stable. Stomach distention after feeding could affect signal-to-noise ratio and thus the results of the analysis. The impact of postprandial distentions as well as the influence of other external noise factors on the quality of the recording could be minimized with the future developments in amplification technology. Future studies could explore the extent of the discovered gastric electrical abnormalities. Work needs to be done to relate more reliably the abnormal patterns of EGG to certain clinical conditions. Clinical manifestation (if any) of the abnormal patterns that were not clinically seen in this study could also be examined. One interesting issue is the interrelation between electrical uncoupling of different parts of the stomach and irregularities of gastric electrical frequency. The problem is whether these two entities manifest themselves independently, or does the appearance of one lead to the appearance of the other as well. An important benefit of the proposed method is that it is realistic. It was clearly demonstrated that the most reliable EGG parameter is gastric electrical frequency. Its good level of stability in all normal subjects and unconscious dogs studied suggested that external artifacts do not drastically affect the recordings when the experiment is carried out carefully. With the described method EGG makes its first steps towards reliable clinical applications. Similarly to electrocardiography in its early years, electrogastrography has the potential to become a routine clinical procedure in the near future.