Evaluation of a shorter algorithm in an automated analysis of sublingual microcirculation

Abstract

OBJECTIVE:

Diagnostic and risk stratification in intensive and emergency medicine must be fast, accurate, and reliable. The assessment of sublingual microcirculation is a promising tool for this purpose. However, its value is limited because the measurement is time-consuming in unstable patients. This proof-of-concept validation study examines the non-inferiority of a reduced frame rate in image acquisition regarding quality, measurement results, and time.

METHODS:

This prospective observational study included healthy volunteers. Sublingual measurement of microcirculation was performed using a sidestream dark field camera (SDF, MicroVision Medical®). Video-quality was evaluated with a modified MIQS (microcirculation image quality score). AVA 4.3C software calculated microcirculatory parameters.

RESULTS:

Thirty-one volunteers were included. There was no impact of the frame rate on the time needed by the software algorithm to measure one video (4.5 ± 0.5 minutes) for AVA 4.3C. 86 frames per video provided non inferior video quality (MIQS 1.8 ± 0.7 for 86 frames versus MIQS 2.2 ± 0.6 for 215 frames, p < 0.05), equal results for all microcirculatory parameters, but did not result in an advantage in terms of speed. No complications occurred.

CONCLUSION:

Video captures with 86 frames offer equal video quality and results for consensus parameters compared to 215 frames. However, there was no advantage regarding the time needed for the overall measurement procedure.

Keywords

Microcirculation risk-stratification SDF-imaging point-of-care measurement

1 Introduction

In the times of limited time resources both in emergency, and intensive care medicine, new fast and reliable tools for outcome prediction and risk stratification are urgently needed. In the past, this triage assessment often focuses on macrocirculatory parameters, but it has become clear that macrocirculation alone does not necessarily reflect factual organ perfusion. Organ perfusion is essential for the outcome but depends mainly on the microcirculation [1 –4]. Focusing on macrohemodynamics might detect an impaired microcirculation too late, and miss severe systemic disorders such as shock in an early phase [1]. The gold standard of microcirculation assessment, serum lactate measurement, has several disadvantages: it is time-delayed, subject to pre-analytical errors, and always gives only a “static snapshot”. Although clearance models can increase the informative value [2], they do not solve the fundamental time delay in diagnostics. Thus, the microcirculation should optimally be measured live and in vivo during the first patient contact [5]. For this purpose, easy to use hand-held devices that assess sublingual microcirculation with a sidestream-darkfield (SDF) camera have been introduced [6]. To eliminate the need for a subjective time-consuming interpretation, a software-based algorithm calculates parameters that correspond to the European Society of Intensive Care Medicine (ESICM) consensus [7]. Technically, SDF-measurements are possible in many circumstances such as weightless [8], the emergency department in dehydrated frail patients [9], in an emergency ambulance [10], or even in small mice [11]. In clinical practice, time is a critical factor in the assessment of critically ill patients. Time is necessary both for video recording and analysis. At the moment, the microcirculation image quality score (MIQS) defines a minimum of 3 seconds (90 frames) as acceptable and ≥5 seconds (150 frames) as well [12]. This corresponds to the consensus statements of the ESICM, which recommend using image sequences of motion-free images of at least 4 seconds (100 frames), but ideally of 20 seconds [7]. However, according to the manufacturer, the AVA-algorithm used in AVA 4.3C needs only 86 stable frames (2 seconds) for a sufficient automatic analysis. SDF-technology still has not reached daily clinical practice. This study investigates whether 86 frames show equal results to 215 frames in terms of quality and the results calculated by AVA 4.3C in healthy volunteers.

2 Methods

2.1 Ethics

The study was approved by the German Ethics Committee of the Medical Faculty of the University Hospital Duesseldorf, Germany. Written informed consent was obtained by all participating probands of the study. The principles of the Declaration of Helsinki and the national rules and regulations on personal data protection were applied.

2.2 Data recording

Medical history and clinical data have been documented. All volunteer’s data were anonymized.

2.3 Quality control

Before analysis, all videos were evaluated by two experienced raters according to a modified microcirculation image quality score (MIQS), that had been originally introduced by Massey et al. [12]. MIQS rates the videos into three categories: “good”, “acceptable”, and “non-acceptable”. In brief, five different criteria were evaluated: Illumination, focus, content, stability, and pressure. Due to the study design, the duration was not rated. A video without significant impairment in all criteria received zero points. Mild impairment resulted in 1 point for each impaired criterion. Severe impairment in one criterion was defined to be rated with 10 points, which results in the category “non-acceptable” (Table 1).

Table 1
The modified MIQS, Microcirculation image quality score (MIQS), adopted from (1)

Description Good (0 points) Acceptable (1 point) Unacceptable (10 points)

Illumination: Brightness and contrast of the video. Even illumination across the entire field of view. Contrast sufficient to see small vessels against a background of tissue. The video borders on being too dark or bright to distinguish vessels from tissue but the vessels are still identifiable. The video is oversaturated/too bright or too dark to make out analyzable features. Insufficient contrast to resolve flow rate.

Focus: Image sharpness in region of interest. Good focus for all vessels (small and large) in the entire field of view. Plasma gaps and red blood cells are visible. <1/2 field of view is out of focus or edges of the vessels are slightly out of focus. Video is completely out of focus such that no small vessels can be seen.

Content: Determination of the types of vessels and/or presence of occluding artifacts in the image. If the video presents overall pathology. Video is free of occlusions. Good distribution of large and small vessels. Less than 30% of the vessels are looped upon themselves. Video may have a few artifacts. Acceptable distribution of large and small vessels. About 30% to 50% of the vessels are looped. Most of the field of view has occluding artifacts such as saliva or bubbles. More than 50% of the vessels are looped upon themselves.

Stability: Frame motion that can be adequately stabilized without motion blur. Movement is within 1/4 of the field of view. No motion blur. Movement is within 1/2 of the field of view. No motion blur. Movement is greater than 1/2 of the field of view and/or motion blur in frame.

Pressure: Iatrogenic mechanical pressure causing misrepresentation of flow. Flow is constant throughout the entire movie. No obvious signs of artificially sluggish or stopped flow. Good flow in the largest vessels. Signs of pressure (localized sluggish flow in a specific large vessels), but flow appears to be unimpeded based on good flow in most large vessels. Obvious pressure artifacts associated with probe movement, and/or flow that starts and stops, reversal of flow. Poor or changing flow in larger venules.

Description	Good (0 points)	Acceptable (1 point)	Unacceptable (10 points)
Illumination:	Brightness and contrast of the video.	Even illumination across the entire field of view. Contrast sufficient to see small vessels against a background of tissue.	The video borders on being too dark or bright to distinguish vessels from tissue but the vessels are still identifiable.	The video is oversaturated/too bright or too dark to make out analyzable features. Insufficient contrast to resolve flow rate.
Focus:	Image sharpness in region of interest.	Good focus for all vessels (small and large) in the entire field of view. Plasma gaps and red blood cells are visible.	<1/2 field of view is out of focus or edges of the vessels are slightly out of focus.	Video is completely out of focus such that no small vessels can be seen.
Content:	Determination of the types of vessels and/or presence of occluding artifacts in the image. If the video presents overall pathology.	Video is free of occlusions. Good distribution of large and small vessels. Less than 30% of the vessels are looped upon themselves.	Video may have a few artifacts. Acceptable distribution of large and small vessels. About 30% to 50% of the vessels are looped.	Most of the field of view has occluding artifacts such as saliva or bubbles. More than 50% of the vessels are looped upon themselves.
Stability:	Frame motion that can be adequately stabilized without motion blur.	Movement is within 1/4 of the field of view. No motion blur.	Movement is within 1/2 of the field of view. No motion blur.	Movement is greater than 1/2 of the field of view and/or motion blur in frame.
Pressure:	Iatrogenic mechanical pressure causing misrepresentation of flow.	Flow is constant throughout the entire movie. No obvious signs of artificially sluggish or stopped flow. Good flow in the largest vessels.	Signs of pressure (localized sluggish flow in a specific large vessels), but flow appears to be unimpeded based on good flow in most large vessels.	Obvious pressure artifacts associated with probe movement, and/or flow that starts and stops, reversal of flow. Poor or changing flow in larger venules.

1. Massey MJ, Larochelle E, Najarro G, Karmacharla A, Arnold R, Trzeciak S, et al. The microcirculation image quality score: development and preliminary evaluation of a proposed approach to grading quality of image acquisition for bedside videomicroscopy. J Crit Care. 2013;28(6):913-7.

2.4 SDF-measurement and automatic video-analysis

The microcirculation was assessed with a sidestream darkfield microscope (MicroScan® device, MicroVision Medical®, Amsterdam, The Netherlands) as described before [8 –10]. On the tip of the device, a highly sensitive camera digitally records the sublingual capillary network. All videos were acquired in different regions under the tongue. At least four videos were taken per area. For video analysis, a tablet computer was used (Microsoft Surface Pro 4, (Redmond, Washington, USA). Once enough videos with sufficient quality had been recorded, a validated automatic algorithm-software (AVA, Version 4.3 C, MicroVision Medical®, Amsterdam, The Netherlands) performed the offline analysis.

Microvascular values provide information about both, density and perfusion capacities [13]: Density is reflected by the De Backer density (a grid-based score that provides the total number of vessels crossing per grid length) and the number of crossing (NC, number of vessels intersections the lines in a grid of 3 equidistant horizontal and vertical lines). The percentage of perfused vessels (PPV) offers information both about convexity and perfusion [14]. PPV was calculated after measuring the proportion of perfused vessels (PPV = 100 * (Total number of perfused vessels/total number of vessels). Both vessel density and perfusion can be combined for perfused DeBacker density (pdBD = the total number of perfused vessels crossing per grid length) and perfused number of crossings (PNC = number of vessel crossings with the continuous flow). Vessels with diameters less than 20 mm can be identified as capillaries. These small vessels are primarily responsible for the microcirculation. The small vessel values can be recognized by their prefix “s” (e.g. sPPV = PPV of small vessels). In this automated analysis, the values for all vessels can be considered as a quality check to exclude for example pressure artifacts [15]. AVA 4.3C is in line with the second consensus on the assessment of sublingual microcirculation in critically ill patients (ESICM) [7].

2.5 Statistics

Analyses were performed with Microsoft® Excel 2010 for Windows, the IBM Statistical Package for the Social Sciences (SPSS) 23.0 for Windows and Graph Pad Prism (Graph Pad Prism Software, Version 5, Graph Pad Software, San Diego, California, USA). The data were checked for normal distribution by the Shapiro-Wilk test. Normally distributed data are given in mean±standard deviation, non-normally distributed data are shown as median with inter-quartile-range. Categorical data are expressed as numbers (percentage). The statistical tests applied were the Mann Whitney test and the t-test, respectively. For paired groups and non-normally distributed data, the Wilcoxon signed-rank test was used. A 2-tailed p-value <0.05 was considered statistically significant.

3 Results

3.1 Participants characteristics

In total, 31 healthy volunteers were included (Fig. 1). Two participants reported chronic arterial hypertension but were sufficiently treated at the time of inclusion with antihypertensive medication. The majority was female sex. The characteristics are shown in Table 2.

Fig. 1

Study protocol.

Table 2

Participants characteristics (Mean±SD for normally distributed values, Median with Interquartile range for not normal distributed values), (n = 31)

Female sex [n (%)]	22 (61.1)
Age [years]	23.5 (±9)
Body weight [kg]	70.6 (±13.5)
Height [cm]	176.3 (±9.1)
BMI	22.6 (±3.1)
Hypertension [n (%)]	2 (5.6)

3.2 Equality of quality

In sum, 946 videos were recorded. In the first step, for each test person and each frame rate, the four best videos were chosen for further analysis. To these four best videos, the modified microcirculation image quality score was applied. 86 frames were evaluated with significantly lower score indicating a higher video quality for pressure (0.56 ± 0.25 for 86 frames versus 0.64 ± 0.23 for 215 frames, p < 0.05), stability (0.29 ± 0.24 for 86 frames versus 0.45 ± 0.32 for 215 frames, p < 0.05), and content (0.28 ± 0.18 for 86 frames versus 0.39 ± 0.30 for 215 frames, p = <0.001, see Fig. 2). Mean MIQS for 86 frames were significantly lower than MIQS for 215 frames (MIQS 1.8 ± 0.7 for 86 frames versus MIQS 2.2 ± 0.6 for 215 frames, p < 0.05).

Fig. 2

Quality scores according to Massey et al. for pressure, focus, stability, content, and brightness (Mean±SD) for 86 and 215 frames, respectively. ^∗= p < 0.05; ^∗∗= p < 0.01.

3.3 Equality of results

No test person showed an impaired microcirculation. The percentage of perfused small vessels was in every test person above 90%. Regarding the percentage of perfused vessels, there was no difference between AVA 4.3C results for 86 frames and 215 frames, neither for all nor for small vessels (94.9% ±3.0% for 86 frames versus 94.0% ±3.4% for 215 frames, p > 0.05, Fig. 3). Accordingly, there was no difference in the results for the other parameters that reflect perfusion (Number of perfused crossings and perfused De Backer Density, Figs. 4 and 5). The density parameters (Number of crossings, De Backer Density) showed no difference between 86 frames and 215 frames as well (Fig. 5).

Fig. 3

Percentage of perfused Vessels. A PPV (Mean±SD) for 86 and 215 frames, respectively. B sPPV (Mean±SD) for 86 and 215 frames, respectively.

Fig. 4

Consensus parameter for all vessels. A Number of crossings (Mean±SD) for 86 and 215 frames, respectively. B Number of perfused crossings (Mean±SD) for 86 and 215 frames, respectively. C De Backer Density (Mean±SD) for 86 and 215 frames, respectively. D Perfused De Backer Density (Mean±SD) for 86 and 215 frames, respectively.

Fig. 5

Consensus parameters for the small vessels. A Number of crossings (Mean±SD) for 86 and 215 frames, respectively. B Number of perfused crossings (Mean±SD) for 86 and 215 frames, respectively. C De Backer Density (Mean±SD) for 86 and 215 frames, respectively. D Perfused De Backer Density (Mean±SD) for 86 and 215 frames, respectively.

3.4 Examination speed

Although the study protocol was not designed as a competition, the time to acquire enough videos for at least four videos with a suitable quality was measured. There was no difference between both approaches, neither regarding the time needed for video-acquisition (7.7 Minutes±4.1 Minutes versus 8.0 Minutes±4.7 Minutes, p = 0.82) nor the software algorithm (duration per video: 4.4 Minutes±0.5 Minutes versus 4.7 Minutes±0.5 Minutes, p = 0.15, Fig. 6).

Fig. 6

Compared duration for video acquisition and analysis. A Duration of video acquisition until at least four videos with sufficient quality were recorded [minutes] (Mean±SD) for 86 and 215 frames, respectively (n = 32). B Duration of the AVA 4.3C analysis per video [seconds] (Mean±SD) for 86 and 215 frames, respectively (n = 20).

3.5 Safety

No adverse events during the procedure had been reported.

4 Discussion

In modern intensive and emergency medicine, a fast, reliable, and accurate method for outcome prediction and risk stratification is urgently needed. The present study demonstrates that for the automatic analysis of microcirculatory imaging, 86 frames offer equal quality and equal measurement results compared to 215 frames. However, without further software improvements, this observation does not result in an advantage in terms of examination speed. Currently, the automatic evaluation of the videos takes up the most time during the entire examination process. A recent study that used sublingual microcirculatory assessment in acute prehospital emergencies identified the time as the most crucial element limiting its value in daily practice [10].

Patients in the intensive care unit still suffer from high mortality despite many improvements during the last decade. Therefore, sufficient monitoring of these critically ill patients in intensive care units is of the highest importance for an early risk prediction and treatment optimization [16]. Several laboratory parameters have successfully been evaluated for risk prediction in this context such as blood urea nitrogen [17], acidosis [18], glycocalyx markers (e.g. syndecan-1) [19], or the combination of values (e.g. lactate/albumin ratio) [20]. For pathophysiology, microcirculation is the compartment that determines the prognosis of the patient [21]. Thus, in daily intensive medicine, serum lactate levels are used to assess the microcirculation. Increased levels of lactate are independently associated with increased mortality and microvascular flow abnormalities [22].

The in vivo measurement of microcirculation has been shown to precede increasing lactate [23] and thus has the power to essentially improve critical care therapy. For example, several small studies used the sublingual microcirculation both for outcome prediction [24] and weaning in patients undergoing extracorporeal membrane oxygenation [25, 26]. Recently, sublingual microcirculation has successfully been used to identify patients who profit from a blood transfusion –regardless of the hemoglobin level [27]. In acute severe heart failure, the administration of angiotensin inhibitors was associated with an independent increase in the proportion of perfused small vessels [28]. However, all these approaches have still not made the leap from a scientific bystander to a therapy target [29].

In the past ten years, several attempts have been made to improve the in vivo visualization of sublingual microcirculation. In addition to the MicroScan (SDF) camera, the Braedius Cytocam, an Incident Dark Field (IDF) video microscope, has been introduced [30]. Some smaller studies found a superior image quality of IDF compared to SDF [31 –33], but the clinical significance remains unclear. Recently, an algorithm-based analysis software (MicroTools) has been created for this device [34]. Furthermore, a new MicroScan that uses USB3 has been developed. This improvement resulted in a significant increase in quality compared to its predecessor (MicroScan Analogue) [35]. Still, neither of both cameras has made the leap into daily practice. A recent study in septic patients compared another automatic algorithm for the IDF camera (CCtools®) to semi-automatic AVA 3.0 [36]. They found that the microcirculation parameters from the automated software were coherent with previous reported semi-quantitative significant numeric differences. However, Carsetti et al. performed a similar comparison between CCtools® (CytoCamTools 1.7.12) and AVA 3.2. This study examined the sublingual microcirculation in patients who underwent cardiothoracic surgery. CCtools® was significantly faster than AVA 3.2. While the total vessel density was comparable, perfusion data differed significantly between both software [37].

For the assessment of sublingual microcirculation, it is important to keep in mind that the inexperienced user must learn two different procedures: First, the basic technique to acquire videos with sufficient quality, and second how to interpret the videos. Both procedures must be learned separately. For the video interpretation, there exist two different approaches: First, the visual and manual interpretation can be standardized into scores such as the POEM-Score (point-of-care microcirculation) [38, 39]. In brief, Naumann et al. developed this score as a point-of-care tool. For this approach, the videos must be assessed manually. The main element of the POEM is the composite assessment of flow and heterogeneity resulting in an ordinal scale from 1 (worst) to 5 (best). This technique must be taught to the user and remains a subjective judgment.

The second approach is the automated analysis by a software algorithm. The main advantage is that the user must learn only the video acquisition technique. Once videos with sufficient quality are recorded, the algorithm provides objective user-independent parameters. Now, AVA 4.3C does not offer information about the flow in terms of heterogeneity. However, the value of microcirculatory flow interpretation such as the MFI (microvascular flow index) is debatable as the MFI failed in the routine daily microcirculatory monitoring as it did not provide any additional information [40].

Currently, two large international multicenter studies are ongoing to evaluate the value of immediate and repetitive sublingual microcirculatory assessment with AVA 4.3C in shock (NCT04173221; NCT04169204).

However, without further software improvements, time remains one of the most important barriers that deter sublingual microcirculatory assessment from daily clinical practice.

4.1 Limitations

This study was not designed as a competition trial in terms of speed between two observers. Due to the short video duration of the 86 frames, no visual control with the POEM was performed. This study investigated only healthy individuals.

4.2 Conclusion

86 frames are not inferior to 215 frames and offer equal microcirculatory parameters. However, there is no advantage regarding the time needed for the measurement procedure. Thus, improved software for analysis is urgently needed that benefits from these shorter frames to calculate the parameters faster.

Conflict of interest

The authors have no conflict of interest to report.

References

Jung

. Assessment of microcirculation in cardiogenic shock. Curr Opin Crit Care. 2019;25(4):410–6.

Masyuk

, Wernly

, Lichtenauer

, Franz

, Kabisch

, Muessig

, et al.Prognostic relevance of serum lactate kinetics in critically ill patients. Intensive Care Med. 2019;45(1):55–61.

Pan

, Su

, Liu

, Wang

. Microcirculation-guided protection strategy in hemodynamic treatment. Clin Hemorheol Microcirc. 2020. DOI: 10.3233/CH-190784

Martini

. The compelling arguments for the need of microvascular investigation in COVID-19 critical patients. Clin Hemorheol Microcirc. 2020;75(1):27–34.

Sharawy

, Mahrous

, Whynot

, George

, Lehmann

. Clinical relevance of early sublingual microcirculation monitoring in septic shock patients. Clin Hemorheol Microcirc. 2018;68(4):347–59.

Sardinha

, MacKinnon

, Lehmann

. Rapid clinical assessment of the sublingual microcirculation - Visual scoring using microVAS in comparison to standard semi-automated analysis. Clin Hemorheol Microcirc.. 2019;72(3):229–38.

Ince

, Boerma

, Cecconi

, De Backer

, Shapiro

, Duranteau

, et al.Second consensus on the assessment of sublingual microcirculation in critically ill patients: results from a task force of the European Society of Intensive Care Medicine. Intensive Care Med. 2018;44(3):281–99.

Bimpong-Buta

, Jirak

, Wernly

, Lichtenauer

, Masyuk

, Muessig

, et al.Analysis of human microcirculation in weightlessness: Study protocol and pre-study experiments. Clin Hemorheol Microcirc. 2018;70(1):119–27.

Bruno

, Masyuk

, Muessig

, Binneboessel

, Bernhard

, Baz

, et al. Sublingual microcirculation in prehospital critical care medicine: A proof-of-concept study. Microcirculation. 2020.

10.

Bruno

, Reed

, Bimpong-Buta

, Muessig

, Masyuk

, Binneboessel

, et al. Sublingual microcirculation in prehospital critical care medicine: A proof-of-concept study. Microcirculation. 2020.

11.

Pistulli

, Quitter

, Andreas

, Rohm

, Kretzschmar

, Figulla

, et al.Intravital microscopy - A novel tool in characterizing congestive heart failure in experimental autoimmune myocarditis. Clin Hemorheol Microcirc. 2015;63(2):153–62.

12.

Massey

, Larochelle

, Najarro

, Karmacharla

, Arnold

, Trzeciak

, et al.The microcirculation image quality score: development and preliminary evaluation of a proposed approach to grading quality of image acquisition for bedside videomicroscopy. J Crit Care. 2013;28(6):913–7.

13.

Massey

, Shapiro

. A guide to human in vivo microcirculatory flow image analysis. Crit Care. 2016;20:35.

14.

Vellinga

, Boerma

, Koopmans

, Donati

, Dubin

, Shapiro

, et al.International study on microcirculatory shock occurrence in acutely ill patients. Crit Care Med. 2015;43(1):48–56.

15.

Rorije

NMG

, Rademaker

, Schrooten

, Wouda

, Homan Van Der Heide

, Van Den Born

, et al.High-salt intake affects sublingual microcirculation and is linked to body weight change in healthy volunteers: a randomized cross-over trial. J Hypertens. 2019;37(6):1254–61.

16.

Jung

, Kelm

. Evaluation of the microcirculation in critically ill patients. Clin Hemorheol Microcirc.. 2015;61(2):213–24.

17.

Wernly

, Lichtenauer

, Vellinga

NAR

, Boerma

, Ince

, Kelm

, et al.Blood urea nitrogen (BUN) independently predicts mortality in critically ill patients admitted to ICU: A multicenter study. Clin Hemorheol Microcirc. 2018;69(1-2):123–31.

18.

Wernly

, Heramvand

, Masyuk

, Rezar

, Bruno

, Kelm

, et al.Acidosis predicts mortality independently from hyperlactatemia in patients with sepsis. Eur J Intern Med. 2020;76:76–81.

19.

Wernly

, Fuernau

, Masyuk

, Muessig

, Pfeiler

, Bruno

, et al.Syndecan-1 Predicts Outcome in Patients with ST-Segment Elevation Infarction Independent from Infarct-related Myocardial Injury. Sci Rep. 2019;9(1):18367.

20.

Lichtenauer

, Wernly

, Ohnewein

, Franz

, Kabisch

, Muessig

, et al.The Lactate/Albumin Ratio: A Valuable Tool for Risk Stratification in Septic Patients Admitted to ICU. Int J Mol Sci. 2017;18(9).

21.

Jung

, Jung

, Kelm

. The microcirculation in hypoxia: The center of the battlefield for oxygen. Clin Hemorheol Microcirc. 2016;63(3):169–72.

22.

Vellinga

NAR

, Boerma

, Koopmans

, Donati

, Dubin

, Shapiro

, et al.Mildly elevated lactate levels are associated with microcirculatory flow abnormalities and increased mortality: a microSOAP post hoc analysis. Crit Care. 2017;21(1):255.

23.

Jung

, Fuernau

, de Waha

, Eitel

, Desch

, Schuler

, et al.Intraaortic balloon counterpulsation and microcirculation in cardiogenic shock complicating myocardial infarction: an IABP-SHOCK II substudy. Clin Res Cardiol. 2015;104(8):679–87.

24.

Yeh

, Lee

, Wang

, Tu

, Lai

, Wang

, et al.Investigation of microcirculation in patients with venoarterial extracorporeal membrane oxygenation life support. Crit Care. 2018;22(1):200.

25.

Kara

, Akin

, Dos Reis Miranda

, Struijs

, Caliskan

, van Thiel

, et al.Microcirculatory assessment of patients under VA-ECMO. Crit Care. 2016;20(1):344.

26.

Akin

, Dos Reis Miranda

, Caliskan

, Soliman

, Guven

, Struijs

, et al.Functional evaluation of sublingual microcirculation indicates successful weaning from VA-ECMO in cardiogenic shock. Crit Care. 2017;21(1):265.

27.

Scheuzger

, Zehnder

, Meier

, Yeginsoy

, Flukiger

, Siegemund

. Sublingual microcirculation does not reflect red blood cell transfusion thresholds in the intensive care unit-a prospective observational study in the intensive care unit. Crit Care. 2020;24(1):18.

28.

Salgado

, Favory

, Rocco

, Silva

, Ortiz

, Donadello

, et al.Microcirculatory effects of angiotensin II inhibitors in patients with severe heart failure. Clin Hemorheol Microcirc. 2013;54(1):87–98.

29.

Jung

, Lauten

, Ferrari

. Microcirculation in cardiogenic shock: from scientific bystander to therapy target. Crit Care. 2010;14(5):193.

30.

Hutchings

, Watts

, Kirkman

. The Cytocam video microscope. A new method for visualising the microcirculation using Incident Dark Field technology. Clin Hemorheol Microcirc. 2016;62(3):261–71.

31.

Gilbert-Kawai

, Coppel

, Bountziouka

, Ince

, Martin

, Caudwell Xtreme

, et al.A comparison of the quality of image acquisition between the incident dark field and sidestream dark field video-microscopes. BMC Med Imaging. 2016;16:10.

32.

van Elteren

, Ince

, Tibboel

, Reiss

, de Jonge

. Cutaneous microcirculation in preterm neonates: comparison between sidestream dark field (SDF) and incident dark field (IDF) imaging. J Clin Monit Comput. 2015;29(5):543–8.

33.

Aykut

, Veenstra

, Scorcella

, Ince

, Boerma

. Cytocam-IDF (incident dark field illumination) imaging for bedside monitoring of the microcirculation. Intensive Care Med Exp. 2015;3(1):40.

34.

Hilty

, Guerci

, Ince

, Toraman

, Ince

. MicroTools enables automated quantification of capillary density and red blood cell velocity in handheld vital microscopy. Commun Biol. 2019;2:217.

35.

Coppel

, Bountziouka

, Martin

, Gilbert-Kawai

. A comparison of the quality of image acquisition between two different sidestream dark field video-microscopes. J Clin Monit Comput. 2020.

36.

Sharawy

, Mukhtar

, Islam

, Mahrous

, Mohamed

, Ali

, et al.Preliminary clinical evaluation of automated analysis of the sublingual microcirculation in the assessment of patients with septic shock: Comparison of automated versus semi-automated software. Clin Hemorheol Microcirc. 2017;67(3-4):489–98.

37.

Carsetti

, Aya

, Pierantozzi

, Bazurro

, Donati

, Rhodes

, et al.Ability and efficiency of an automatic analysis software to measure microvascular parameters. J Clin Monit Comput. 2017;31(4):669–76.

38.

Naumann

, Hazeldine

, Bishop

, Midwinter

, Harrison

, Nash

, et al.Impact of plasma viscosity on microcirculatory flow after traumatic haemorrhagic shock: A prospective observational study. Clin Hemorheol Microcirc. 2019;71(1):71–82.

39.

Watchorn

, Fargaly

, Gilani

, Assadi

, Deitchman

, Naumann

, et al. The Reproducibility of the Point of care Microcirculation (Poem) Scorewhen used to Assess Critically Ill Patients: A Multicenter Prospective Observational Study. Shock. 2019.

40.

Scorcella

, Damiani

, Domizi

, Pierantozzi

, Tondi

, Carsetti

, et al.MicroDAIMON study: Microcirculatory DAIly MONitoring in critically ill patients: a prospective observational study. Ann Intensive Care. 2018;8(1):64.