Middle of the page     Bottom of the page    Selected Topics on Electrogastrography

5. ACCURACY OF EGG

5.1. INTRODUCTION.

In the last 10 years there have been several publications describing abnormalities in the gastric electrical activity (GEA) and cutaneous electrogastrogram (EGG) in a variety of clinical conditions (78 - 93). Unfortunately, the diagnostic value of EGG remains in question because although reportedly reliable in recognition of normal frequency and tachygastria, cutaneous recordings have not been shown to be accurate in assessing uncoupling and contractions (39, 42). Ideally abnormalities of gastric electrical activity would first be recognized with implanted electrodes and the diagnostic value of cutaneous EGG be confirmed by simultaneous recording with both techniques. Unfortunately this is rarely possible. In the current study invasively obtained gastric electrical and mechanical activity were compared to EGG and evaluated the reliability and usefulness of the four major properties of the cutaneous signal i.e., frequency, amplitude, phase lag and wave form.

5.1.1. FREQUENCY.

The gastric slow wave frequency can occasionally be recognized visually in cutaneous recordings, but it is most reliably assessed using computer-aided running spectrum analysis (RSA). Such analysis depends upon wave shape. Invasive recordings obtained with short distance (2-8 mm) bipolar (SDB) electrodes produce signals with waveshapes that preclude recognition of the fundamental frequency with RSA. Spectral comparison of records obtained cutaneously with those obtained from implanted electrodes is therefore difficult if not impossible. A technique has been developed which allows internal recordings to be analyzed with the fast Hartley transform and compared directly with cutaneous EGG signals. A quantitative comparison of the frequency changes in internal and cutaneous gastric electrical activity was also done.

5.1.2. AMPLITUDE.

Increased signal amplitude is often attributed to the appearance of electrical response activity (ERA), (12) and it is therefore assumed to indicate the presence of contractile activity. A significant increase in the amplitude of cutaneous gastric signals has been observed postprandially (9, 11, 12). In this study we determined the reliability of the relationship between amplitude changes in cutaneous EGG and the occurrence of gastric contractions.

5.1.3. PHASE LAG.

The direction of spread of gastric electrical activity over the stomach determines the direction of propagation of contractions. Chen et al.(41) described phase lag (time shift) between different cutaneous EGG channels recorded from fasting and postprandial patients. In the present study the use of cutaneous electrodes to recognize phase lag is reexamined.

5.1.4. WAVE FORM.

Familoni et al. (39) suggested that the propagation direction could be determined from the wave form of the cutaneous EGG. If with a specified electrode configuration, the ascending portion (arm) of each wave was shorter than the descending arm propagation of the EGG wave was considered aboral. This concept has been reexamined.

5.2. METHODS.

5.2.1. SUBJECTS.

These studies were performed on one patient with surgically implanted and cutaneous electrodes, four volunteers with cutaneous electrodes, another four volunteers with an intraluminal pressure (ILP) recording tube and cutaneous electrodes, and four anaesthesized dogs with serosal electrodes, serosal force transducers (FT) and cutaneous electrodes. In two of the dogs an intraluminal distending balloon was introduced also. The clinical study was approved by the Ethics Committee of the University of Alberta Hospitals and informed written consent was obtained from each subject before the procedure. The animal study was approved by the University of Alberta Health Sciences Animal Welfare Committee.

The patient was a 42 year old female with a 10 year history of recurrent anorexia, nausea, vomiting, heartburn and weight loss. Three years earlier a vagotomy and pyloroplasty had been done for duodenal ulcer without improvement in her symptoms. Investigation revealed slow gastric emptying with no organic obstruction and gastroesophageal reflux. She underwent fundoplication and insertion of three sets (3 cm apart) of stainless steel wire bipolar electrodes (0.5 cm apart) into the serosa of the antrum at operation. The technique was previously described (10). The electrodes were removed nine days following operation. Four abdominal surface electrodes (Hewlett Packard 14445) were attached on the abdominal surface overlying the serosal electrodes. Three 2-hour recordings of serosal and cutaneous electrical activity of the patient in fasting state were performed on subsequent days. Five additional recordings of 1 hour fasting and 1 hour postprandial activity were also performed on subsequent days. A standard 500 Kcal meal containing 68g of CHO (52%), 24g of protein (19%) and 17g fats (29%) was ingested during a twenty-minute break between the fasting and postprandial recordings.

Four volunteers underwent recording of the cutaneous EGG for 1 hour fasting and 1 hour after the test meal. It was assumed that the lesser curvature begins at the Xiphisternum and ends at the point where the midclavicular line meets the costal margin. The greater curvature is situated to the left and inferior to that line depending on gastric distention. The most proximal electrode was placed 5 cm left of the Xiphisternum on the costal margin, and a row of 4 more electrodes (3 cm apart) was placed linearly between the first electrode and the junction of the midclavicular line with the right costal margin. In four other volunteers a four-tube perfused intraluminal pressure monitoring assembly was introduced and positioned fluoroscopically in the gastric antrum. The recording assembly had four openings at 5 cm intervals from the distal tip. The tubes were perfused with water by an Arndorfer pump. Simultaneous 2-hour recordings of ILP and cutaneous EGG in the fasting state were obtained from each subject.

Four fasting dogs underwent a laparotomy under Penthotal anesthesia. Three sets of bipolar electrodes (0.5 cm apart) and force transducers were sutured to the antral serosal surface at 3 cm intervals. In two of the dogs a fundal gastrotomy was made and a 7 mm diameter tube with attached latex balloon for stomach distention placed into the antrum. Four cutaneous electrodes (3 cm apart) were placed on the shaved anterior abdominal wall. In the dogs and the patient with implanted electrodes two types of recordings of serosal electrical activity were made. The usual SDB recordings from pairs of electrodes placed close together were obtained. In addition by using electrodes from different SDB pairs new pairs were obtained to give a long distance bipolar (LDB) recording. The increment in distance between internal serosal bipolar electrodes produced signals with similar waveshape and spectral characteristics as those obtained cutaneously. Spike activity clearly recognized in SDB channels could not be seen in LDB recordings. After amplification and filtering the data were sampled at a rate of 10 Hz and an analog-to-digital conversion made.

5.2.2. FREQUENCY.

To study gastric frequency 2-hour experiments were performed on the patient (2 LDB, 2 SDB and 2 cutaneous bipolar channels, fasting and postprandial state), 4 healthy volunteers (2 cutaneous bipolar channels, fasting and postprandial state) and four unconscious fasting dogs (2 SDB, 2 LDB and 2 cutaneous bipolar channels). The experiments on the patient were repeated on 5 consecutive days after the operation. For the LDB and cutaneous channels a time constant of 10 s (cut-off frequency of the high pass filter of 0.017 Hz) and cut-off frequency of 0.3 Hz of the low pass filter were used, while for the SDB channels these parameters were 0.3 s and 4 Hz respectively.

To determine the reliability of registration of rhythm changes in LDB and cutaneous channels in the time domain, periods from the SDB with more then 25% change in duration than the preceding period were compared to the same periods in LDB and cutaneous recordings. When representing the frequency dynamics, best results were achieved by normalizing each spectrum in the power 3-D plot so that the fundamentals of all spectra had the same height. Power changes over time were therefore obscured, but the dynamics of the frequency peaks were clearly seen. Time-frequency plots of each channel were plotted as well using the methodology described in the previous section.

Top of the page    Bottom of the page    Selected Topics on Electrogastrography

5.2.3. AMPLITUDE.

Three different experiments were performed to determine the relationship between antral contractions and cutaneous EGG amplitude changes. A comparison was made between contractions noted in ILP recordings and increments in amplitude seen with cutaneous electrodes in 2-hour experiments with 4 healthy fasting volunteers (2 channels cutaneous EGG, 3 channels ILP). Recognizable contractions in an ILP channel were defined when the recorded pressure waves exceeded 25% of the maximal pressure recorded during the whole experiment in this channel. Similarly, recognizable increments in each separate cutaneous EGG channel were defined when the amplitude exceeded 25% of the maximal recorded EGG amplitude in this particular channel. In two fasting dogs (2 channels SDB, 2 channels LDB, 2 channels FT and 2 channels cutaneous EGG), contractions as demonstrated by serosal FT were compared with the changes in SDB, LDB and cutaneous EGG. The periods of spike bursts in SDB were also compared to contractions and amplitude changes in LDB and cutaneous EGG. Antral contractile activity was stimulated with 0.25 mg/kg Pentagastrin IV on four separate occasions. Twenty minutes after each Pentagastrin injection mechanical activity was blocked with 0.25 mg Glucagon. At the end of each experiment all antral contractile activity was blocked with 0.25 mg Atropine and 0.25 mg Glucagon. The antral balloon was then distended intermittently in stages after 200, 400 and 600 ml air were introduced. Each air inflation was done for approximately 20 seconds and the new air volume was maintained for 5 minutes. Four distentions were performed in each dog. Changes in cutaneous EGG amplitude were noted. The examined amplitudes were considered increased when they reached or exceeded the level of 25% from the maximum both for the EGG and the FT channels.

5.2.4. PHASE LAG.

In three 2-hour recordings from the patient in the fasting state, LDB and SDB serosal signals obtained from different areas of the stomach were compared with 4 channels cutaneous EGG, recorded from the orthogonal projections of those areas on the abdominal wall. Separate 2-hour experiments were also performed on 4 fasting dogs using a similar electrode configuration. In four volunteers 4-channel cutaneous EGG was recorded for 2 hours (1 hour fasting and 1 subsequent hour postprandially) using the above cutaneous electrode configuration. In all recordings from the patient, volunteers, and dogs the signal from the proximal electrode was compared to the signal from the distal one. It was pointed out previously (see the previous section) that with running cross-correlation analysis mean time shifts could be obtained. This approach was applied for 4.27 minute intervals and the obtained time shifts were arranged in a plot against time (time shift plot). The recordings were filtered (bandpass digital filter, 0.02-0.1 Hz) so that only the fundamental gastric component remained. The processing was done using the fast Hartley transform (FHT, 66) and the cross-correlation functions were successively obtained from the transform. The possibility for direct calculation of the cross-correlation function from the Hartley coefficients was mentioned by Bracewell (65, 66) and described previously. It was found to be more convenient for the time shift study than the method proposed by Chen et al. (41). Recorded time shifts were defined as "significant" if they exceeded 2.5 s (approximately a quarter of the half-period of the signals).

5.2.5. WAVE FORM.

The wave form was studied in the same experiments that measured gastric frequency. A cut-off frequency of 0.3 Hz was used instead of 0.5 or 1 Hz as recommended before (41) and thus most of the respiration artifacts were avoided, preserving the first three harmonics of the signal. To assess the changes in the wave form both ascending and descending arms of each wave were compared by a computer program which utilized the gradient method described in the previous chapter. The study was done separately in the fasting and postprandial state for the humans and before and after gastric distentions in dogs.

5.3. RESULTS.

5.3.1. FREQUENCY.

There was good correlation between the different recording methods. A comparison of visually evident frequency changes recorded with cutaneous and implanted electrodes is given in Table 5.1. Intermittent irregularities in gastric electrical activity were recorded from the patient during the first postoperative day. A well-defined peak of the fundamental gastric frequency in the spectral range of 2.54-3.67 cycles per minute (cpm) was recognized after the application of the Fast Hartley Transform (FHT) on successive 4.27 minute intervals of human cutaneous recordings. Use of the FHT (instead of the FFT) increased both the speed of spectral procedures and conversion speed .

Table 5.1. Recognition of changes in gastric electrical frequency by different recording methods.

   Computer frequency analysis could not be reliably applied to recordings obtained with electrodes implanted into the stomach in the usual manner i.e., short distance bipolar (SDB) recordings. The rapid bi-directional wave obtained in such studies yielded a broad frequency spectrum, in which the fundamental EGG component can be obscured. The use of pairs of implanted electrodes separated by several centimeters (long distance bipolar recordings, LDB) resulted in a signal that approximated a sine wave and resembled the cutaneous signal. This could be easily analyzed by the FHT. Comparison of frequency spectra obtained by implanted and cutaneous electrodes then became possible. Similar wave shapes and spectral characteristics were seen in LDB recordings and cutaneous EGG. Normalized 3-D plots were also very similar. The plots clearly showed postprandial frequency increments as reported before [9, 11, 12, 17, 45].

5.3.2. AMPLITUDE.

The amplitude of electrical signals from internal electrodes varied with distance from the pylorus in the patient (0.6 - 2.2 mV) and dogs (0.4 - 2.0 mV). Cutaneous EGG were also of higher amplitude in electrodes near the pylorus in humans (0.05 - 0.52 mV) and dogs (0.01 - 0.45 mV). Force transducer amplitudes ranged from 0 to 0.54 N and intraluminal recordings ranged from 0 to 127 cm H20. In dogs and humans amplitude changes in cutaneous EGG did not reliably reflect gastric contractions. Only a minority of increased EGG signal amplitudes was associated with contractions and the majority of contractions was not associated with increased EGG amplitude (Fig.5.1, Fig.5.2).

Figure 5.1. Relationship between gastric contractions and changes in EGG amplitude in normal volunteers: (A) incidence of increased EGG amplitudes during gastric contractions and (B) incidence of gastric contractions during increased EGG amplitudes.

Figure 5.2. Incidence of gastric contractions when (Left) SDB channels displayed spikes; (Center) Increased signal amplitude was observed in LDB channels; and (Right) Increased signal amplitude was observed in EGG recordings.

Distention of the atonic canine stomach with a balloon increased the amplitude of EGG signals even when gastric contractile activity was completely abolished by Atropine and Glucagon (Fig.5.3).

Figure 5.3. Effect of gastric distention on incidence of increased EGG signal amplitudes.

Most of the cutaneous EGG amplitudes increased significantly after air infiltration, while force transducer channels and signals recorded with internal electrodes did not change (Fig. 5.4).

Figure 5.4. Increased EGG amplitudes (lower tracing) after an air distention of a canine stomach. An arrow indicates the exact moment of distention. Contractile activity (upper tracing) was blocked with athropine and glucagon.

5.3.3. PHASE LAG.

In the patient with implanted and cutaneous electrodes time shift (phase lag) between the most distal and the most proximal implanted electrodes was consistently between 5 and 8 seconds. In cutaneous recordings, however, no time shift was recorded in 2 of the 6 recording hours. In the remaining 4 hours the time shift was very small ranging from 0.5 to 1.8 seconds. The time shift obtained from the four volunteers was less consistent in the fasting state, probably due to the worse signal-to-noise ratio. After a test meal it became stable but remained close to 0. The time shift obtained from internal electrodes in dogs varied from 3 to 7 seconds. Cutaneous recordings, however, revealed small but consistent time shift of 0.5 to 1.8 seconds for 64 % of the 8 hour recording time.

5.3.4. WAVE FORM.

In the fasting state wave forms displayed small predominance of the descending arms or symmetry in all subjects. After feeding of the humans the descending arm in cutaneous waves dominated more evidently, but still this predominance was insignificant. Distention of the canine stomachs had little effect on the wave form of LDB signals, but slightly increased the preponderance of the descending arms in cutaneous EGG waves. The differences between the ascending and descending arms in all recordings were small and gradient plots in all measurements showed very little deviation from symmetry in favor of the descending arms.

5.4. DISCUSSION.

Each single gastric antral muscle cell undergoes periodic depolarizations and the muscular array simulates a series of relaxation oscillators i.e., higher frequency oscillators pull up lower frequency oscillators to the higher frequency. The highest frequency oscillator in man lies high in the greater curvature and periodically depolarizes 3 times per minute. Entraining all distal oscillators whose intrinsic frequency is lower to this frequency results in an aboral spreading wave of depolarization. Electrical response activity (ERA) which heralds contractions is superimposed upon electrical control activity (ECA) so the frequency and direction of propagation of ECA waves determines the frequency and direction of propagation of contractions.

Abnormalities in gastric electrical activity could have a profound effect on gastric motor function so the recent quickening of interest in the electrogastrography is very understandable. Using EGG, a large and growing number of clinical conditions have been associated with gastric electrical abnormalities. Before accepting these findings as "true bills", two vital questions demand answers:

1. Does the EGG accurately reflect gastric electrical activity?

2. Does the reported electrical abnormality have an effect on gastric motor function?

Most recorded EGG abnormalities have been changes in frequency, usually reported as bradygastria or tachygastria. Several authors have suggested that increments in EGG amplitude are associated with antral contractions (9, 11, 12, 45). The present study clearly demonstrates that the amplitude of EGG signals can increase significantly when there are neither antral contractions nor spikes in the SDB channels. A changed or changing position of the antrum with respect to the skin electrodes is sufficient to cause increased amplitude in cutaneous EGG. Such changes could for example be the result of feeding or any other mechanical displacement, which leads to a change in the distance between the source of the electrical field in the stomach and the cutaneous electrodes. Running cross-correlation analysis based on direct application of the FHT facilitated the study of time shift dynamics.

Chen et al. (41) have reported occasional significant phase lags (or time shifts) between different sets of cutaneous electrodes. However we could not confirm these results. No significant (above 2.5 s) time shift were noted in the different sets of cutaneous electrodes. It is possible that the EGG signal is so integrated, that the actual time shifts recorded internally cannot be registered on the abdominal wall or they are very small.

Familoni et al.(39) suggested that orad propagation of gastric electrical signals would result in prolongation of the ascending arm of the waves. In this study the descending arm dominated in both internal LDB and cutaneous recordings in humans and dogs. After feeding or gastric distention this preponderance became more marked in cutaneous recordings, probably due to an improved signal-to-noise ratio. Preponderance of the ascending arm of the waves was rarely observed. Unfortunately orad propagation of gastric slow waves was not seen during this study and the hypothesis suggested in (39), could not be directly tested.

Electrogastrography, although a very appealing non-invasive technique is not currently of clinical diagnostic value. There is insufficient evidence that 1) abnormalities in EGG are commonly present in patients with gastric motor dysfunction, and 2) cutaneous EGG can actually record gastric electrical abnormalities. Only gastric electrical frequency can be assessed with good reliability by EGG. Therefore, at the present time the information coded in gastric frequency determines the clinical value of EGG.

Top of the page    Middle of the page    Selected Topics on Electrogastrography