The Cytocam video microscope. A new method for visualising the microcirculation using Incident Dark Field technology

Abstract

We report a new microcirculatory assessment device, the Braedius Cytocam, an Incident Dark Field (IDF) video microscope, and compare it with a precursor device utilising side stream dark field (SDF) imaging.

METHODS:

Time matched measurements were made with both devices from the sublingual microcirculation of pigs subjected to traumatic injury and hemorrhagic shock at baseline and during a shock phase. Images were analysed for vessel density, microcirculatory flow and image quality.

RESULTS:

There were no differences in density or flow data recorded from the two devices at baseline [TVD IDF 14.2 ± 2.4/TVD SDF 13.2 ± 2.0, p 0.17] [MFI IDF 3 (2.8–3.0)/MFI SDF 3 (2.9–3.0), p 0.36] or during the shock state [TVD IDF 11.64 ± 3.3/TVD SDF 11.4 ± 4.0 p = 0.98] [MFI IDF 1.9 (0.6–2.7)/MFI SDF 1.7 (0.3–2.6) p 0.55]. Bland and Altman analysis showed no evidence of significant bias. Vessel contrast was significantly better with the IDF device for both capillaries [17.1 ± 3.9 (IDF) v 3.4 ± 3.6 (SDF), p = 0.0006] and venules [36.1 ± 11.4 (IDF) v 26.4 ± 7.1 (SDF) p 0.014]

CONCLUSION:

The Braedius Cytocam showed comparable vessel detection to a precursor device during both baseline and low flow (shock) states.

Keywords

Microcirculation sidestream dark field imaging incident dark field imaging haemorrhagic shock

1 Introduction

The microcirculation plays a critical role in the delivery of oxygen and substrates at a cellular level. Microcirculatory impairment can lead to significantly increased morbidity and mortality across a range of shock states [9 , 14] and significant impairment to microcirculatory flow may be present in around a fifth of unselected patients within critical care units [15]. Assessment of the microcirculation can occur through two broad and separate approaches; the first approach involves assessment of tissue substrate delivery and uptake and the second relies on direct visualisation of microcirculatory vessels in order to determine flow and density characteristics. Examples of the former approach include Near Infra Red Spectroscopy (NIRS) [10] and fluorescence quench photometry [6] which provide an estimation of tissue oxygenation and sublingual capnometry which, by assessing the degree of carbon dioxide production in an individual tissue bed, provides a surrogate marker of perfusion [2, 16]. These methods produce an aggregated overview of microcirculatory performance but may be influenced by the heterogeneity that is often found in microcirculatory vessel beds. The second broad method of microcirculatory assessment relies on the direct visualisation of small vessels (arterioles, venules and capillaries) at various sites throughout the body. Historically this required the use of large microscopes and tissue dyes, so called intra vital microscopy, and was therefore mainly limited to experimental work in small animal models. The development of handheld microscopes has revolutionised this area of research. The first generation handheld devices used Orthogonal Polarised Spectroscopy (OPS) imaging [5] which produced a somewhat limited field of view, whilst the devices were relatively bulky. The next generation devices, used a Sidestream Dark Field (SDF) technique [4]. Similar to OPS in theory, SDF technology involves illuminating the edges of the examined field with visible green light in an 530 nm wavelength from a circular array of diodes. The light illuminates the target tissue from the edges, whilst the field itself is excluded from external light. Haemoglobin (both oxygenated and deoxygenated) absorbs the green light and thus appears black. Thus perfused blood vessels appear as black lines on a white/greyscale background. Lighter gaps within blood vessels are caused by plasma or by leucocytes. Producing qualitative data from SDF images requires editing and processing into discrete sequences followed by off line analysis. Such analysis produces data on vessel density and flow. At its most basic this analysis can be entirely manual consisting of a subjective assessment of flow using a qualitative scale. More commonly semi automated software is used to assist in the process, but a significant amount of user interaction is still required and each ten to twenty second video sequence can take up to thirty minutes to process, even with an experienced analyst. Given the widespread heterogeneity seen within the microcirculation during shock states it is important to obtain a large number of video sequences, ideally five, at each time point being examined, thus increasing the amount of video sequences that need to be analysed. This prolonged off line analysis has, thus far, limited the use of this technique to research rather than as a clinical point of care perfusion test.

The latest generation of optical device for assessing the microcirculation is the Braedius Cytocam (Fig. 1), which utilises an Incident Dark Field (IDF) technique. This technique, originally described by Sherman and colleagues in the 1970s [12] is similar to Sidestream dark field illumination, in that it uses a ring of circumferential LEDs to illuminate the target tissue tangentially, the illuminating light being excluded from the central column of the microscope. However, the LED strobe speed is significantly slower in the Cytocam device (2 ms versus 16 ms); this theoretically produces less distortion of the erythrocyte image potentially allowing more accurate automated flow analysis. The Cytocam is a small handheld camera, which weighs considerably less than the previous generation Microscan SDF device (Fig. 2) (115 versus 350 grams). It is also less bulky than the SDF device and this has the effect of making it easier for the operator to manipulate (Fig. 3). In turn this has the potential to reduce pressure artifacts caused by the tip of the device occluding flow in microcirculatory vessels. The Cytocam has a higher optical resolution than the Microscan (3.12 versus 4.54 micrometers) and a wider field of view (1.79 mm² versus 0.84 mm²). This larger field of view enables the user to more quickly identify suitable areas of microcirculation for analysis. The Cytocam has an electronically driven focusing motor as opposed to the analogue focusing mechanism of the Microscan making it easier to carry out fine focusing actions. Additionally the device returns to the same focus depth for subsequent measurements, speeding up the time taken to acquire images. The Cytocam is linked to a dedicated computer with integral software which controls image acquisition, editing and potentially on line analysis. This is in contrast to the previous generation devices which required analogue to digital signal conversion through a signal adapter as well as bulky separate batteries and a non proprietary laptop computer with adapted software.

In this paper we report a series of experiments designed to assess the performance of the Cytocam IDF device and to compare and contrast it to it’s immediate predecessor, the Microscan SDF device. In the first part of the study, the Cytocam is directly compared with the Microscan in an existing large animal model of haemorrhagic shock and complex injury. The aim is to compare the performance of the devices with a particular focus on vessel identification. The second part of study is a direct comparison of image quality between the two devices, using matched sublingual microcirculatory images, obtained from one of the study investigators.

2 Methods

Conduct of this research was approved under license from the United Kingdom Home Office: Animals (Scientific Procedures Act) 1986.

2.1 Subjects

Briefly, six terminally anaesthetised and surgically prepared large white pigs, were subjected to a controlled haemorrhage and reproducible extremity injury and, in some instances, blast wave exposure. Following a 30 minute shock phase the animals were resuscitated using differing resuscitation strategies that are outside the scope of this paper, but which have been reported elsewhere [7, 8]. The experiment ended at four hours after the commencement of resuscitation at which point the animals were euthanised.

The aspect of the study examining image contrast required two closely matched microcirculatory images, obtained from the subject devices. In practice, this was not possible within the constraints of our existing experimental setup. We therefore obtained these images from the sublingual region of one of the investigators (SDH).

2.2 Vessel identification and flow assessment

Two devices were used to record video images of the sub lingual microcirculation, utilising Incident Dark Field (IDF) technology (Cytocam, Braedius Medical, Amsterdam, NL) or Sidestream Dark Field (SDF) technology (Microscan, Microvision Medical, Amsterdam, NL).

Serial video images of the sublingual microcirculation were recorded at the following experimental time points: Baseline (a steady state time point prior to injury or blood loss) and Shock (15 minutes after extremity trauma, controlled blood loss and blast wave injury). The shock time point represents a low flow state, with a degree of dynamic response within the microcirculation in contrast to the predictable steady state seen at baseline. At each time point at least three and ideally five ten second sequences were recorded using each device, in accordance with accepted consensus opinion on assessment of the microcirculation [3]. Images were recorded using the SDF device (Microscan), immediately followed by the IDF device (Cytocam). Images were either directly recorded onto a dedicated integral panel PC using integral software (Cytocam Tools v. 7.1, Braedius Medical) or via an analogue to digital signal converter (Canopus, ADVC 110) onto a stand alone laptop computer (Microscan). Microscan acquired images were saved as digital DV-AVI files, whilst those acquired by the Cytocam were converted to this format during export using integral software. As previously mentioned the field of view from the Cytocam is considerably wider than that obtained using the Microscan necessitating a reduction in the field of view during export to DV-AVI format. Video captures were all performed by a single operator. All images were recorded from the same part of the sublingual area, to the left side of the midline. Care was taken to minimise pressure artefact, by applying the probe tip and pulling back until contact was just lost and then reapplying with the minimum amount of pressure required to produce an image. Focus was adjusted by the operator in order to produce the best possible vessel definition, either manually (Microscan) or via the integral software linked to the electronic focusing mechanism (Cytocam). Videos were assigned a five digit random number prior to archiving in order to reduce bias during interpretation. The videos were linked to study time point and device by reference to a separate database. Videos were exported in DV-AVI format and with standardised characteristics. Video files were saved on an external hard drive prior to off line analysis.

Analysis of video sequences was carried out by a single operator using dedicated software (AVA 3.0, Automated Vascular Analysis, Microvision Medical, NL). The operator was blinded to both the device used for recording and the time point of the video sequence. Only video sequences that conformed to pre determined standards of stability, focus, illumination, length and absence of pressure artefacts were included in the analysis. These standards were detailed by Massey et al. in a previous study [11]. All visible vessels were manually traced by the operator and flow characteristics were assigned to each individual vessel segment, using a four point scale originally described by Boerma et al. [1]. The mean flow across all segments was calculated in order to produce an overall Microcirculatory Flow Index (MFI) for each video sequence. Analysis of each video sequence produced data for the total number of vessel segments compared to the size of the analysis area (Total Vessel Density, TVD).

2.3 Vessel contrast assessment

Vessel contrast and sharpness was analysed using a technique originally described by Goedhart etl al. [4] To compare vessel contrast between the two devices a single operator obtained two near identical images from their own sublingual microcirculation, using the two subject devices. Optimal image characteristics were obtained with respect to focus, illumination and avoidance of pressure. Video sequences were examined and one video frame, demonstrating the highest quality of focus and illumination, was selected for each device. This video frame was saved as a Portable Network Graphic (PNG) file. In each video frame image, ten venule segments and ten capillary segments were selected for quality analysis. (Fig. 4) To determine vessel contrast and sharpness, cross sectional grey scale profiles (greyscale value 0 corresponding to black and 255 to white) were obtained using Image J software (freeware developed by the US National Institute for Health). The contrast was defined as the absolute difference between the minimum value within the vessel and the maximum value on either side of the vessel wall (average of the two sides of the vessel).

2.4 Statistical analysis

Statistical Analysis was performed using GraphPad Prism version 6. Data was tested for normality using D’Agostino & Pearson omnibus normality test. Students paired t tests were used to compare vessel density data which was normally distributed. Mann Whitney U tests were used to compare vessel flow data, which did not have a normal distribution. Bland – Altman distribution analysis was performed on the TVD results from the two devices. Contrast values for the two devices were analysed using paired t tests after confirming normal distribution. A p value of < 0.05 was considered to be statistically significant.

3 Results

3.1 Vessel identification and flow assessment

Results from six animals were included in the study. Forty four video sequences were obtained with each device, giving a total of eighty eight matched video sequences. In order to assess for possible variance at different flow states video sequences were subdivided into Baseline (n = 40) and Shock (n = 48) time points. As expected there was a wider spread of results during the shock time point as the microcirculation demonstrated increased heterogeneity. By contrast the Baseline time point was more homogenous representing a relatively steady state at this stage of the experiment. One experimental time point was excluded as at least three video sequences of sufficient quality could not be obtained.

There were no significant differences in vessel detection, expressed as the Total Vessel Density, recorded by the two devices at either the baseline or shock time points (Table 1).

There was also no difference in the observed Microvascular Flow Index, recorded by the two devices at either experimental time point. These findings are consistent when the data is aggregated (Fig. 5) or examined for each individual animal (Table 2).

Bland Altman analysis confirms the comparability of data collection from each device and shows no evidence of bias (Fig. 6).

3.2 Vessel contrast and sharpness

Vessel contrast was significantly higher in the image obtained from the Cytocam IDF device. This applied to both capillary and venules (Table 3, Fig. 4).

4 Discussion

We have described a series of experiments designed to assess the performance of a new device for assessing the microcirculation, the Braedius Cytocam. To our knowledge this is the first such evaluation of the Cytocam IDF device performed over a range of flow states, in an experimental model of shock. Our experimental setup was designed to replicate as closely as possible conditions found within a clinical environment. We feel that this adds considerable value to this study as it enables a comparison of the two imaging devices under conditions that closely replicate that which will be encountered in clinical practice. The inclusion of imaging comparisons at low flow shock states is often omitted from validation studies, which are usually conducted on healthy volunteers under controlled conditions. By including these low flow images in this study we have produced a comparison between the two devices over the full operating range in which they would be used in clinical practice.

In the first part of the study we showed that, in the pragmatic setting of an existing trial, the two devices performed in a comparable fashion and the data obtained, with respect to both vessel density and flow, was virtually indistinguishable. By selecting a baseline steady state timing point we attempted to reduce any changes in the microcirculation seen between the recording of the SDF and the IDF acquired images. The addition of the shock timing point is important because it allows a comparison of the two devices at a low flow state, where vessel identification may be predicted to be more difficult. However, at this time point the microcirculation was in a state of flux, induced by haemorrhage and injury, and this can be seen in the much wider spread of data. Despite this dynamism and heterogeneity there is still a remarkable similarity between the results collected from the two devices. As far as possible we attempted to reduce the time between image acquisition, which in practice was achieved within five minutes from the start of SDF collection to the end of IDF collection.

We acknowledge that this approach has limitations and that it would be optimal to compare identical images derived from the two devices. However, in practical terms it is extremely difficult to find even two such images, given the magnification and field of view constraints as well as the instabilityinherent in the use of handheld cameras. In reality it is impossible to obtain matched images in a dynamic study that includes animals or patients with shocked microcirculations. The first part of our study therefore represents an attempt to answer the question of comparability of data generated by the two devices, rather than a completely objective assessment of image identification under perfect operating conditions.

In the second part of this study we examined whether the image quality produced by the IDF device was superior to that of the SDF device. Subjectively, we felt that the IDF images were sharper and had greater definition than the precursor SDF images and were able to confirm this objectively on one matched image using pixel analysis software. Further studies are needed to replicate this finding which is of potential importance as sharper images and improved contrast between blood vessels and background tissue potentially make the automated analysis of such images by computer software more achievable. Automated analysis programs cannot yet reliably measure vessel density in recorded video-microscopy images and improvements in image quality are a necessary prerequisite for the development of this technology.

In summary the Braedius Cytocam is a new device for imaging the microcirculation which we have validated against a common precursor device. Microcirculatory data produced by the two devices is comparable and this should allow researchers to confidently transition to the new device whilst maintaining the integrity of data previously recorded with precursor technology.

Conflicts of interest

None declared.

Footnotes

Acknowledgments

Prof. Can Ince, University of Amsterdam for valuable assistance and advice over the course of this research.

Mr. Frank Messie for assistance with the technical aspects of device comparison.

References

Boerma

E.C.

, Mathura

K.R.

, van der Voort

P.H.J.

, Spronk

P.E.

and Ince

, Quantifying bedside-derived imaging of microcirculatory abnormalities in septic patients: A prospective validation study,R601–R, Crit Care 9(6) (2005), 606.

Creteur

, De Backer

, Sakr

, Koch

and Vincent

J.-L.

, Sublingual capnometry tracks microcirculatory changes in septic patients, Intensive Care Med 32(4) (2006), 516–523.

De Backer

, Hollenberg

, Boerma

, et al. How to evaluate the microcirculation: Report of a round table conference, Crit Care (11) (2007), R101.

Goedhart

P.T.

, Khalilzada

, Bezemer

, Merza

and Ince

, Sidestream Dark Field (SDF) imaging: A novel stroboscopic LED ring-based imaging modality for clinical assessment of the microcirculation, Opt Express 15(23) (2007), 15101–15114.

Groner

, Winkelman

J.W.

, Harris

A.G.

, et al., Orthogonal polarization spectral imaging: A new method for study of the microcirculation, Nat Med 199(5), 1209–1212.

Guerci

, Tran

, Menu

, et al., Impact of fluid resuscitation with hypertonic-hydroxyethyl starch versus lactated ringer on hemorheology and microcirculation in hemorrhagic shock, Clinical Hemorheology and Microcirculation 56 (2014), 301–317.

Hutchings

S.D.

, Wendon

, Watts

and Kirkman

, Assessing the microcirculation using Sidestream Dark Field video microscopy in a porcine model of complex haemorrhagic shock and traumatic injury, Intensive Care Med S2(39) (2013), S436–S437.

Hutchings

S.D.

, Wendon

, Watts

and Kirkman

, Microcirculatory and macrocirculatory responses in a porcine model of traumatic haemorrhagic shock and resuscitation, Br J Anaesth 112(1) (2014), 185–186P.

Jhanji

, Lee

, Watson

, Hinds

and Pearse

R.M.

, Microvascular flow and tissue oxygenation after major abdominal surgery: Association with post-operative complications, Intensive Care Med 35(4) (2009), 671–677.

10.

Lipcsey

, Woinarski

N.C.

and Bellomo

, Near infrared spectroscopy (NIRS) of the thenar eminence in anesthesia and intensive care, Ann Intensive Care 2(1) (2012), 11.

11.

Massey

M.J.

, Larochelle

, Najarro

, 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 28(6) (2013), 913–917.

12.

Sherman

, Klausner

and Cook

W.A.

, Incident dark-field illumination: A new method for microcirculatory study, Angiology 22(5) (1971), 295–303.

13.

Spanos

, Jhanji

, Vivian-Smith

, Harris

and Pearse

R.M.

, Early microvascular changes in sepsis and severe sepsis, Shock 33(4) (2010), 387–391.

14.

Tachon

, Harrois

, Tanaka

, et al., Microcirculatory alterations in traumatic hemorrhagic shock, Critical Care Medicine 42(6) (2014), 1433–1441.

15.

Vellinga

N.A.

and Boerma

E.C.

, Koopmans M for the microSOAP investigators. International Study on Microcirculatory Shock Occurrence in Acutely Ill Patients, Critical Care Medicine 43 (2015), 58–56.

16.

J.J.

, Ma

L.L.

, Sun

S.S.

, et al., Fluid resuscitation guided by sublingual partial pressure of carbon dioxide during hemorrhagic shock in a porcine model, Shock 39(4) (2013), 361–365.