This paper presents ‘GEMS’, a new software package for the analysis and visualization of multi-electrode GI electrical recordings. This software platform effectively incorporates a number of recent analytical advances in the field of GI mapping into a coherent framework coupled to an intuitive and user-friendly GUI. The most significant value of GEMS is that it allows for substantial gains in efficiency and productivity, by automating laborious functions and quantitative analyses that must otherwise be performed manually (Table). The package also allows for the rapid generation of high-quality graphical outputs suitable for presentation and publication.
As outlined in Table, the software has undergone rigorous quantitative validation, with qualitative validation also achieved through application in recent experimental and clinical works [10
]. These studies demonstrate the utility of GEMS in accelerating efforts to better understand the causes and consequences of abnormal slow wave activity occurring in disease states. Specific problems suitable to GEMS analyses include defining abnormal slow wave initiation and conduction patterns, including in clinical disorders such as gastroparesis, in which ICC numbers are reduced [10
]. Importantly, these slow wave initiation and conduction abnormalities may often occur within normal frequency ranges, potentially being missed by techniques lacking in spatial resolution, like cutaneous EGG [10
]. GEMS analyses have also recently been applied to show, for the first time, that high-velocity and high-amplitude activity routinely emerges in association with many dysrhythmic events [28
]. In future, this finding may help in the localization and treatment of dysrhythmic sources.
Activation time marking, cycle clustering and isochronal mapping are all complex activities that must be undertaken with significant care to ensure accuracy is maintained and assumptions are reasonable [34
]. The automated mapping algorithms currently employed in GEMS are validated and capable of producing accurate spatiotemporal maps, but must be used with caution, knowledgeable application of parameters, and in association with thorough manual review, particularly when the raw data quality is variable. To this end, GEMS is equipped with a broad range of manual analysis options.
As in cardiac mapping, some unresolved difficulties remain in the mapping process [27
]. In particular, multiphasic ‘fractionated’ electrograms of long duration can occur in normal activity in the corpus [7
], or during complex sequences, potentially introducing ambiguity into FEVT or manually-derived activation time marks [34
]. Such complex activation events may arise due to electrical complexity in the propagation of wavefronts through the underlying tissue structure [35
]. Currently, we adhere to a convention that the activation time of such events be marked to the first major deflection in the multiphasic deflection, and manual adjustments to FEVT results may occasionally be required. In addition, cycle clustering may be challenging when slow wave activity becomes highly disorganized, as can occur during complex dysrhythmias [9
], potentially resulting in unreliable results from REGROUPS [14
]. In these circumstances, as in cardiac mapping, resorting to propagation movies can be a productive solution [27
], and the movie capability within GEMS is therefore a significant asset.
Another issue is that automated contour generating algorithms such as the ones employed by GEMS may incorrectly ‘assume’ that it is always permissible to interpolate between two given activation times [27
]. This assumption may lead to incorrect and potentially misleading ‘crowding’ of isochrones in the presence of an activation block, which is now known to occur during a range of gastric dysrhythmias [10
], requiring manual correction of the maps. Improvement on the current automated mapping algorithm to account for conduction blocks is therefore a focus of current work, as achieved previously in the cardiac field [36
GEMS provides an extensible framework for data analysis, and we anticipate that future enhancements will continue to be added by the user community. One particular focus of interest is the application of using GEMS to analyze mapping data from other sections of the GI tract, notably the small intestine. Small intestine motility has been the focus of several HR electrical mapping studies in recent years, performed by Lammers et al. in SmoothMap [37
]. With the recent steps toward successful human and clinical translation of HR gastric mapping [8
], the opportunity now exists to similarly expand small intestinal mapping applications, and GEMS could be a valuable tool. However, the re-optimization of key algorithms such as FEVT to small intestinal slow waves will be necessary to ensure that accuracy and efficiency is maintained [33
]. Frequency-domain analyses, such as fast-Fourier transform methods, could also be added to the software, if desired by the research community. HR electrical mapping is also now being productively applied in other excitable smooth muscle organs, such as the ureter, presenting further potential applications for GEMS [39
To date, GEMS has only been applied to analyze GI slow wave activity. ‘Spikes’ are also described as smooth muscle action potentials, and these events have been shown to propagate in specific patterns [40
]. Work is currently being undertaken to expand GEMS to allow semi-automated spike detection and mapping in the future.
Currently, GEMS is an off-line analysis system, for use after the completion of studies. Efforts are now being directed to further develop GEMS into an online mapping system suitable for real-time use [32
]. In future, the software could also be adapted for analysing cutaneous signals (electrogastrography; EGG), potentially supporting the development of HR body surface EGG potential mapping, as proposed recently by Du et al. [41
]. However, the algorithms are presently designed for serosal data and would therefore need extensive modifications to be useful for EGG studies.