Abstract
Background
Measuring mitral annular plane systolic excursion (MAPSE) serially in a single wall may be an effective method for monitoring global left ventricular (LV) function, especially when automated with a novel deep learning method using transesophageal echocardiography, called autoMAPSE. However, this assumption is untested in postoperative ICU patients, and the impact of regional wall motion abnormalities (RWMA) is unclear.
Aim
To assess the ability of single-wall autoMAPSE in tracking changes in global LV function (i.e., trending ability) using manual global MAPSE averaged from four walls as reference, and to explore how this ability was affected by RWMA.
Methods
This study was a secondary analysis of a prospective observational study. The changes in single-wall autoMAPSE and manual global MAPSE were calculated in 49 patients undergoing cardiac surgery. Trending ability evaluates how well a novel method detects changes in a reference method, which we assessed using four-quadrant plots with concordance rates. To explore the effects of RWMA, we classified RWMA by temporal behavior (persistent or dynamic) and by location relative to single-wall autoMAPSE (same or remote wall), yielding four patterns for analysis of trending ability: (i) remote wall, dynamic RWMA, (ii) remote wall, persistent RWMA, (iii) same wall, dynamic RWMA, and (iv) same wall, persistent RWMA.
Results
Overall, single-wall autoMAPSE had adequate trending ability (concordance rate 91%). All LV walls showed adequate trending ability (concordance rate ≥ 90%), except for the septal wall (concordance rate 88%). The effect of RWMA was negligible, as the concordance rates of the four patterns ranged from 89 to 96%.
Conclusions
In this secondary analysis, single-wall autoMAPSE tracked changes in global LV function in 9 out of 10 cases, irrespective of RWMA.
Graphical Abstract
Keywords: MAPSE, Left ventricular function, Transesophageal echocardiography, Deep learning, Artificial intelligence, Hemodynamic monitoring, Regional wall motion abnormalities
Introduction
Monitoring global left ventricular (LV) function with transesophageal echocardiography (TEE) during cardiac surgery is an important yet challenging task [1]. Due to competing demands, clinicians usually evaluate LV function by qualitative visual assessment alone. This practice is effective for detecting severe deteriorations in LV function [2]. However, such severe changes tend to be delayed with potentially irreversible complications. A real-time quantitative method that is quick and precise would, therefore, be useful to detect subtler and earlier changes in LV function [3, 4]. This goal could potentially be achieved by measuring the mitral annular plane systolic excursion (MAPSE) in a single wall [5]. MAPSE reflects global LV longitudinal function [3, 6], is more sensitive than left ventricular ejection fraction (LVEF), and can even be measured with TEE without visualizing the entire left ventricle. However, manual echocardiography remains challenging during rapid changes in the patient’s condition and when clinicians face competing tasks.
The potential to improve echocardiographic monitoring has motivated us to develop autoMAPSE, a novel method that combines 2D TEE with deep learning to monitor MAPSE automatically [5, 7, 8] and continuously [9]. Several studies have shown that MAPSE decreases after myocardial infarction in all the walls of the left ventricle, even in those remote from the infarcted territory [10–14]. However, by only reporting the average effects, the cited studies [10–14] do not address whether changes in MAPSE measured from a single wall can be used to adequately track changes in global LV function in individual patients with regional wall motion abnormalities (RWMA). This uncertainty must be resolved, not only because RWMA is common in both perioperative and critically ill patients [1, 15–18], but the excellent feasibility for monitoring LV function with autoMAPSE relies on measurements from a single wall [5, 7, 9].
Thus, this study aimed to test the ability of single-wall autoMAPSE in tracking changes in manual global MAPSE (i.e., trending ability) and to explore how this trending ability was affected by RWMA.
Methods
The present study is a secondary analysis of a patient cohort for which other data have been published previously [8, 9, 19–22]. The inclusion criteria were adult patients scheduled for on-pump cardiac surgery with intraoperative TEE. Exclusion criteria were age less than 18 years, unwillingness to participate, and contraindications to TEE. This study adheres to the principles of the Declaration of Helsinki. Approval was granted by the Regional Committee for Ethics in Medicine (REK number 2022/345556), and all participants provided written informed consent before inclusion. We adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist when composing this manuscript [23].
Echocardiographic image acquisition
We obtained 2D midesophageal two- and four-chamber images of the left ventricle at two time-points: (1) before commencing cardiopulmonary bypass (CPB) and (2) prior to extubation in the postoperative intensive care unit (ICU). The recordings were acquired by retroflecting and turning the probe (6VT-D connected to an E95 scanner, GE Healthcare, Horten, Norway) with the omniplane at 90 degrees to acquire a non-foreshortened midesophageal two-chamber (2C) view. Simultaneous biplane imaging (Multi-D®, GE Healthcare, Horten, Norway) was used to minimize LV foreshortening by aligning the secondary sector through the LV apex and rotating the omniplane angle in the secondary sector to approximately 0–30 degrees to obtain an optimized midesophageal four-chamber (4C) view with minimal foreshortening [22]. Finally, the sector width was maximized, and the temporal resolution was increased to at least 40 frames per second.
Automatic and manual measurement of MAPSE
AutoMAPSE and manual MAPSE were analyzed in the same pre-CPB and postoperative ICU recordings. Their respective changes from pre-CPB to ICU were calculated. MAPSE was measured as the mean of three to five cardiac cycles in the pre-CPB recordings and the mean of up to 10 cycles in the ICU [9].
The technical details and validation of autoMAPSE have been reported previously in detail [7–9]. Briefly, autoMAPSE is based on a deep learning convolutional network trained under supervision to detect the mitral annulus in 2D TEE recordings. The method was trained on recordings obtained from cardiology patients [7, 8] and validated in patients after cardiac surgery [5, 9]. AutoMAPSE detects the mitral annulus in each echocardiography frame, then measures MAPSE as the longitudinal distance traveled by the annulus from end-diastole to end-systole. Here, end-diastole is defined as the mitral annular point closest to the probe timed around the R-wave of the ECG, whereas end-systole is the point of nadir. To avoid erroneous outliers, the method rejects a cardiac cycle if the mitral annulus has a frame-to-frame distance of more than 5 mm, if the annulus is detected in less than 60% of the frames per cycle, or if the highest point does not correspond to the definition of end-diastole. Finally, autoMAPSE reports the MAPSE of each specific wall as the mean of all feasible cardiac cycles in that specific recording. Thus, using autoMAPSE in the 2C and 4C views provides MAPSE in four walls: anterior, inferior, lateral, and septal. AutoMAPSE is defined as feasible when it reports the MAPSE from a single wall, irrespective of the number of cardiac cycles measured [9].
Manual MAPSE was measured in B-mode recordings of the same four walls using EchoPAC software version 2.06 (GE Healthcare, Horten, Norway) [9], and the average MAPSE of these walls (global MAPSE) was used to define global LV function. We chose this definition of global LV function for two reasons. First, parameters of LV longitudinal function, such as MAPSE, are more sensitive to changes in LV systolic function than LVEF [6]. Indeed, our previous reports in this patient sample show that MAPSE decreases immediately after surgery [9, 22], whereas several studies show that LVEF remains relatively unchanged [19, 22, 24, 25]. Second, speckle-tracking analysis of global longitudinal strain is not possible with simultaneous biplane imaging (Multi-D®, GE Healthcare, Horten, Norway) with the present vendor (EchoPAC software version 2.06, GE Healthcare, Horten, Norway). Furthermore, global longitudinal strain requires pristine image quality of the left ventricle in all three standardized views, including the apex, which is generally challenging in the perioperative setting. The feasibility of global MAPSE was defined as the average manual MAPSE of all available walls and did not require measurements from all four.
Analysis of regional wall motion abnormalities
RWMA was analyzed in the same four walls as MAPSE (anterior, inferior, lateral, and septal) and dichotomously reported as present/absent if any segment of that wall was either hypokinetic, akinetic, or dyskinetic [26]. Two independent observers (JY and OG) assessed all the recordings for RWMA. Cases of disagreement were settled by a third independent observer (EARB) who was blinded to the conclusions of the first two observers. If disagreement was due to one observer rating the image as inadequate and the other rating it adequate, the final decision of image adequacy and RWMA was determined by the third observer, irrespective of the evaluation performed by the first two.
To explore how RWMA affects the ability of single-wall autoMAPSE in tracking global LV function, we first classified whether the RWMA was in the same wall as single-wall autoMAPSE or in a remote wall, where a remote wall was defined as any wall other than the one monitored by autoMAPSE. Next, we classified the evolution of those RWMA to be either persistent or dynamic, where dynamic RWMA comprised either newly developed RWMA postoperatively or the resolution of a pre-existing one. These two definitions yielded four RWMA–autoMAPSE patterns for analysis: (i) remote wall, dynamic RWMA, (ii) remote wall, persistent RWMA, (iii) same wall, dynamic RWMA, and (iv) same wall, persistent RWMA.
Statistical analysis
The primary objective of the present study was the trending ability of single-wall autoMAPSE as compared to manual global MAPSE. We analyzed trending ability using the four-quadrant plot and reported the primary outcome as the concordance rate with an exclusion zone [27]. Here, a concordant change means that autoMAPSE and manual global MAPSE change in the same direction, irrespective of magnitude. In this analysis, the exclusion zone was defined as the least significant change of autoMAPSE by hands-on imaging (1.8 mm), reported previously [9]. In general, a concordance rate of 90% is considered adequate [28, 29]. The secondary objective was to explore the impact of the four RWMA–autoMAPSE patterns on the trending ability of single-wall autoMAPSE.
The absolute change in autoMAPSE and manual global MAPSE was compared using paired t tests. Agreement for RWMA analysis between the initial two observers was assessed using Cohen’s kappa. Data are reported as the mean ± standard deviation or median [25–75 percentiles]. The statistical analysis plan was established prior to seeing the results. Missing data were not replaced. As the present study is a secondary analysis, the sample size was fixed to that of the primary study [9]. Statistical significance was defined as a P < 0.05. Stata 19.5 (StataCorp LLC) was used for all analyses.
Results
Fifty one patients were enrolled, of whom 49 remained for final analysis (Table 1). One patient was excluded from the ICU due to an airway issue unrelated to TEE, and another patient had missing pre-CPB recordings. The feasibility of autoMAPSE in the present study was 97% in the anterior wall, 84% in the inferior wall, 89% in the lateral wall, and 92% in the septal wall. The feasibility of manual global MAPSE was 100%, where 75% of all measurements comprised all four walls, 87% at least three, and 100% comprised at least two walls, averaging 3.6 ± 0.7 walls per measurement of manual global MAPSE.
Table 1.
Patient characteristics
| Variables | Values (N = 49) |
|---|---|
| Age (years) | 67 ± 8 |
| Male/female (n, %) | 38/11 (78/22) |
| Height (cm) | 175 ± 8 |
| Weight (kg) | 83 ± 16 |
| BMI (kg/m2) | 27 ± 5 |
| Coronary artery disease (n, %) | 30 (61) |
| Left main stenosis | 9 (18) |
| 1-vessel disease | 5 (10) |
| 2-vessel disease | 8 (16) |
| 3-vessel disease | 17 (35) |
| Recent myocardial infarction (n, %) | 14 (29) |
| Preoperative troponin-T (ng/L) | 16 [11–51] |
| Preoperative NT-pro BNP (ng/L) | 481 [141–1342] |
| Preoperative LVEF (n, %) | |
| > 50% | 35 (71) |
| 31–50% | 10 (20) |
| 21–30% | 4 (8) |
| Preoperative RWMA (n, %) | 17 (35) |
| MAPSE pre-CPB (mm) | |
| Anterior autoMAPSE | 8.7 ± 2.7 |
| Inferior autoMAPSE | 11.2 ± 2.9 |
| Lateral autoMAPSE | 11.3 ± 3.0 |
| Septal autoMAPSE | 8.1 ± 2.0 |
| Manual global MAPSE | 11.1 ± 2.7 |
| MAPSE postoperative ICU (mm) | |
| Anterior autoMAPSE | 6.7 ± 2.0 |
| Inferior autoMAPSE | 7.5 ± 2.6 |
| Lateral autoMAPSE | 8.1 ± 2.4 |
| Septal autoMAPSE | 6.4 ± 1.9 |
| Manual global MAPSE | 7.6 ± 2.1 |
| Operation (n, %) | |
| Isolated CABG | 23 (47) |
| Aortic valve replacement | 20 (41) |
| Mitral valve replacement | 4 (8) |
| Mitral valve repair | 1 (2) |
| Surgery on the thoracic aorta | 6 (12) |
| Redo cardiac surgery | 3 (6) |
| EuroScoreII (%) | 1.6 [1.0–2.66] |
| Aortic cross-clamp time (min) | 72 ± 35 |
| Cardiopulmonary bypass time (min) | 98 ± 46 |
Bold signifies a category with subgroups
Values are mean ± SD, n (%), or median [interquartile range]
AutoMAPSE automatic measurements of mitral annular plane systolic excursion in a single wall using deep learning and transesophageal echocardiography, BMI body mass index, CABG coronary artery bypass grafting, CPB cardiopulmonary bypass, LVEF left ventricular ejection fraction, MAPSE mitral annular plane systolic excursion measured manually on hands-on images, NT-pro BNP N-terminal pro B-type natriuretic peptide, RWMA regional wall motion abnormality
Trending ability of single-wall autoMAPSE
Overall, single-wall autoMAPSE had adequate trending ability in all but 12 of 138 cases, with a concordance rate of 91% (Fig. 1). None of the 12 failed cases showed signs of intraventricular dyssynchrony or a mobile apex, nor did they have mitral valve prosthesis. Trending ability for each wall was also adequate except for the septal wall, which failed to meet the adequacy threshold by 2% (Fig. 2).
Fig. 1.
Four-quadrant plot showing the pooled trending ability of single-wall autoMAPSE versus manual global MAPSE (n = 138 paired observations). The origin box indicates a central exclusion zone of 1.8 mm, which corresponds to the least significant change of autoMAPSE. The dashed line shows the identity line. Single-wall Single-wall Auto automatic measurements of mitral annular plane systolic excursion in a single wall using deep learning and transesophageal echocardiography, MAPSE mitral annular plane systolic excursion
Fig. 2.
Four-quadrant plots showing the trending ability of single-wall autoMAPSE for each of the four left ventricular walls. The origin box indicates a central exclusion zone of 1.8 mm, which corresponds to the least significant change of autoMAPSE. The dashed line shows the identity line. After applying the exclusion zone, the number of paired observations for each wall was 39 for the anterior wall, 32 for the inferior wall, 34 for the lateral wall, and 33 for the septal wall. AutoMAPSE automatic measurements of mitral annular plane systolic excursion in a single wall using deep learning and transesophageal echocardiography, MAPSE mitral annular plane systolic excursion
Single-wall autoMAPSE decreased significantly in all four walls from pre-CPB to the ICU (anterior −2.0 mm, inferior −3.6 mm, lateral −3.1 mm, and septal −1.8 mm; P < 0.001 for all), as did manual global MAPSE (−3.5 mm, P < 0.001). The absolute changes in single-wall autoMAPSE differed significantly from those of global MAPSE in the anterior and septal wall (Table 2), suggesting a lack of interchangeability between these two parameters.
Table 2.
Changes in single-wall autoMAPSE compared to paired changes in manual global MAPSE
| AutoMAPSE (wall) | Observations (n) | ΔAutoMAPSE (mm) | ΔManual global MAPSE (mm) | Difference (mm) | P-value |
|---|---|---|---|---|---|
| Anterior | 45 | −1.9 ± 2.7 | −3.3 ± 2.8 | 1.4 ± 2.0 | < 0.001 |
| Inferior | 34 | −3.2 ± 3.4 | −3.1 ± 3.1 | −0.1 ± 1.7 | 0.71 |
| Lateral | 37 | −3.1 ± 4.0 | −3.2 ± 3.0 | 0.1 ± 2.4 | 0.88 |
| Septal | 39 | −1.8 ± 2.3 | −3.0 ± 2.6 | 1.22 ± 2.1 | < 0.001 |
The values for ΔManual global MAPSE vary for each row, because they show four paired comparisons with each wall. Values are mean ± SD
AutoMAPSE automatic measurements of mitral annular plane systolic excursion in a single wall using deep learning and transesophageal echocardiography, MAPSE mitral annular plane systolic excursion
Notably, the overall trending ability for manual single-wall MAPSE was also adequate (concordance rate 95%, pooled from all four walls). A subgroup analysis showed that trending ability was also adequate for each of the four walls with single-wall manual MAPSE, with concordance rates ranging from 93% (lateral wall) to 100% (anterior wall).
Effects of regional wall motion abnormalities
RWMA was detected in the anterior wall in 27% of the recordings (indeterminate 3%), inferior in 21% (indeterminate 11%), septal in 29% (indeterminate 1%), and lateral in 13% of the recordings (indeterminate 16%). The RWMA analysis had moderate agreement between the initial two observers before the third observer intervened (kappa = 0.59).
The effects of the four RWMA patterns on the trending ability of single-wall AutoMAPSE were negligible, as only one of the patterns showed a marginally inadequate concordance rate of 89% (Fig. 3).
Fig. 3.
Four-quadrant plots showing the trending ability of single-wall autoMAPSE for the four patterns of regional wall motion abnormality. The origin box indicates a central exclusion zone of 1.8 mm, which corresponds to the least significant change of autoMAPSE. The dashed line shows the identity line. AutoMAPSE automatic measurements of mitral annular plane systolic excursion in a single wall using deep learning and transesophageal echocardiography, MAPSE mitral annular plane systolic excursion, RWMA regional wall motion abnormality
Discussion
Our main finding is that single-wall autoMAPSE tracks changes in global LV function in 9 out of 10 cases. This was also the case when single-wall MAPSE was measured manually. The effect of RWMA on this trending ability was negligible. Thus, it appears that MAPSE from one wall, measured manually or by autoMAPSE, may track changes in global LV function, irrespective of RWMA.
RWMA was relatively prevalent in this study, but it is doubtful that all these cases represent myocardial ischemia. Not only can myocardial stunning or the use of pacemakers induce RWMA, but postoperative patients may suffer from relative or absolute hypovolemia, which can cause both global and regional hypokinesia [30, 31]. RWMA of any etiology was relevant for the study’s objective, even if they did not necessarily represent myocardial ischemia.
Why single-wall MAPSE tracks global left ventricular function irrespective of regional wall motion abnormalities
The ability of single-wall MAPSE in tracking global LV function, even during RWMA, may be explained by two complementary mechanisms: first, the geometric relationship, whereby global MAPSE is proportional to stroke volume and inversely proportional to the LV outer cross-sectional area [32], and second, the effects of tethering, where all the contractile movements in the left ventricle affect the entire mitral annulus and can be monitored by quantifying MAPSE at a single wall.
An acute RWMA of a certain magnitude reduces LV stroke volume and emptying, which in turn increases LV end-systolic and end-diastolic volumes [33]. This acute ventricular dilatation increases afterload in early-systole according to the Law of Laplace and preload in end-diastole, which may or may not maintain stroke volume depending on the magnitude of the RWMA, the position of the left ventricle on the Frank–Starling curve, and the contractile status of the remaining myocardium. Irrespective of the effects on stroke volume, this LV dilatation will increase the LV outer cross-sectional area and, therefore, reduce the global MAPSE, given the inverse relationship between MAPSE and LV outer cross-sectional area [32]. The effect of RWMA on global MAPSE is, therefore, determined by the balance between LV stroke volume and its cross-sectional area. As the entire annular ring is tethered to the left ventricle, RWMA in one wall will affect the MAPSE in a remote wall, depending on the factors mentioned. The same mechanisms explain why it is still useful to monitor the single-wall MAPSE of a persistently akinetic wall. The akinetic wall is anchored to the mitral annulus and the remaining functional segments. When these segments shorten, traction is applied to the akinetic region, and this motion is translated into the mitral annulus. However, these mechanisms may be less reliable in patients with intraventricular dyssynchrony, mitral valve prosthesis, or an abnormal mobile apex. Although we did not identify these potential limitations in any of the failed cases, they may still be important. Nevertheless, single-wall MAPSE does not reflect regional function, but the global interactions between LV loading conditions and contractility.
Clinical implications
These findings have some practical implications when using MAPSE to monitor LV function: MAPSE from a single wall is likely adequate for serial monitoring of global LV function. This means that the already-simple method of averaging MAPSE from the septal and lateral walls can be further simplified to one wall for clinicians with competing tasks. When only a single wall is required for MAPSE, the advantage is that the clinician can choose the view with optimal image quality, which can improve the measurement reproducibility and precision during monitoring.
Regarding the use of single-wall autoMAPSE, we should note that the absolute values of MAPSE [5, 9, 19, 34] and their respective changes differ between walls. Thus, the absolute measurement of single-wall MAPSE is not necessarily interchangeable between walls or with the global MAPSE. Furthermore, measurements of MAPSE with B-mode are not interchangeable with M-mode [35]. Thus, we should not blindly apply normal values of MAPSE derived from larger epidemiological studies. AutoMAPSE in the anterior wall has repeatedly been the most feasible one [5, 9], whereas this present study suggests that the changes in lateral wall autoMAPSE are most interchangeable with the changes in manual global MAPSE. Finally, in line with previous studies [10–14], we should not use MAPSE in a wall to detect new-onset RWMA.
Limitations
The first limitation is that the sample size was relatively small. Thus, the study may have been underpowered to exclude an effect of RWMA on the trending ability of single-wall autoMAPSE. However, the concordance rates in the four patterns of RWMA suggest that any undetected effect of RWMA on trending ability is relatively small. Second, the septal wall autoMAPSE did not show adequate trending ability (concordance rate 88%). However, a single observation in the other direction would have tipped the results to meet the threshold due to the small sample size. Notably, the same septal wall had adequate trending ability when assessed by single-wall manual MAPSE. It may, therefore, be possible that the septal wall autoMAPSE can be used if no other walls are visible. Nevertheless, the findings of this study, together with previous studies suggesting limited monitoring feasibility in this wall [7, 9], suggest that the septal wall MAPSE should not be the wall-of-choice, especially in patients undergoing cardiac surgery. Third, our data did not allow comparison of changes in MAPSE versus those of global longitudinal strain. However, the two parameters are highly correlated [36], and MAPSE was even found to have superior prognostic abilities compared to global longitudinal strain in a head-to-head comparison using cardiac magnetic resonance imaging in a cardiology cohort [37]. Finally, the subjective visual analysis of RWMA was performed by two echocardiographic researchers and not by expert cardiologists. However, visual analysis has suboptimal reproducibility even among experts [38, 39]. There are currently no gold standards for RWMA analysis, and the method used in this study is the current clinical standard in perioperative and critical care echocardiography. We did not use ultrasound contrast agents to optimize RWMA assessment, as the present study is a secondary analysis of already-obtained data. Novel methods for more accurate quantification of regional function are underway, but none have yet reached clinical application [21, 40]. Given the uncertainty of the assessment of RWMA and the lack of a gold standard, our approach using a third independent observer to settle the disagreement between the first two yielded a consistent agreement among at least two observers.
Conclusion
In this secondary analysis, MAPSE from a single wall tracked changes in global LV function in 9 out of 10 cases, irrespective of RWMA. Single-wall autoMAPSE has the advantage of being fully automated, quick, and precise, and is, therefore, a feasible tool for automatic monitoring of LV function in patients undergoing cardiac surgery.
Take-home message
Measuring mitral annular plane systolic excursion (MAPSE) in a single wall is a potentially effective method for monitoring global left ventricular function, especially when automated with a novel deep learning method using transesophageal echocardiography, called autoMAPSE. This study found that MAPSE from a single wall, measured manually or by autoMAPSE, adequately tracks changes in global left ventricular function in 9 out of 10 cases, irrespective of regional wall motion abnormalities.
Acknowledgements
We would like to thank all the patients for their voluntary participation, the Department of Anesthesia and Intensive Care, St. Olavs Hospital, Trondheim University Hospital, Norway, for the use of their clinical equipment, and Simon Fougner Hartmanns Familiefond, Copenhagen, Denmark, for the generous and unrestricted supply of funding for a dedicated E95-scanner to this project.
Abbreviations
- 2C
Midesophageal two-chamber view
- 4C
Midesophageal four-chamber view
- AutoMAPSE
Automatic measurement of mitral annular plane systolic excursion
- CPB
Cardiopulmonary bypass
- ICU
Intensive care unit
- LV
Left ventricular
- LVEF
Left ventricular ejection fraction
- MAPSE
Mitral annular plane systolic excursion
- TEE
Transesophageal echocardiography
- RWMA
Regional wall motion abnormality
Author contributions
Conceptualization: JY, OG, EARB, AAT, IKG, BG, GK, and SA; data curation: JY, OG, and EARB; formal analysis: JY, OG, EARB, and AAT; funding acquisition: SA; investigation: JY and OG; methodology: JY, OG, EARB, AAT, IKG, BG, GK, and SA; project administration: JY, SA, and GK; resources: JY, AAT, EARB, TDT, KHS, IKG, BG, GK, and SA; software: JY, AAT, EARB, BG, GK, and SA; supervision: SA and GK; validation: JY; visualization: JY, OG, AAT, EARB, and SA; writing—original draft: JY and OG; writing—review and editing: JY, OG, EARB, AAT, TDT, KHS, IKG, BG, GK, and SA.
Funding
Open access funding provided by NTNU Norwegian University of Science and Technology (incl St. Olavs Hospital - Trondheim University Hospital). The current project was funded by Samarbeidsorganet, Central Norway Regional Health Authority (Helse Midt-Norge).
Data availability
The datasets used and analyzed during the current study are unavailable to other parties due to a lack of patient consent for this purpose.
Declarations
Ethics approval and consent to participate
This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Regional Committee for Ethics in Medicine (REK number 2022/345556).
Consent for publication
Consent to publish anonymized patient data was obtained from all patients.
Competing interests
BG and EAB have received research funding from the Centre for Innovative Ultrasound Solutions (CIUS)—a Norwegian Research Council (NRC) center for research-based innovation, where GE HealthCare, Horten, Norway, was one of the institutional partners. BG holds a position at Precision Health Center for Optimized Cardiac Care (ProCardio)—another NRC center for research-based innovation, where GE HealthCare is a partner. SA was provided an E95-scanner by GE Healthcare for our institution (Dept of Circulation and Medical Imaging, NTNU) without restrictions for research.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Swaminathan M, Morris RW, De Meyts DD et al (2007) Deterioration of regional wall motion immediately after coronary artery bypass graft surgery is associated with long-term major adverse cardiac events. Anesthesiology 107(5):739–745 [DOI] [PubMed] [Google Scholar]
- 2.Vieillard-Baron A, Charron C, Chergui K, Peyrouset O, Jardin F (2006) Bedside echocardiographic evaluation of hemodynamics in sepsis: is a qualitative evaluation sufficient? Intensive Care Med 32(10):1547–1552 [DOI] [PubMed] [Google Scholar]
- 3.Yu J, Aakhus S, Chew MS (2025) Monitoring left ventricular function with MAPSE and critical care echocardiography. Intensive Care Med 51(6):1176–1178 [DOI] [PubMed] [Google Scholar]
- 4.Squara P, Cecconi M, Rhodes A, Singer M, Chiche JD (2009) Tracking changes in cardiac output: methodological considerations for the validation of monitoring devices. Intensive Care Med 35(10):1801–1808 [DOI] [PubMed] [Google Scholar]
- 5.Yu J, Taskén AA, Flade HM et al (2024) Automatic assessment of left ventricular function for hemodynamic monitoring using artificial intelligence and transesophageal echocardiography. J Clin Monit Comput 38(2):281–291 [DOI] [PubMed] [Google Scholar]
- 6.Hu K, Liu D, Herrmann S et al (2013) Clinical implication of mitral annular plane systolic excursion for patients with cardiovascular disease. Eur Heart J Cardiovasc Imaging 14(3):205–212 [DOI] [PubMed] [Google Scholar]
- 7.Berg EAR, Taskén AA, Nordal T et al (2023) Fully automatic estimation of global left ventricular systolic function using deep learning in transoesophageal echocardiography. Eur Heart J 1(1):qyad007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Taskén AA, Yu J, Berg EAR et al (2024) Automatic detection and tracking of anatomical landmarks in transesophageal echocardiography for quantification of left ventricular function. Ultrasound Med Biol 50(6):797–804 [DOI] [PubMed] [Google Scholar]
- 9.Yu J, Taskén AA, Berg EAR et al (2024) Continuous monitoring of left ventricular function in postoperative intensive care patients using artificial intelligence and transesophageal echocardiography. Intensive Care Med Exp 12(1):54 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Pahlm U, Ostenfeld E, Seemann F et al (2020) Evolution of left ventricular function among subjects with ST-elevation myocardial infarction after percutaneous coronary intervention. BMC Cardiovasc Disord 20(1):309 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Pahlm U, Seemann F, Engblom H et al (2018) Longitudinal left ventricular function is globally depressed within a week of STEMI. Clin Physiol Funct Imaging 38(6):1029–1037 [DOI] [PubMed] [Google Scholar]
- 12.Støylen A, Skjærpe T (2003) Systolic long axis function of the left ventricle. Global and regional information. Scand Cardiovasc J 37(5):253–258 [DOI] [PubMed] [Google Scholar]
- 13.Berg J, Jablonowski R, Nordlund D et al (2020) Decreased atrioventricular plane displacement after acute myocardial infarction yields a concomitant decrease in stroke volume. J Appl Physiol 128(2):252–263 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Höglund C, Alam M, Thorstrand C (1989) Effects of acute myocardial infarction on the displacement of the atrioventricular plane: an echocardiographic study. J Intern Med 226(4):251–256 [DOI] [PubMed] [Google Scholar]
- 15.Cavefors O, Holmqvist J, Bech-Hanssen O et al (2021) Regional left ventricular systolic dysfunction associated with critical illness: incidence and effect on outcome. ESC Heart Fail 8(6):5415–5423 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.London MJ, Tubau JF, Wong MG et al (1990) The “natural history” of segmental wall motion abnormalities in patients undergoing noncardiac surgery. Anesthesiology 73(4):644–655 [DOI] [PubMed] [Google Scholar]
- 17.Comunale ME, Body SC, Ley C et al (1998) The concordance of intraoperative left ventricular wall-motion abnormalities and electrocardiographic S-T segment changes: association with outcome after coronary revascularization. Anesthesiology 88(4):945 [DOI] [PubMed] [Google Scholar]
- 18.Leung JM, O’Kelly B, Browner WS, Tubau J, Hollenberg M, Mangano DT (1989) Prognostic importance of postbypass regional wall-motion abnormalities in patients undergoing coronary artery bypass graft surgery. Anesthesiology 71(1):16 [DOI] [PubMed] [Google Scholar]
- 19.Yu J, Taskén AA, Berg EAR et al (2025) Continuous three-dimensional transesophageal echocardiography and deep learning for perioperative monitoring of left ventricular longitudinal function. Eur Heart J Imaging Methods Pract 3(1):qyaf052 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Yu J, Høyland K, Berg EAR, Tannvik TD, Grenne B, Aakhus S (2025) Right ventricular ejection fraction assessed by single-beat three-dimensional transesophageal echocardiography in mechanically ventilated patients. Anesth Analg 141(2):435 [DOI] [PubMed] [Google Scholar]
- 21.Taskén AA, Judge T, Berg EAR et al (2025) Estimation of segmental longitudinal strain in transesophageal echocardiography by deep learning. IEEE J Biomed Health Inform. 10.1109/JBHI.2025.3605793 [DOI] [PubMed] [Google Scholar]
- 22.Yu J, Tannvik TD, Taskén AA et al (2025) Monitoring systemic ventriculoarterial coupling after cardiac surgery using continuous transoesophageal echocardiography and deep learning. J Clin Monit Comput. 10.1007/s10877-025-01328-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP (2007) Strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ 335(7624):806–808 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Dybos Tannvik T, Kiss G, Torp H, Eskeland Rimehaug A, Kirkeby-Garstad I (2020) No evidence of cardiac stunning or decoupling immediately after cardiopulmonary bypass for elective coronary surgery. Acta Anaesthesiol Scand 64(8):1128–1135 [DOI] [PubMed] [Google Scholar]
- 25.Chen W, Ortiz-Leon XA, Posada-Martinez EL et al (2021) Acute changes of left ventricular function during surgical revascularization by 3D speckle tracking. Echocardiography 38(4):623–631 [DOI] [PubMed] [Google Scholar]
- 26.Lang RM, Badano LP, Mor-Avi V et al (2015) Recommendations for cardiac chamber quantification by echocardiography in adults: an update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging. J Am Soc Echocardiogr 28(1):1-39.e14 [DOI] [PubMed] [Google Scholar]
- 27.Saugel B, Grothe O, Wagner JY (2015) Tracking changes in cardiac output: statistical considerations on the 4-quadrant plot and the polar plot methodology. Anesth Analg 121(2):514–524 [DOI] [PubMed] [Google Scholar]
- 28.Lamarche-Fontaneto R, Oud L, Howell KD et al (2025) Cardiac output monitors in septic shock: do they deliver what matters? A systematic review and meta-analysis. Crit Care 29(1):299 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Critchley LA, Lee A, Ho AMH (2010) A critical review of the ability of continuous cardiac output monitors to measure trends in cardiac output. Anesth Analg 111(5):1180–1192 [DOI] [PubMed] [Google Scholar]
- 30.Seeberger MD, Cahalan MK, Rouine-Rapp K et al (1997) Acute hypovolemia may cause segmental wall motion abnormalities in the absence of myocardial ischemia. Anesth Analg 85(6):1252 [DOI] [PubMed] [Google Scholar]
- 31.Feldheiser A, Juhl-Olsen P, Nordine M et al (2023) A comprehensive echocardiographic analysis during simulated hypovolaemia: an observational study. Eur J Anaesthesiol. 10.1097/EJA.0000000000001863 [DOI] [PubMed] [Google Scholar]
- 32.Carlsson M, Ugander M, Mosén H, Buhre T, Arheden H (2007) Atrioventricular plane displacement is the major contributor to left ventricular pumping in healthy adults, athletes, and patients with dilated cardiomyopathy. Am J Physiol Heart Circ Physiol 292(3):H1452–H1459 [DOI] [PubMed] [Google Scholar]
- 33.Katz AM (2002) Ernest Henry Starling, his predecessors, and the “Law of The Heart”. Circulation 106(23):2986–2992 [DOI] [PubMed] [Google Scholar]
- 34.Støylen A, Mølmen HE, Dalen H (2023) Regional motion of the AV-plane is related to the cardiac anatomy and deformation of the AV-plane. Data from the HUNT study. Clin Physiol Funct Imaging 43(6):453–462 [DOI] [PubMed] [Google Scholar]
- 35.Wang YH, Sun L, Li SW et al (2023) Normal reference values for mitral annular plane systolic excursion by motion-mode and speckle tracking echocardiography: a prospective, multicentre, population-based study. Eur Heart J 24(10):1384–1393 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Huang SJ, Ting I, Huang AM, Slama M, McLean AS (2017) Longitudinal wall fractional shortening: an M-mode index based on mitral annular plane systolic excursion (MAPSE) that correlates and predicts left ventricular longitudinal strain (LVLS) in intensive care patients. Crit Care 21(1):292 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Xue H, Artico J, Davies RH et al (2022) Automated in-line artificial intelligence measured global longitudinal shortening and mitral annular plane systolic excursion: reproducibility and prognostic significance. J Am Heart Assoc 11(4):e023849 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Hoffmann R, von Bardeleben S, Kasprzak JD et al (2006) Analysis of regional left ventricular function by cineventriculography, cardiac magnetic resonance imaging, and unenhanced and contrast-enhanced echocardiography: a multicenter comparison of methods. J Am Coll Cardiol 47(1):121–128 [DOI] [PubMed] [Google Scholar]
- 39.Hoffmann R, von Bardeleben S, Barletta G et al (2014) Comparison of two- and three-dimensional unenhanced and contrast-enhanced echocardiographies versus cineventriculography versus cardiac magnetic resonance for determination of left ventricular function. Am J Cardiol 113(2):395–401 [DOI] [PubMed] [Google Scholar]
- 40.Nyberg J, Østvik A, Salte IM et al (2024) Deep learning improves test–retest reproducibility of regional strain in echocardiography. Eur Heart J 2(4):qyae092 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets used and analyzed during the current study are unavailable to other parties due to a lack of patient consent for this purpose.