Perioperative Gastrointestinal Myoelectric Activity Measurement Using Wireless External Patches

Abstract

Background

Stomach, small intestine, and colon have distinct patterns of contraction related to their function to mix and propel enteric contents. In this study, we aim to measure gut myoelectric activity in the perioperative course using external patches in an animal model.

Methods

Four external patches were placed on the abdominal skin of female Yucatan pigs to record gastrointestinal myoelectric signals for 3 to 5 days. Pigs subsequently underwent anesthesia and placement of internal electrodes on stomach, small intestine, and colon. Signals were collected by a wireless transmitter. Frequencies associated with peristalsis were analyzed for both systems for 6 days postoperatively.

Results

In awake pigs, we found frequency peaks in several ranges, from 4 to 6.5 cycles per minute (CPM), 8 to 11CPM, and 14 to 18 CPM, which were comparable between subjects and concordant between internal and external recordings. The possible effect of anesthesia during the 1 or 2 hours before surgical manipulation was observed as a 59% (± 36%) decrease in overall myoelectric activity compared to the immediate time before anesthesia. The myoelectrical activity recovered quickly postoperatively. Comparing the absolute post-surgery activity levels to the baseline for each pig revealed higher overall activity after surgery by a factor of 1.69±0.3.

Conclusions

External patch measurements correlated with internal electrode recordings. Anesthesia and surgery impacted gastrointestinal myoelectric activity. Recordings demonstrated a rebound phenomenon in myoelectric activity in the postoperative period. The ability to monitor gastrointestinal tract myoelectric activity non-invasively over multiple days could be a useful tool in diagnosing gastrointestinal motility disorders.

Introduction

Gastrointestinal (GI) motility is a complex process, driven by the interplay between the enteric nervous system, paracrine and endocrine signals, and inherent visceral smooth muscular contractions to generate peristaltic activity from the esophagus to the rectum. Each organ has a distinct rhythmic contraction pattern to mix and propel ingested contents related to its digestive function along the gastrointestinal tract. Slow wave activity (rhythmic smooth muscle membrane depolarization and repolarization) exists from the stomach to the colon, which is initiated by interstitial cells of Cajal that facilitate neuromuscular interaction [1]. While these slow waves do not cause smooth muscle contraction, they appear to drive the natural peristaltic frequency of each organ when contractions do occur: in humans, stomach at 3 cycles per minute (CPM), duodenum at 12 CPM diminishing to 6 to 8 CPM in the ileum, and colon ranging up to 40 CPM [2,3].

The natural functions of these organs are often altered in the surgical setting, frequently leading to post-operative ileus or gastrointestinal dysfunction. The estimated incidence of postoperative ileus in the adult population ranges from 10 to 50%, which costs the United States healthcare system approximately $1.5 billion for the index hospitalization and 30-day readmission [4].

The pathophysiology of ileus after surgery is quite complex, with many risk factors associated with its development, including increasing age, length of surgery, and opioid use [5=7]. Our group developed a wireless skin patch myoelectric measurement system to detect myoelectric signals of different parts of the GI tract [8-10]. We have demonstrated the correlation between external patch measurements and internally measured myoelectrical signals in an animal model. In that study, we demonstrated concordance between internal electrode and external patch recordings in four pigs, primarily qualitatively, and reported the measured organ frequencies of this commonly used animal model [11]. Cutaneous patches were also used to record gastrointestinal myoelectric activity in adult and pediatric patients post-operatively [8-10]. Based on our results, in adult patients after colonic surgery, early colonic motility predicted flatus up to five days in advance [10], and monitoring gastric myoelectric activity in post-operative pancreaticoduodenectomy patients predicted patients at risk for delayed gastric emptying [8].

In this study, we aim to further investigate the accuracy of external patch measurements for detecting alterations in the myoelectrical activity of the GI system throughout the perioperative course. The current publication presents comprehensive results of the full study cohort of seven pigs, with quantitative analyses of concordance between internal and external measurements, the intra-pig consistency of the organ frequencies, and perioperative organ frequencies and activity levels. In particular, we examined the effects of anesthesia on the myoelectrical activity before surgery, during the surgical intervention, and in the postoperative period following anesthesia.

Material and Methods

Patch technology

The myoelectric activity in the GI tract refers to the electrical current generated by the smooth muscle cells (SMCs). These signals can be detected on the skin surface similar to how the electrocardiogram detects heart signals. The G-Tech system employs disposable wireless patches applied to the skin to detect this activity. Each patch (manufactured by G-Tech Medical, Fogarty Innovation, Mountain View, CA) contains four pairs of bipolar electrodes arranged in two horizontal and two vertical pairs. This configuration allows for redundancy checks from the parallel pairs and greater spatial coverage by the perpendicular pairs. The patches continuously record the myoelectric activity of the GI tract and transmit the data to a paired iPhone with an iOS operating system (Apple, Cupertino, CA). Each external patch has the potential to detect signals from multiple organs. A designated application on the iPhone collects the data, encrypts it, and relays it to a secure cloud server, where it is processed to represent organ-specific waves across the GI tract [11,12].

Animal care and surgery

This experiment involving animal care and surgical procedures was approved by the Stanford Administrative Panel on Laboratory Animal Care under protocol number 33264. Seven female, juvenile, miniature Yucatan pigs (Sus Scrofa, S&S Farms, Ramona, CA) were used for these experiments. Male pigs were avoided since midline incision might interrupt their urethral anatomy [13]. Four external skin patches were placed on four quadrants of the abdominal wall with similar orientation and labeling in all animals three to five days prior to anesthesia, with three scheduled meals per day. Pigs were kept in a pen that allowed unhindered movement. Fabric sleeves were worn to help protect the patches and reduce damage to the electronics. On the day of the procedure, animals underwent induction of general anesthesia with injected ketamine and xylazine and at least 1.5% inhaled isoflurane. Pigs were observed for 1 to 2 hours after induction of anesthesia to record the myoelectric activity before surgical manipulation. Subsequently, the patches were removed, and a midline incision was made to insert intraabdominal strip internal electrodes (AD-Tech, Oak Creek, WI). The electrodes used consisted of a linear strip of eight individual sensing elements spaced 1 cm apart. For this test, they were wired as 4 sequential bipolar pairs with the first two electrodes forming channel a, the next two channel b, etc. Thus, the pairs formed a sensing element with a 1 cm separation, spaced 2 cm apart from center to center. Three 8-element commercial linear arrays of electrodes were sutured to the serosa of the stomach, the proximal jejunum, and the proximal spiral colon as described in the previous study [11]. The wires were exteriorized through a new lateral incision. Both incisions were closed. The digitized signals were collected by a wireless data acquisition receptor defined by G-Tech's technology. Four new external patches were placed in a similar manner to record the signals through the skin and transfer the data to the same data acquisition system. The pigs were fed solid food approximately three hours post-operatively when they were fully awake. Data were collected simultaneously from both systems for the next six days. The pigs were euthanized on postoperative day 6.

Data Analysis

A customized program developed in LabVIEW version 19.0 was utilized to analyze the raw data collected by internal electrodes and external patches as described in the previous study [11]. In summary, the data were processed by removing large amplitude artifacts, and Fourier transformation was performed to create frequency spectra in 10-minute intervals for each internal and external channel. Peak detection algorithms were applied to these spectra to identify rhythmic activity episodes. The G-Tech algorithm was used to distinguish between true peaks and noise by identifying the frequencies, peak width, height, area above the background, and quality parameters from clean peaks in the spectra. The G-Tech algorithm differentiates the source of signals by spectral peak characteristics, primarily frequency and peak width. Dot plots of frequencies versus time (hereafter referred to as “peak frequency plots” or “peak plots” for brevity) were created based on the center frequencies of peaks detected in the 10-minute spectra, as well as histograms of these peak frequencies weighted by their area (“peak frequency histograms”, or “peak histograms”). For external patches, the signals were attenuated while traveling to the surface, which was compensated for by G-Tech's patented algorithms when calculating peak volumes (peak area times duration). To evaluate the concordance between the external and internal measurements in terms of frequency and time, the peak frequencies detected by the electrodes located on each organ were compared with those detected by the external patches by visual examination of the plots and by a second custom LabVIEW program which calculated the correlation by tallying up matches between internal and external peak frequencies within each time segment.

Results

Subject demographics

Seven pigs were included in the study. The age was approximately seven to nine weeks. Weight ranged from 7.7 kg to 14 kg, with an average ± standard deviation of 10.3 kg ± 2.1 kg. All survived the procedure and did not have any postoperative complications. They tolerated solid diet and had normal bowel movements.

Internal electrode and external patch measurements post-surgery

Both internal electrodes and the external patches acquired continuous time series data at approximately 5 Hz. Representative recordings of internal electrodes and external patches showed the nominal slow waves of each organ for the same two-minute period (Figure 1). The rhythmic activity was transformed into peaks in the frequency spectra over the course of the experiment in all pigs (Figure 2). Higher frequency “spike bursts” were apparent from the internal electrode recordings of the stomach and small intestine.

Figure 1.

Simultaneous two-minute time series data from internal electrodes; A: stomach; B: small intestine; C: colon; and D: one channel from an external patch, all showing rhythmic variation.

Figure 2.

Peak frequency plots as labeled for (left to right) stomach, small intestine, and colon from the internal electrodes, for (top to bottom) Pig 1 through Pig 7. Empty plots reflected failed electrodes or connections. The horizontal axes span six days in all cases to facilitate visual comparison.

Each internal strip of eight electrodes was paired to produce four separate signals spaced 2 cm from each other. The stomach and small intestine time series data had both low and high frequency components while the colon recordings had mostly low frequency (Figure 1). In the stomach, a progression from one electrode pair to the next was apparent in Figure 1A. There was a time difference of approximately 9 to 10 seconds from the first electrode pair to the fourth electrode pair, implying a signal propagation speed of about 0.65 cm/sec in the stomach.

Frequency peaks detected by internal electrodes for each organ are shown for each pig after surgery (Figure 2) Peak frequency histograms averaged over all pigs, and the same for individual pigs, are shown in Figure 3. For the stomach, the higher frequencies were multiples of the lowest frequency, representing harmonic content. The colon exhibited a similar frequency as the stomach, but the harmonic content was completely or nearly absent. In the small intestine, there were frequencies similar to those of the stomach, plus additional frequencies in the 14 to 18 CPM range. Variation over time in both frequency and amplitude was evident, particularly in the first hours after surgery. Across the seven pigs, there were strong similarities in the frequencies in each organ, with variability in their relative strengths and time evolution. There was a strong overlap in frequency in the 4 to 5 CPM range for all three organs. Similarities in the trajectory of peaks traced out between the stomach and small intestine at 4 to 5 CPM and its first harmonic around 9 CPM can be seen in the frequency plots of Figure 2. This may be evidence of electrical crosstalk between the organs measured by the electrodes (Appendix).

Figure 3.

Peak histograms averaged across pigs for the A: stomach; B: small intestine; and C: colon from the internal electrodes. Individual peak histograms for all pigs in the D: stomach; E: small intestine; and F: colon. To account for differences in overall signal strength between pigs, the individual histograms are first scaled to equal areas.

External patch measurements – pre-surgery baseline and post-surgery

The frequency plots and peak histograms for the external patch measurements before surgery, external patch measurements after surgery, and the sum of the post-surgery internal electrode frequencies are shown for all 7 pigs (Figures 4,5). Note that for the first few pigs, the duration of baseline prior to surgery was shorter. Overall, the baseline and post-surgery frequency plots from external patch measurements were similar, with higher activity after surgery than before surgery in most cases. The internal electrode frequency plots showed more details and had a better signal-to-noise ratio, evidenced by a cleaner appearance, with a higher proportion of the peak frequencies concentrated in connected continuous patterns, or “peak trajectories.” The dominant frequencies in those peak trajectories appeared also in the plots of external patch data and correlated well with internal measurements.

Figure 4.

Peak frequency plots as labeled for (left to right) pre-surgery patch, post-surgery patch, and the sum of post-surgery internal electrodes, for (top to bottom) Pig 1 through Pig 7. Empty plots reflected failed electrodes or connections. The horizontal axes span six days in all cases to facilitate visual comparison.

Figure 5.

Peak histograms averaged across pigs for the A: pre-surgery patch; B: post-surgery patch; and C: sum of post-surgery internal electrodes. Individual peak histograms for all pigs in the D: pre-surgery patch; E: post-surgery patch; and F: sum of post-surgery internal electrodes. To account for differences in overall signal strength between pigs, the individual histograms are first scaled to equal areas.

Comparison of external patch to internal electrode measurements

The frequencies that emerged as the primary frequency for each pig, with averages and standard deviations across pigs were compared between internal electrode and external patch measurements (Table 1). The primary peaks as directly measured by the internal electrodes for the stomach and colon were at 4.5 ± 0.23 and 4.5 ± 0.37 CPM, respectively, while for the small intestine, it was 16 ±1.2 CPM. For the external patch measurements, the pre-surgery baseline was 4.5 ± 0.32 CPM with post-surgery at 4.4 ± 0.26 CPM. Variation from pig to pig in the primary frequency tracked closely between baseline and post-surgery, with an average difference pre- to post-surgery of 0.9 CPM, and these also tracked with the internal stomach frequency. Similarities in primary frequency as measured at each organ across animals suggest a possibility of crosstalk between organs (Appendix).

Table 1:

Primary and secondary rhythmic frequencies from each pig as measured by internal electrodes and external patches.

Pig number	Stomach	Small intestine	Colon	External, pre-surgery	External, post-surgery
Pig 1	4.6	16.7	4.7	4.8	4.7
Pig 2	4.1	14	4.1	4.1	4.1
Pig 3	n/a	16	4.7	4.7	4.3
Pig 4	4.8	15.5	4.8	4.8	4.8
Pig 5	4.3	15.8	4.7	4.3	4.3
Pig 6	4.3	n/a	4.3	4.1	4.2
Pig 7	4.9	18.1	4.8	4.9	4.7
Primary peak	Avg=4.5 (SD=0.23)	Avg=16 (SD=1.24)	Avg=4.5 (SD=0.37)	Avg=4.5 (SD=0.32)	Avg=4.4 (SD=0.26)
Other peaks	Avg=9.4	Avg=4.4 Avg=9.3	Avg=6.6 Avg=14	Approximately 6, 8, 13 to 15	Approximately 6, 8, 13 to 15, 18

Comparing peak frequencies between internal and external measurements can establish a valuable measure of concordance, but that approach uses average values over the full test, and when there is an overlap in the frequencies between organs, the results can be harder to interpret. A stronger analysis involves comparing the frequency peaks detected in each 10-minute time segment. The fraction of peaks detected by the external patch that was also detected by internal electrodes in the time interval was analyzed. Since the time segments for each of the readouts were not time-aligned, adjacent 10-minute segments were included in the analysis, with a precision range of 1 CPM. The analysis was conducted for all 7 pigs for the full post-operative period of 6 days.

Figure 6 shows an illustrative example of the analysis for a three-day subset chosen for clarity. The peaks detected by the internal electrodes are plotted as blue crosses and those detected by the external patches as red circles to show overlap (Figure 6A). The frequencies where there was agreement between internal and external measurements are shown in Figure 6B. In this example, there were 2,075 peaks detected by the internal electrodes, of which 1,722 (83%) were also detected by the external patches. Such peak detection analysis was performed for each pig over six days post-surgery (Table 2). The average fraction of the internal electrode-detected peaks that were simultaneously detected by the external patch was 70%, with a range of 55% to 81%. This peak-by-peak analysis is the functional equivalent of comparing time series data in small periods but applied to multiple days on multiple channels to provide comprehensive information on the agreement between recordings.

Figure 6.

Peak volumes from external patches averaged across all pigs during the pre-surgery time period. All data sets were aligned at the beginning of surgery (at t=0 for surgery start).

Table 2:

Measures of concordance between external patch and internal electrode peaks

Pig number	Patch peak matches Internal peak	Internal peak matches Patch peak
Pig 1	0.796	0.561
Pig 2	0.730	0.673
Pig 3 *	0.813	0.566
Pig 4	0.658	0.625
Pig 5	0.651	0.651
Pig 6 **	0.721	0.590
Pig 7	0.553	0.566

^*internal stomach electrode failed

^**internal small intestine electrode failed

Variations in absolute signal strength due to animal girth, skin condition, and similar effects not related to actual motor activity levels are compensated for during data processing. This applies also to the internal vs. external recordings. Overall, the internal recordings have approximately five times higher signal voltages than external, in this configuration of electrode pairings. In the internal recordings, the measured voltages are in an approximate ratio of 6 to 3 to 1 for the stomach, small intestine, and colon. We emphasize that the measured voltages are from bipolar electrode pairs, spaced 1 cm apart for the internal electrodes and just under 5 cm apart for the external patches. Different spacings are likely to produce different signal strengths.

Effect of anesthesia prior to surgery

Pigs were anesthetized for either 1 or 2 hours prior to the start of surgery to implant internal electrodes. Pre-surgery patch data from each pig were normalized to 1 to give each an equal weight for averaging and time-aligned at the beginning of surgery. The overall myoelectrical activity averaged across pigs, exhibited a periodicity over the last 48 hours likely attributable to a diurnal effect, and fluctuations in the trend (Figure 7). After removing the trend, as isolated using a Savitsky-Golay filter, the residual hour-to-hour variance is 21%. Activity in the last two hours, with anesthesia in effect, is 41% lower than the average activity for the pre-surgery period and 55% lower than the prior 12-hour period, consistent with a suppression of activity by anesthesia. However, the last two hours also coincide with a low activity period in the ostensible diurnal pattern, which could alternatively account for some or all of the suppression.

Figure 7.

A: and B: peak volumes from internal electrodes averaged across all pigs; C: and D: the same from external patches. E, F, and G: average peak volumes from internal electrodes for the stomach, small intestine, and colon.

Recovery of myoelectric activity following surgery

In the hours immediately following surgery, myoelectrical activity was substantially suppressed (Figure 8). The summed activity increased quickly and reached a maximum at about 12 hours after surgery in the external patch measurements. This increase was more substantial in the external patch measurements. For the internal electrode measurements, a rapid increase was observed during the first day, followed by a daily maximal activity in the next 5 days. The stomach typically recovered most quickly, while the colon’s recovery time was the longest. The slower recovery of the external patch recordings may be due to sampling more from the colon than the other organs. The myoelectrical activity from internal colonic electrodes and external patch measurements tracked one another closely, with maxima at 11 and 40 hours.

Figure 8.

Peak volumes from external patches averaged across all pigs in the postoperative course, compared to the baseline average (orange line).

Comparison of post-surgery to pre-surgery baseline myoelectric activity

Post-surgery external patch measurements showed an increase compared to the overall baseline prior to anesthesia and surgery (Figure 9). As with the baseline data treatment, combining the post-surgery recordings across pigs, each data set was scaled to have an average value of 1 to normalize the results from each pig. Since equal weighting of pig data required scaling, calculating the ratio of post- to pre-surgery activity was done on a pig-by-pig basis. The ratio of post-operative to baseline data was calculated for each pig. Three of the individual ratios were excluded from the calculation because their baseline data were incomplete.

Figure 9.

Peak histograms from external patches averaged across all pigs for A: before and B: after surgery.

The average ratio of post- to pre-surgery activity was 1.69 with a standard deviation of 0.30. The average post-surgery activity initially increased significantly and decreased to baseline over the six-day recovery period (Figure 9). Another way of looking at this rebound effect is through the averaged (but not scaled) peak histograms from before and after surgery (Figure 10). This view of the data was averaged over the six-day period and provided insight into the frequency distribution. Post-surgery, all frequencies were increased. The ratio of areas of the two histograms was 1.82, consistent with the ratio of 1.69 above. To our knowledge, this is the first demonstration of such a rebound phenomenon in the postoperative course.

Figure 10.

A: Peak frequency plots (three-day subset) for internal electrodes (blue crosses) and external patches (red circles) superimposed. B: frequency plot of all peaks for which the external patches matched the internal peak frequency.

Discussion

We report the myoelectric activity of the porcine gastrointestinal tract in the perioperative course from 3 to 5 days before, to 6 days after abdominal surgery. We compared internal and external measurements after the surgery to validate external measurements. In the awake pig, we found frequency peaks in several ranges, from 4 to 6.5 CPM, 8 to 11 CPM, and 14 to 18 CPM which were comparable between subjects and in agreement between internal and external recordings.

The primary aim of this study was to measure the rhythmic characteristics of the digestive tract of a sample population of mini pigs as a proxy for humans in neurostimulation studies, and to assess the potential value of non-invasive cutaneous myoelectric measurements as a monitor of that activity in pig, and eventually, in human studies. The ability of the external system to detect a representative sample of the internally measured signals is of primary importance. Through comparison of the primary frequencies observed over the full duration of the tests via peak histograms, and by peak-by-peak comparisons in close frequency and time proximity, a strong concordance has been established. An additional aim was to discern in which organ a signal originates, although for testing and tuning neurostimulation known as targeting a particular organ, this may not be absolutely necessary.

Possible effects of anesthesia during the 1 or 2 hours before surgical manipulation was observed as a 59% decrease in overall myoelectric activity compared to the immediately preceding time period. It is not possible to confirm that the observed decrease is solely due to anesthesia due to the natural hour-to-hour variation in activity as reflected by the 36% standard deviation. Also, animal handling, transport, and preparation may have induced stress that would confound an attempt to see the effect. The 1- or 2-hour time period, chosen for practical reasons, does not afford enough time to allow for a definitive answer. A longer time period under anesthesia would allow us to determine if the drop is systematic and consistent and would provide additional time for the anesthesia to affect motor activity.

Postoperatively, there is a consistent effect across all pigs of initial minimal activity with recovery in approximately 4 hours for the stomach, 9 hours for the small intestine, and 11 hours for the colon. This pattern mirrors what is seen in human patients postoperatively in terms of the return of function as well as the order of recovery of the organs but at a faster rate [16, 17]. In the first hours after surgery, residual anesthesia may play a role in the reduced activity, with organ response to the surgery playing a larger role later. The surgery in this case consisted of sewing strip electrodes to the serosa of each organ and was less of an insult than would be the case in typical gastrointestinal interventions in humans.

Comparing the absolute post-surgery activity levels to the baseline for each pig reveals higher overall activity postoperatively by a factor of 1.69, averaged over the six-day recovery period. The comparison is complicated experimentally by a change in patches after surgery, possibly slightly different patch locations, variable skin conditions, and the automatic data normalization process, all of which must be considered as possible sources of systematic error. However, the trend of an initial higher ratio to one approaching unity at the end is encouraging.

Cutaneous patches have been used in several human studies to correlate activity with the return of bowel function [8-10]. For the first time to our knowledge, a postoperative rebound phenomenon of overall GI motility was recorded here.

Additional information regarding organ-specific internal measurements and their comparison with external readings can be found in our previous report as well [11]. The human GI tract has distinct frequencies of myoelectric activity associated with each part: stomach at 3 CPM, duodenum at 12 CPM, ileum at 6 to 8 CPM, and colon ranging from 0 to 40 CPM [2,3]. In the pig, there is a strong overlap between the organs as measured by the internal electrodes in this study, particularly in the region of 4 to 5 CPM. Hence data from external patches are unable to specify the specific organ as readily as in humans. However, the patterns of multiple frequencies seen in the peak spectra provide additional information. Specifically, the colon has a single dominant frequency at about 5 to 6 CPM while the stomach has a 4.5 to 5 CPM peak plus its harmonic multiples, and the small intestine has its strongest peak centered around 15 to 18 CPM, distinguishing it from the stomach. In the Appendix, we consider the effects of crosstalk between organs and propose an alternative frequency assignment scenario using the peak-by-peak concordance algorithm used to compare external with internal measurements.

Isoflurane is an inhaled anesthetic agent, commonly used in the maintenance of general anesthesia. It belongs to a class of halogenated ethers; these gasses are thought to enhance the release and action of gamma-aminobutyric acid (GABA) in the brain, an inhibitory neurotransmitter [18,19]. In animal models, it has been shown to slow GI motility. A pig model utilizing orogastric-instilled radionuclide-tagged nutritive or non-nutritive liquid demonstrated high-dose isoflurane (~2%) delayed gastric emptying [20]; this rate of anesthesia delivery is comparable to our subjects. Similarly, in rodents, brief exposure to isoflurane resulted in a 50% reduction in the propulsion of ingested charcoal [21], and fasted rats showed a decrease in myoelectric complex spiking derived from electrodes implanted directly into the muscularis of the intestines during ether anesthesia [22]. Despite these motility changes, regional gastrointestinal blood flow appears to remain constant during surgery in a pig model [23]. We corroborate these findings of slowed GI activity with cutaneous patches during anesthesia demonstrating a relative decrease in myoelectric signal after cumulative anesthesia as compared to pre-anesthesia.

The goal of utilizing EMG-sensing cutaneous patches is to predict the recovery of GI function after an abdominal operation. The external patch measurements were well correlated with internal measurements visually and statistically. Organ-by-organ measurements postoperatively, demonstrated that the stomach has the fastest and the colon the slowest recovery. We also observed a rebound phenomenon in peak volumes of GI motility in the postoperative course.

Our data on GI motility measurement in the postoperative course of pediatric patients showed that, in heterogeneous pediatric surgical patients, there was an increase in activity from immediately post-operative to first bowel movement, which supports a role for myoelectric activity in the determination of return of physiologic function versus ileus [9]. Being able to adequately predict those patients at higher risk for postoperative ileus may modify patient management strategies and lead to a better understanding of the underlying pathophysiology [8,10].

One of the limitations of our study is the short duration of anesthesia time before surgical intervention which does not afford enough time to allow for a definitive result. Additionally, it was challenging to apply patches on pigs' skin and keep them on for about 6 days since these patches are designed for human use. The primary limitation of the GutTracker system is that it measures only the signals that are detectable at the skin surface, which is reduced in amplitude from those that can be measured internally. The location of the signal origin is compromised compared to an internal measurement due to the spatial separation and the fact that the signals disperse as they propagate. Mitigating these effects is the fact that the external measurements detect the internal signals on average 70% of the time as we show, and that there is internal electrical crosstalk between organs that makes signals from one organ appear on the electrodes of a different organ as we also show.

Using only female pigs to avoid potential urethral injury was another limitation of the study. Although employing only one sex would reduce the generalizability of the results and eliminate the possibility of evaluating sex-based differences in ileus and GI myoelectric activity, the absence of one gender had minimal impact on the primary goals of our study which were detecting the concordance of external and internal signals, reproducing organ-specific frequencies from pig to pig, and defining the impact of anesthesia on GI myoelectric activity.

Conclusion

In this study, the perioperative myoelectric activity of the GI system was measured with external patches and internal electrodes sutured to the serosa of the stomach, small intestine, and colon, for 3 to 5 days prior to surgery and 6 days afterward. Unique to this study to the best of our knowledge, during the 6 days of recovery the animals were fully ambulatory within pens. The time series data was processed in ten-minute time segments to search for and quantify the frequency and area of peaks in FFT spectra. Histograms of the peak frequencies detected showed strong agreement between internal and external measurements. A peak-by-peak, time segment-by-segment statistical analysis showed that on average 70% of the internally measured peaks were also detected externally in the same time segment. This result provides confidence that an external patch system could be useful in determining the return of digestive organ function following abdominal surgery.

External patch measurements show a strong concordance with internal electrodes placed on the serosa of the stomach, small intestine, and colon of pigs, allowing for monitoring of their motor activity non-invasively. Recovery of the organs from surgery and anesthesia was recorded consistently by both systems. The stomach recovered the fastest while the colon was the slowest. Overall, postoperative GI myoelectric activity after initial recovery was higher than in the preoperative course, demonstrating a rebound phenomenon in the first few postoperative days. Anesthesia was applied either 1 or 2 hours before surgery to study its effect on motility. A decrease in the last two hours of approximately 40 to 55% was observed, but the short duration of the anesthesia and an overall diurnal trend make the attribution of the decrease in activity to anesthesia less than conclusive. Increasing knowledge of GI motility during the perioperative course would facilitate the prediction, diagnosis, and early treatment of postoperative ileus. Similarly, non-invasive measurements of the myoelectric activity of the full GI tract over multiple days in an ambulatory setting may provide new insights into GI motility disorders at the population and individual levels.

Supplementary Material

2

Figure A1. A: Frequency plot comparison of Internal small intestine (blue crosses) and internal stomach (red circles) electrodes superimposed. B: frequency plot of those peaks for which the small intestine and stomach have the same frequency at the same time, zoomed in to the 3 to 12 CPM range. C: frequency plot as in A but comparing colon electrode (blue crosses) to stomach (red circles).

Highlights:

• Both external patch measurements and internal recordings concordantly show changes in the gastrointestinal myoelectric activity in the perioperative course.

• Overall, postoperative GI myoelectric activity after initial recovery was higher than preoperative course, demonstrating a rebound phenomenon in the first few postoperative days.

• Increasing knowledge on gastrointestinal motility during the perioperative course may facilitate prediction, diagnosis, and early treatment of postoperative ileus.

APPENDIX

Internal electrical crosstalk between organ electrodes

Just as the myoelectric signals are able to travel from organs to skin surface and are detectable there, we might reasonably assume they likewise travel to and are detectable at other locations inside the body, specifically at the other digestive organs. The similarity in time evolution of the peak frequencies between stomach and small intestine seen in Figure 2 suggests crosstalk may be a source of the 4-5 CPM peaks (and its harmonics at two and three times this) in the small intestine. The peak by peak, time segment by segment concordance analysis provides a test that can shed a brighter light on whether there is significant crosstalk between the organs. Figure A 1A shows peaks from the small intestine (blue) and stomach (red) plotted on top of one another for a representative time and frequency range for clarity (data from Pig 1). In the 5 and 10 CPM trajectories it is hard to distinguish one data series from the other, but at the higher frequencies, the blue small intestine peaks stand on their own. In Figure A 1B, the plotted points represent only those peaks which were found to be in agreement in time and frequency, showing that the small intestine peaks are in common with the stomach. The peak-to-peak agreement between stomach and small intestine in the 4 to 6 CPM range and 8 to 11 CPM range is 93% and 89% respectively. Therefore, the peaks we measure in the small intestine electrode are likely to be crosstalk from the stronger stomach myoelectric activity. Only in the 11 to 18 CPM range (24% agreement) are the true peaks from the small intestine to be found. By contrast in Figure A 1C, we compare colon peaks in blue with the stomach peaks, where we see far less overlap with the exception of some at 5 CPM in the first 12 hours after surgery. In fact, the stomach signals were relatively strong during that time, more than twice as strong as afterward, which may explain why they are seen in the colon electrode data. Table A1 shows the proposed nominal Yucatan mini-pig rhythmic frequency ranges we infer from our analysis. A frequency range is assigned to the small intestine or colon whenever it appears only on that internal electrode at a given time and cannot be assigned to crosstalk from the stomach. These values are consistent with what can be seen in the peak histograms of Figure 3, where in the case of the colon, the small shoulder peak at about 6 CPM and isolated bump at 14 can be seen for what they represent.

Table A.1:

Proposed nominal rhythmic frequencies of pig organs after elimination of crosstalk effects.

Organ	Proposed nominal frequencies
Stomach	4 to 5 CPM and multiples
Small intestine	14 to 17 CPM
Colon	5 to 6.5 CPM primary, 14 CPM secondary