Validation of antroduodenal motility measurements made by echo-planar magnetic resonance imaging

Authors


Wright Department of Surgery, University Hospital, Nottingham NG7 2UH, UK. Tel: 44 1159709249; fax: 44 1159709428.

Abstract

Echo planar imaging, a development of magnetic resonance imaging, can produce snapshot images of the stomach and antroduodenal segment in as little as 64 msec and can be more useful than conventional techniques when assessing motility. The aim of this study was to compare antroduodenal motility measured by simultaneous perfused tube manometry and echo planar imaging. Ten volunteers were studied following the ingestion of 500 mL water or 500 mL porridge. Antroduodenal images, with acquisition times of 130 msec, were taken at 3-sec intervals, synchronized with motility traces and presented as a split-screen video. This allowed direct visual comparison of gastric wall movement and motility to be made. Contractions were confined to either the stomach or the duodenum or propagated across the antroduodenal segment. Over 4550 images were available for analysis. A larger number of propagated contractions were recorded with echo planar imaging in both water (P = 0.03) and food (P = 0.02) groups, whereas manometry detected a greater number of isolated duodenal pressure waves (P = 0.005). The contraction rate for water and food studies was similar, but direct visualization indicated that the manometric technique under-detected propagated events. The ability of echo planar imaging to record antroduodenal contractile activity provides a new insight into the role of occlusive and nonocclusive contractions during gastric emptying.

INTRODUCTION

Magnetic resonance imaging (MRI) has been available for many years and is a widely used technique in medical diagnostics, producing high-definition static images of internal organs without the use of X-rays or ionizing radiation. MRI has replaced conventional radiographic and scintigraphic techniques in many medical imaging procedures. Conventional MRI, however, typically requires at least tens of seconds of imaging time to build up an acceptable picture. To date, this has meant that MRI has only been suitable for stationery organs or organs undergoing periodic motion.

Echo planar imaging (EPI)1 is a relatively new magnetic resonance technique whereby a complete picture can be produced in a fraction of a second, thus enabling pictures of moving organs to be produced. We have previously reported the use of EPI in the evaluation of gastrointestinal motility for short time sequences2 and more recently with longer imaging periods over many hours.3

To date, however, MR imaging of the gastrointestinal tract has not been validated by comparison with conventional techniques for the measurement of gastrointestinal motility. The aim of this study was to compare dynamic images produced by EPI simultaneously with perfused tube manometry and, in addition, we sought to compare the accuracy of EPI and manometry when differentiating between propagated antroduodenal contractions and those which were confined to either the antrum or the duodenum alone.

MATERIALS AND METHODS

Subjects

Ten healthy male volunteers, with a median age of 24 years (range 18–46 years), and with no history of gastrointestinal disorders, were recruited for combined EPI and antroduodenal manometry. Subjects were studied fasted and were then fed during the course of the experiments. Subjects were all of slim build, with a median body weight of 71 kg (range 64–77 kg), thereby maximizing manoeuvrability within the magnet. All completed standard MRI safety checks. The protocol for the study had previously been approved by the Nottingham University Medical School Ethics Committee.

Echo planar imaging

Dynamic images of the antroduodenum were obtained using a 0.5-Tesla, purpose-built, echo planar magnetic resonance imaging system. Ultrafast magnetic resonance imaging was performed using a modulus blipped echo-planar single-pulse technique (MBEST)4 which produced a two-dimensional transverse image in an acquisition time of 130 msec for a 128 × 128-pixel matrix image, with a slice thickness of 1 cm. In order to assess gastric wall movement each set of data consisted of 128 images acquired at 3-sec intervals. During imaging, subjects were required to time their respiration so that it was gated to the scan frequency, thus avoiding changes in the imaging plane during the respiratory cycle.

Perfused tube manometry

Conventional manometry was recorded simultaneously with echo planar imaging using a seven-channel water perfused tube assembly. The catheter configuration is shown in Fig. 1 and consisted of six pressure channels, the proximal four channels being positioned at 1-cm intervals across the antropyloric segment and the distal channels at 5 and 7 cm from the pylorus. A small latex balloon was attached to the catheter at the level of the mid antropyloric ports and was inflated with gadolinium to confirm the catheter position during imaging. Two pH channels, one proximal to the antral measuring ports and one distal to the pyloric measuring ports, attached to portable dataloggers, were incorporated in the catheter assembly. Gastric and duodenal pH values were used to further confirm the position of the antropyloric pressure units during imaging and could be synchronized with the manometric data by use of the time stamp recorded with each pH value.

Figure 1.

Perfused tube assembly: the relative positions of the manometry side-holes are indicated.

The catheter assembly was connected to a polygraph chart recorder via in-line pressure transducers. A low-compliance pneumohydraulic pump, driven by oxygen-free nitrogen at 15 psi was used to continually perfuse the catheter with distilled water at a delivery rate of 0.2 mL min−1. The overall frequency response of the system measured at the catheter was better than 300 cmH2O sec−1. Manometry data were simultaneously recorded onto a seven channel FM tape recorder, at a tape speed of 15/16 in sec−1 with a band width of DC to 313 Hz, for subsequent analysis. The pump was sited adjacent to the volunteer, in order to maintain the compliance of the manometric system, whereas the recording system and associated hardware was sited at least 3 m away from the scanner to prevent any corruption of the data stored on magnetic media. The internal catheter components were required to be constructed of nonferrous material in order to avoid image degradation during scanning.

Study protocols

Volunteers reported to the laboratory on the morning of the study, after an overnight fast, and were intubated with the catheter assembly via the nose into the stomach. The catheter was allowed to pass by normal motility such that the distal end of the tube was transported across the pylorus into the duodenal bulb. This was confirmed by a sustained pH change in the distal channel to a high value (> pH 5), with an acid value (< pH 3) being maintained in the proximal pH channel. Pressure recordings from the six pressure ports confirmed that the distal two ports were sited in the duodenum.

The subject entered the MRI scanner and was positioned such that the antropyloroduodenal slice was visible. To confirm the position of the catheter within the empty stomach, the latex balloon was inflated and deflated with 5–10 mL of gadolinium whilst scanning was in process. The bright image caused by the gadolinium in the balloon could be seen to appear and disappear on inflation and deflation and further fine tuning of the catheter was made until it was assessed to be in the correct position ( Fig. 2). The catheter was anchored in position by taping to the cheek and combined recordings were commenced. The position of the catheter was checked at intervals throughout the study, using pH, waveform and balloon inflation. The period of balloon inflation was kept to a minimum to avoid affecting the motility. In addition, by asking the volunteer to alter their position within the scanner from supine to 45° left lateral tilt, it was possible to optimally visualize the position of the tube and also the gravitational effects on the water level ( Fig. 3).

Figure 2.

EPI images before (upper) and after (lower) balloon inflation with gadolinium. The position of the balloon can be determined by the appearance of a bright spot (arrowed) upon inflation.

Figure 3.

The position of the manometry catheter can be more easily detected when the volunteer rotates 45° in the scanner. The horizontal water level in the antrum can be clearly seen.

Water imaging

The subjects were given an initial oral dose of 500 mL of tap water, which acted as a gastric contrast agent. Simultaneous supine imaging and manometry was then performed and datasets of 128 EPI images were collected as described previously. Imaging continued in this way at intervals until the water had emptied from the stomach. Subjects remained in the scanner and were topped up with further volumes of tap water, via a flexible tube, and the scanning procedure was continued.

Food imaging

Following water imaging, subjects were brought out of the scanner and rested for a short period before being given a standardized meal consisting of 500 mL of porridge made with water. Subjects were again placed in the scanner and motility imaging with simultaneous conventional tube manometry was continued for up to 3 h.

Interpretation of data

Echo planar images were reconstructed and displayed as `movie loops' and then recorded on a video tape at speeds of four and eight times real time. EPI images were synchronized to the tape-recorded manometry traces by using the electronic trigger pulse produced by the scanner electronics exactly at the start of image acquisition. This was stored in an additional channel on the taped manometry recording. It was possible, by video editing, to produce split-screen images of the EPI and reconstructed manometry data ( Fig. 4), thus enabling a direct visual comparison to be made between the total number of contractions recorded by either EPI or by conventional manometry. An arbitrary fixed point on the EP image denoted the start of each contraction and the number and frequency of contractions passing that point was counted by two of the authors independently. In cases of disagreement, contractions were assessed jointly by a panel of three authors.

Figure 4.

Single frame from a split- screen video showing synchronized EPI and manometry data. The uppermost manometric channel corresponds to proximal antral activity.

Contractions were identified as follows.

EPI data

(i) Antroduodenal contractions – arising within the antral segment and progressing across the pylorus and seen as further peristaltic contractions within the duodenum; (ii) antral contractions – arising in the antrum but not propagated across the pyloric segment; (iii) duodenal contractions – seen only in the duodenum, progressing caudally but not associated with any antral contractility.

Manometric data

All pressure waves with an amplitude threshold greater than 10 cmH2O and a duration of 15–25 sec in the antrum and 4–10 sec in the duodenum were analysed. Wave velocities of 0.62 cm sec−1 (range 0.37–0.87, antrum) and 1.94 cm sec−1 (range 1.29–2.59, duodenum) have been reported by others.5, 6 These data have been utilized, and a propagation event was assumed to occur if a pressure wave appeared in at least two channels within this predetermined time window.

Statistical analysis

Results were analysed on a personal computer using the Statistical Package for Social Sciences (SPSS Ltd, Surrey, UK). Bland and Altman plots7 were used to calculate the agreement between the two methods. The bias, or mean difference between the techniques, showed how closely the methods agreed while the 95% confidence intervals and two standard deviations showed the precision and limits of the agreement. The null hypothesis that there should be no difference between the number of contractions recorded by EPI and manometry was tested using the χ2 analysis. A probability value of < 0.05 was accepted as statistically significant.

RESULTS

Results from seven volunteers were available for analysis. The remainder were lost primarily because of poor manometer positioning (migration of the catheter back into the stomach). Those pressure waves which occurred when the catheter assembly moved in an aborad direction (as detected by pH change and waveform) were ignored on analysis. Gastric wall movement, indicative of both occlusive and nonocclusive contractions, was clearly visible with EPI ( Fig. 5). Split-screen videos, comprising synchronized EPI images and motility traces, were analysed by two observers to assess the origin and percentage propagation of contractions.

Figure 5.

EPI images acquired at 6-sec intervals following a liquid meal. Water, seen as a bright image, fills the stomach and delineates the antrum (a), and the duodenal bulb (d). The pylorus (p) can be identified as the wave of contraction passes from the antrum into the duodenum.

Results from over 4550 images are presented in Fig. 6 and demonstrate the differences in detection of antroduodenal contractions between EPI and perfused tube manometry. Manometry detected a larger proportion of isolated duodenal events than EPI in the water and food groups (3:1, = 0.005). In the water studies, 83% of EPI detected contractions were propagated as opposed to 57% measured by the manometry and similarly in the food studies 92% of EPI contractions were propagated compared with 68% in the manometry group, the differences being statistically significant (χ2 = 4.74 = 0.030 and χ2 = 5.13 = 0.024, respectively).

Figure 6.

Results from over 4550 images demonstrate the differences in contractions detected by EPI and manometry during fasting and fed studies.

There was no significant difference in the rate of contractions for either group between the water and food studies. The Bland and Altman plots demonstrated an excellent agreement between methods for total contraction numbers, with a bias of −1.2 over a range of values 3–30 ( Fig. 7A). The ratio of percentage propagation between manometry and EPI was similar in both cases (water, 0.69 and food, 0.74), with ≈ 3 contractions detected by EPI for every 2 pressure waves detected by manometry, shown by a negative bias ( Fig. 7B). The ratio of nonpropagated events detected by manometry and EPI was ≈ 3:1, resulting in a positive bias ( Fig. 7C).

Figure 7.

Bland and Altman plots showing a small negative bias (A) indicating excellent agreement between methods when assessing total contraction numbers. The negative bias of plot (B) indicates that a greater proportion of propagated contractions were detected by EPI than manometry. The positive bias (C) indicates that a greater number of nonpropagated events were detected by manometry than EPI.

DISCUSSION

Magnetic resonance imaging is now an established diagnostic tool in clinical medicine, but the variation in image acquisition times, from several seconds to several minutes, has made the technique highly susceptible to motion artefact and unreliable for motility measurement. The development of echo planar imaging has enabled snapshot images of the gastrointestinal tract to be collected in as little as 130 msec, thereby allowing gastric function to be imaged in real time. Previous studies have used this noninvasive method to assess simultaneous morphology and function,2 and to visualize the phases of the MMC and the contractile activity post food.3

We have sought to validate the EPI method by comparing it with conventional perfused tube manometry. The design of the catheter is perfectly adequate for this validation, but we acknowledge the limitations of the tube design in that the spacing of the recording ports would probably make it unsuitable for the evaluation of antroduodenal coordination using manometry alone. Split screen video, however, allowed us to simultaneously display EPI images synchronized with their corresponding motility traces at speeds greater than real time, thus enabling us to make direct visual comparisons of gastroduodenal wall movement and contractile activity and also to assess antroduodenal motility.

EPI events do not represent 100% of all detected events, some being detected by manometry but not EPI. The majority of these were isolated duodenal pressure waves. Manometry detected waves at a distance of 5 cm from the pylorus but since the depth of image plane was only 1 cm EPI, in a few instances, was unable to detect these. This is obviously a limitation of the method but may be resolved in future studies by multislice imaging.

It was possible to visualize occlusive contractions occurring in the region of the pylorus using EPI ( Fig. 5), but as the image depth was only 1 cm, it became increasingly difficult to differentiate between nonocclusive and occlusive contractions in the more proximal antrum. Conventional manometric techniques are said to register only lumen-occlusive events and may not be able to detect subtle pressure changes following nonocclusive contractions.8 We did, however, find excellent agreement between the number of events recorded by EPI (431) and by manometry (404) with an overall 93% concordance. The overall detection rate for antroduodenal propagated events was, however, approximately three contractions detected by EPI for every two pressure waves detected by manometry, with a reversal of these figures in the duodenum. Since a visual assessment of wall movement was made for each EPI image and each propagating event could be accurately followed, these results imply an under-detection of propagated events as recorded by manometry. Similar results were seen when scintigraphy was validated by manometry.9 These differences may indeed result from nonocclusive events, which may play an important role in gastric emptying,10 or from primarily tonic changes since it has been shown that many propagating antral waves begin proximally as low-amplitude contractions and only occlude the lumen as they pass distally towards the gastroduodenal junction.11 Fone et al.8 also demonstrated that perfused tube manometry under-detected antral events by 41% post food when compared with changes in wall movement.

These results did, however, parallel other studies in which extraluminal strain gauges, sutured on the serosal surface of the bowel, were compared with similarly positioned perfused tubes.12 Valori et al.13 reported 98% agreement in the antrum and 90% in the duodenum, but these studies were performed in anaesthetized dogs and used artificial means to stimulate the antrum and so must be reviewed with caution. These results did, however, show that the perfused tube was particularly reliable when recording phasic contractions, but less so when monitoring tonic changes.

Ahluwalia et al.,14 using a novel intraluminal traction device to detect small intestinal propulsive activity, demonstrated that conventional manometry was not always able to predict propulsive activity. They measured simultaneous manometric activity at 2 cm proximal and distal to the device and showed that during the fasting phase III activity, 11% of manometrically propagated events produced no corresponding propulsive activity and, following food, showed even greater dissociation which was indicative of mixing activity.

There are still limitations with EPI. For example, subjects can be imaged only in the horizontal position and within the confines of the surrounding magnet, thereby imposing both postural and movement constraints. Routine radioisotopic measurements of gastric emptying are currently performed in a variety of positions, but recent work has highlighted the effects of gravity on gastric distribution and emptying of non-nutrient10 and nutrient meals.15, 16 The choice of test meal is also currently a limiting factor since our EPI method relies on the imaging qualities of water as a contrast medium, thereby restricting calorific meals to semisolids. While this is perfectly physiological it may, in some respects, be seen as a limitation of the system. Other workers17 have enhanced liquid contrast with gadolinium, but to date no suitable agent has been found for solids.

It is important to bear in mind the infancy of the technique together with these constraints when evaluating the present data. EPI is, however, the only current technique which has the potential to allow virtually simultaneous measurement of gastric emptying, motility and contractility, and gastric mixing and distribution. We conclude that continued research using EPI in the GI tract will enhance our knowledge of gut physiology and it is envisaged that in future EPI will be available for clinical studies thereby aiding the assessment of functional disorders of the GI system.

ACKNOWLEDGMENTS

We would like to thank Dr Alan Freeman, Valerie Adams, Dr Jon Hykin, Dr Paul Harvey and Rachel Moore for their help with data acquisition and production of the manuscript. We would also like to acknowledge the major support of the Medical Research Council for the funding of the echo-planar imaging programme.

Ancillary