Guideline for In Vivo Assessment of Adherent and Rolling Leukocytes in Human Skin Microvasculature via Reflectance Confocal Videomicroscopy

Abstract

Objective:

To develop a guideline that reliably identifies cutaneous adherent and rolling leukocytes from mimicking scenarios via in vivo reflectance confocal videomicroscopy.

Methods:

We used a clinical reflectance confocal microscope, the VivaScope 1500, to acquire 1522 videos of the upper dermal microcirculation from 12 healthy subjects and 60 patients after allogeneic hematopoietic cell transplantation. Blinded to clinical information, two trained raters independently counted the number of adherent and rolling leukocytes in 88 videos. Based on discrepancies in the initial assessments, we developed a guideline to identify both types of leukocyte-endothelial interactions via a modified Delphi method (without anonymity). To test the guideline’s ability to improve the inter-rater reliability, the two raters assessed the remaining 1434 videos by using the guideline.

Results:

We demonstrate a guideline that consists of definitions, a step-by-step flowchart, and corresponding visuals of adherent and rolling leukocytes and mimicking scenarios. The guideline improved the inter-rater reliability of the manual assessment of both interactions. The intraclass correlation coefficient (ICC) of adherent leukocyte counts increased from 0.056 (95% confidence interval: 0-0.236, n=88 videos, N=10 subjects) to 0.791 (0.770-0.809, n=1434, N=67). The ICC of rolling leukocyte counts increased from 0.385 (0.191-0.550, n=88, N=10) to 0.626 (0.593-0.657, n=1434, N=67). Intra-rater ICC post-guideline was 0.953 (0.886-0.981, n=20, N=12) and 0.956 (0.894-0.983, n=20, N=12) for adherent and rolling, respectively.

Conclusion:

The guideline aids in the manual identification of adherent and rolling leukocytes via in vivo reflectance confocal videomicroscopy.

INTRODUCTION

Adherent and rolling leukocyte counts are altered in a variety of inflammatory diseases.^1–7 Although in vivo leukocyte-endothelial interactions have been well explored in animal models, studies in human microcirculation remain limited. Noninvasive imaging of leukocytes in the sublingual mucosa^1–3 have demonstrated prognostic value in septic shock.¹ These potential clinical biomarkers have also been explored in conjunctiva.⁷ Leukocyte motion in cutaneous microcirculation was first visualized in 1995 via reflectance confocal microscopy (RCM).⁸ However, there has been a lack of descriptive and clinical studies further exploring these interactions. We previously characterized individual cell motion in healthy cutaneous microvasculature by RCM via nine quantitative parameters, including the number of adherent and rolling leukocytes.⁹ We visualized adherent and rolling leukocytes in acute graft-versus-host disease (GVHD)⁴ but only adherent leukocytes in healthy cutaneous microvasculature.⁹ However, there is a lack of consensus on the manual identification of adherent and rolling leukocytes in RCM videos.

RCM enables noninvasive, direct visualization of the tissue at a cellular level. FDA-approved systems (i.e., Vivascope 1500 and 3000) acquire 0.5x0.5 to 0.75x0.75 mm² videos at 6-9 frames per second and up to ~200 μm deep into the skin. The rapid image acquisition allows real-time visualization of individual cell movement in the upper dermal microvasculature, including leukocyte-endothelial interactions. However, due to limited imaging depth and reduced image quality with increasing depth,^10,11 individual cells cannot be clearly visualized in vessels deeper than the upper dermis. Additional challenges include limited visualization of the blood vessel wall⁹ and motion artifacts.¹²

The leukocyte inflammatory cascade has been well-studied in animals and consists of five main steps: rolling, adhesion, crawling, passage through the endothelium, retainment, and passage through the basement membrane.^13–19 Rolling leukocytes have low-affinity adhesion to the endothelium and undergo rotational motion due to the force of the blood flow.²⁰ After rolling, leukocytes can arrest on the endothelium.¹⁶ Adherent leukocytes have high-affinity adhesion to the endothelium that is sufficient to withstand the forces of the blood flow and remain stationary.²⁰ Using intravital and electron microscopy, animal studies have revealed that all types of white blood cells undergo leukocyte extravasation.^21–30 Although leukocyte rolling and adhesion are often perceived as abnormal processes, they have been observed in healthy and inflammatory states in humans^1–4,7,9 and animals.^{14,17,31–33} Rolling and adhesion occur primarily in the post-capillary venules^20,34, but have been reported in arterioles during inflammatory and rarely in healthy states.³⁵ Compared to non-inflammatory states, inflammation decreases rolling velocity^7,19,36 and increases the number of adherent leukocytes.¹⁷ Interestingly, mechanical stimulation alone in mice can increase leukocyte rolling by 32%.³²

Visualizing leukocytes in cutaneous microcirculation may provide clinical value, similar to what has been shown in the sublingual microcirculation of patients with septic shock.¹ Though clinical RCM devices enable noninvasive imaging of leukocyte motion in skin, assessments of leukocyte-endothelial interactions are manual. To enable the study of cutaneous leukocytes as potential biomarkers, we developed a guideline for identifying adherent and rolling leukocytes in RCM videos of the cutaneous microcirculation. We hypothesized that the guideline can improve the inter-rater reliability of identifying these leukocytes via reflectance confocal videomicroscopy.

MATERIALS & METHODS

We conducted a cross-sectional study at the Vanderbilt University Medical Center and the Nashville Veterans Affairs, after obtaining approval from their Institutional Review Boards. All subjects gave informed consent.

Study Participants

To ensure the inclusion of inflammatory and non-inflammatory states, we enrolled 12 healthy subjects and 60 patients after allogeneic hematopoietic cell transplantation (HCT). This resulted in 1522 RCM videos. Healthy subjects (N=12; 6 Male, 6 Female, Age Range: 18 - 76) felt well and had no visible rash on imaging day. Patients after HCT (N=60; 37 male, 23 female, age range: 19 - 73) had suspected GVHD (N=36) or no GVHD symptoms (N=26) on imaging day. Three patients were imaged twice, first as patients after HCT without GVHD and later as suspected GVHD patients.

Data Collection and Processing

We visualized cutaneous microvasculature with a reflectance confocal microscope, the VivaScope 1500 (Caliber Imaging and Diagnostics, Rochester, New York). It was approved by the FDA and Centers for Medicare & Medicaid Services for direct patient care reimbursement.¹⁰ The VivaScope 1500 uses a near-infrared diode laser (830 nm) to acquire 0.5x0.5 mm² videos at 9 frames per second (fps) up to 200 μm deep in the tissue. An RCM image is formed by collecting the light reflected from the tissue through a pinhole. Melanin, melanosomes, connective tissue, and other structures with high refractive index appear bright in RCM images.³⁷ Our protocol⁹ entailed imaging two standard sites (left volar forearm and left upper anterior chest) and suspected cutaneous GVHD sites. Within the 8x8 mm² total imaging field of view per body site, we aimed to take at least ten 0.5x0.5 mm² videos with visible blood flow for a minimum length of 30 seconds.^3,9 Each video was processed by Fiji software³⁸ plugin Linear Stack Alignment with SIFT^9,39 to remove motion artifacts (e.g., due to breathing).

Manual Counts of Adherent and Rolling Leukocytes in RCM Videos

An experienced investigator (IS) trained two medical students (ZZ, RP) in a two-hour session to identify adherent and rolling leukocytes. IS selected 11 RCM videos that demonstrated clear examples of adhesion and rolling. Guidance on how to differentiate adherent and rolling leukocytes from mimicking scenarios was not provided at this timepoint. Blinded to diagnoses, the raters independently analyzed each video. Each rater viewed the video at least once and counted the number of adherent and rolling leukocytes. At times, the raters viewed the videos at a slower frame rate to better distinguish adhesion and rolling leukocytes from mimickers. For post-guideline analyses of the videos, the raters used the developed guideline.

Inter- and Intra-rater Reliability of Adherent and Rolling Leukocyte Counts in RCM Videos

The inter-rater reliability of the counts was determined before and after guideline development (henceforth referred to as pre-guideline and post-guideline, respectively). We evaluated 88 distinct videos pre-guideline (median video length: 56.78 sec; average: 48.04 sec; range: 6.67 sec – 93.67 sec). Next, we evaluated 1434 videos post-guideline (median: 32.44 sec; average: 34.57; range: 0.78 sec - 147.78 sec).

One rater’s (ZZ) intra-rater reliability was determined post-guideline in 20 videos. The selection of videos representative of the full dataset was ensured by a process detailed in the Supplementary Information section. Blinded to prior ratings, ZZ re-analyzed the 20 videos more than a month after the initial ratings.

Guideline Development

To create a consensus on identifying adherent and rolling leukocytes via reflectance confocal videomicroscopy, we used a modified Delphi method (without anonymity) conducted between September 2019 and June 2020. We convened a panel of two trainees (ZZ, RP) and an experienced investigator (IS). A board-certified dermatologist and RCM clinic director (ERT) was frequently consulted throughout the development of this guideline. Over the past three years, IS has acquired and analyzed hundreds of RCM videos. IS and ET have discussed their observations of leukocyte-endothelial interactions with confocal videomicroscopy experts including Milind Rajadhyaksha and Salvador Gonzalez who first described the phenomenon in human skin.⁴⁰

The panel conducted three rounds of discussions over a period of nine months. After the initial two-hour training by IS to count adherent and rolling leukocytes, each rater (ZZ, RP) independently reviewed 88 RCM videos, blinded to prior evaluations by IS of these same videos. Of the 88 videos, 36 videos had discrepancies between at least two of the three raters. In round one, the panel (ZZ, RP, IS) met in person for a three-hour meeting to carefully review each of the 36 videos. Based on the discussion of the discrepancies, the panel identified key features defining adherent and rolling leukocytes in RCM videos, as well as the key mimickers of each interaction. In round two, the panel consulted with ERT, who provided additional suggestions for the guideline. Upon finishing the analysis of all 1522 videos, ZZ and RP met with IS and ERT in round three to identify the best videos that illustrate the guideline’s key points (Figures 2–3, Videos S1–S8).

Figure 2.

An adherent leukocyte (blue circle) and associated mimicking scenarios (red circles). A) An adherent leukocyte is stationary ≥ 30 seconds near the periphery of the interconnective tissue space (yellow dashed lines) and exhibit fluctuations in surface pixel brightness. The adherent leukocyte is typically located near the periphery of the blood flow (cyan dotted lines), usually but not always present. B) A cell stationary < 30 seconds is not an adherent leukocyte. In this example, it is a slow-rolling leukocyte that would be counted as a rolling leukocyte. C) Multiple fast-flowing leukocytes may give the illusion of an adherent leukocyte, but are not visible in all frames throughout the 30 seconds. D) A cell without fluctuations in pixel brightness is most likely a basal keratinocyte or melanocyte and not an adherent leukocyte. Basal keratinocytes and melanocytes may act as mimickers even though they are extravascular, due to the inability to clearly visualize the endothelial wall in RCM videos.

Figure 3.

A rolling leukocyte (blue circle with arrow) and associated mimicking scenarios (red circles). A) A rolling leukocyte with forward rotational motion is located near the periphery of the interconnective tissue space (yellow dashed lines) and moves slower than the rest of the blood flow (cyan dotted lines). The rolling leukocyte is typically located near the blood flow periphery. B) A cell without forward rotational motion (red circle) is not considered a rolling leukocyte. In this example, the cell is sliding. C) A cell rotating within the blood flow (red circle) is likely rotating on its axis (i.e., tumbling) rather than rolling. D) A cell rotating and flowing at the same speed as the blood flow (red circle) also more likely tumbling than rolling.

We conducted a literature review of adherent and rolling leukocyte definitions in prior studies.^{14,17,46,20,31,36,41–45} We incorporated our own observations of the pitfalls that can cause misidentification of leukocytes in RCM videos as well as prior observations of leukocyte motion into a step-by-step flow chart (Figure 1).

Figure 1.

A guideline to identify A) adherent and (B) rolling leukocytes and differentiate them from mimicking scenarios. Viewing the videos at a slower frame rate enables better identification of adhesion and rolling leukocytes.

Statistical Analysis

We calculated the intraclass correlation coefficient (ICC) as a measure of reliability using the irr package in RStudio V1.1.463. We used the following parameters in the ICC calculation: single-rating, absolute-agreement, two-way random/mixed-effects model.^47,48

RESULTS

Pre-Guideline Inter-Rater Reliability of Adherent and Rolling Leukocyte Counts

Reflectance confocal videomicroscopy enables visualization of leukocyte motion in cutaneous microcirculation. Once attached to the skin, the microscope enables video acquisition of a 0.5x0.5 mm² en face view of the skin within an 8x8 mm² total field of view at 9 frames per second. To determine the pre-guideline inter-rater reliability of adherent and rolling leukocyte counts, two trained raters analyzed 88 RCM videos. The baseline (i.e. pre-guideline) intra-class correlation coefficients (ICC) among two raters were 0.056 (95% confidence interval: 0-0.236; n=88 videos, N=10 subjects) for adherent and 0.385 (0.191-0.550; n=88, N=10) for rolling leukocyte counts (Table 2).

Table 2.

The inter-rater intraclass correlation coefficients (ICCs) of adherent and rolling leukocyte counts among two blinded raters, conducted pre- and post-guideline. The intra-rater ICCs of the leukocyte counts in one blinded rater, conducted more than a month after the initial ratings.

	INTER-RATERICCs		INTRA-RATERICCs
	Pre-Guideline (2 raters, 88 videos)	Post-Guideline (2 raters, 1434 videos)	(1 rater, 20 videos)
Adherent	0.056 (CI*: 0, 0.236)	0.791 (0.770, 0.809)	0.953 (0.886, 0.981)
	5 cells (rater 1)	154 cells (rater 1)	15 cells (rater 1: 1st review)
	47 cells (rater 2)	193 cells (rater 2)	16 cells (rater 1: 2nd review)
Rolling	0.385 (0.191, 0.550)	0.626 (0.593, 0. 657)	0.956 (0.894, 0.983)
	37 cells (rater 1)	66 cells (rater 1)	20 cells (rater 1: 1st review)
	29 cells (rater 2)	134 cells (rater 2)	19 cells (rater 1: 2nd review)

^*95% Confidence Interval

Guideline for Identifying Adherent and Rolling Leukocytes in Upper Dermal Microvasculature

Using a modified Delphi method (without anonymity), we created a guideline to reliably identify adherent or rolling leukocytes via reflectance confocal videomicroscopy. The complete guideline is comprised of definitions (Table 1), a step-by-step flow chart (Figure 1), and corresponding images and videos to aid in the identification of each interaction from its mimicking scenarios (Figures 2–3, Videos S1–S8).

Table 1.

Definitions of adherent and rolling leukocytes visualized via RCM. Blood vessel walls are not clearly visible in RCM videos. We use the periphery of the interconnective tissue space as an approximate location of the blood vessel wall ⁹.

	Definitions	Mimickers
Adherent Leukocyte	A cell that is stationary ≥ 30 seconds near the periphery of the interconnective tissue space (i.e. dark space between the bright-appearing connective tissue). It exhibits rapid fluctuations in surface pixel brightness that occur multiple times per second over time (Figure 2A, Video S1). It should be seen in each frame for at least 30 seconds unless the imaging view shifts momentarily due to patient movements.	(1) Slow-rolling leukocytes move forward at a velocity <5 μm per second16 (Figure 2B, Video S2). (2) Multiple fast-flowing leukocytes appearing near each other in serial RCM video frames may leave the impression of a single stationary leukocyte (Figure 2C, Video S3). (3) Basal keratinocytes or melanocytes may appear as bright cells due to backscattering caused by melanin (Figure 2D, Video S4).8
Rolling Leukocyte	A cell that is rotating in a forward rotational motion near the periphery of the interconnective tissue space and moves slower than the rest of the blood flow. (Figure 3A, Video S5). This definition includes slow rolling.	(1) Sliding leukocytes move due to the propelling motion of the blood flow.50 The fact that they do not move slower than the blood flow nor rotate suggests that they are not part of the inflammatory cascade (Figure 3B, Video S6). (2) Tumbling leukocytes50 rotate but do not move slower than the blood flow, suggesting the absence of leukocyte-endothelial interactions (Figure 3C–D, Video S7–8).

Definition of adherent leukocytes in RCM videos.

We define an adherent leukocyte based on the following criteria: (i) stationary ≥ 30 seconds near the periphery of the interconnective tissue space (i.e., dark space between the bright-appearing connective tissue⁹), (ii) seen in each frame for at least 30 seconds unless the imaging view or depth shifts momentarily due to patient movements, (iii) exhibits rapid fluctuations in surface pixel brightness that occur multiple times per second over time (Figure 2A, Video S1, Table 1). Because blood vessel walls are not clearly visible in RCM videos, we used the periphery of the interconnective tissue space⁹ as an approximate location of the endothelial lining of the blood vessel wall. Mimicking scenarios include slow-rolling (Figure 2B, Video S2) and fast-flowing leukocytes (Figure 2C, Video S3), basal keratinocytes (Figure 2D, Video S4), and melanocytes (Figure 2D, Video S4).

The duration of an adherent leukocyte’s attachment to the endothelial wall has been variably defined in the literature, ranging anywhere from 1 second⁴⁵ to 10 minutes or more.^{14,17,36,41–44} In an induced inflammatory state, most leukocytes remain adherent for ≥ 30 seconds. Furthermore, < 1% of leukocytes that adhere for < 30 seconds extravasate.^43,49 Thus, we selected 30 seconds as the minimum duration in our adherent leukocyte definition.

We included additional criteria to address common mimicking scenarios of adherent leukocytes. When viewing a video at 9 fps, fast freely-flowing leukocytes in narrow vessels perpendicular to the skin surface may appear as a single adherent leukocyte (Figure 2C). However, if the imaging depth and view remain stable, an adherent leukocyte is visible in all consecutive frames for ≥ 30 seconds. The rater may need to watch the video at a slower frame rate or review each frame to confirm the presence of an adherent leukocyte. Additionally, due to the difficulty in visualizing the blood vessel wall via RCM, rapid fluctuations in surface pixel brightness is a criterion to differentiate adherent leukocytes from mimickers. This cell surface change may be attributable to cytoskeletal rearrangements that adherent leukocytes are known to undergo.²⁵ The cytoskeletal rearrangements involve cytoplasmic extensions of the adherent leukocyte (pseudopods) that probe the endothelium, in search of entrance into the connective tissue.²⁵ In contrast, rolling and extravasated leukocytes do not undergo cytoskeletal rearrangements (Figure 2D).

Definition of rolling leukocytes in RCM videos.

We define a rolling leukocyte as a cell that is rotating in a forward motion near the periphery of the interconnective tissue space and moves slower than the rest of the blood flow (Table 1, Figure 3A, Video S5). Rolling leukocyte mimickers include sliding (Figure 3B, Video S6) and tumbling leukocytes (Figure 3C, D; Video S7–S8).

It is important to note that leukocytes can rotate not only due to leukocyte-endothelial interactions but also due to blood flow turbulence or twists and turns of the vessels.⁵⁰ Sliding and tumbling leukocytes do not move slower than the blood flow,^16,50 which is an important feature that helps distinguish them from rolling leukocytes. The limited visualization of the blood vessel wall increases the difficulty of determining the presence of such interactions. If the leukocyte is rolling near the endothelium (i.e. the periphery of the interconnective space) and slower than the rest of the blood flow, it is more likely to be a rolling leukocyte than one merely rotating on its axis.^17,20,31

If a leukocyte transitions from rolling to adhesion or vice versa and fits criteria for both, it counts as an adherent and a rolling leukocyte. We have not yet observed this occurrence.

Guideline Improves the Inter-Rater Reliability of Adherent and Rolling Leukocyte Counts

To assess the value of the guideline, the two raters independently evaluated 1434 additional RCM videos from 67 subjects. Post-guideline inter-rater reliability improved significantly in adherent (0.791, 0.770-0.809; n=1434 videos; N=67 subjects) and rolling (0.626, 0.593-0.657; n=1434; N=67) leukocyte counts, compared to pre-guideline assessments (Table 2). To determine the intra-rater reliability of the counts post-guideline, one rater reanalyzed 20 selected videos one month after the initial assessments. Post-guideline intra-rater reliability of the adherent and rolling leukocytes was 0.953 (0.886, 0.981; n=20, N=12) and 0.956 (0.894, 0.983; n=20, N=12), respectively (Table 2).

DISCUSSION

The clinical value of noninvasive imaging of cutaneous leukocyte-endothelial interactions remains to be explored. However, distinguishing adherent and rolling leukocytes from mimicking scenarios in cutaneous microcirculation via in vivo reflectance confocal videomicroscopy can be challenging. We present a guideline to aid in the manual identification of adherent and rolling leukocytes in RCM videos. In 1522 videos of 72 human subjects, the guideline improved the inter-rater ICCs of adherent leukocyte counts from 0.056 to 0.791 and rolling counts from 0.385 to 0.626.

Noninvasive imaging of leukocytes in skin, the largest and most accessible organ, could be potential biomarkers in a variety diseases where the leukocyte inflammatory cascade plays a role.^1–7 Clinical RCM devices have not been widely used for the study of the cutaneous microcirculation,^10,51 which may be partially attributable to a lack of guidance on interpreting videomicroscopy features. This guideline will allow further investigation of the clinical significance of observable leukocyte motion via reflectance confocal videomicroscopy.^1–7

Limitations of the guideline include the probable exclusion of some leukocytes (limited sensitivity) due to the stringent criteria (high specificity). Additionally, there is no external FDA-approved modality available to confirm in human subjects that the identified cells are leukocytes. The guideline is only appropriate for videos of sufficient duration and quality, which can be difficult to obtain in the setting of movement artifacts (e.g., heavy breathing and tremors). Additionally, because the raters gain experience from analyzing each additional video, the guideline is likely not solely responsible for the improvement in the ICCs of the leukocyte counts. Despite these limitations, the guideline provides a systematic way of identifying leukocytes via reflectance confocal videomicroscopy.

We encountered various challenges in identifying adherent and rolling leukocytes in cutaneous microcirculation via RCM. Animal vessels can be surgically exposed and cells can be labeled, enabling clear visualization of the entire leukocyte inflammatory cascade, including active extravasation.^13–19 However, in vivo studies of the human microcirculation are not amenable to these techniques and remain limited in image quality.^13,14,17 This results in limited visualization of the blood vessel wall and individual cells, causing difficulty in identifying adherent and rolling leukocytes from mimicking scenarios. These limitations may also explain why we did not clearly observe the other steps of the leukocyte inflammatory cascade, such as active extravasation. There is also a lack of reliable automated algorithms to identify and count leukocyte-endothelial interactions in RCM videos. Given these challenges, the guideline can assist in identifying adherent and rolling leukocytes, as suggested by the increase in the post-guideline ICCs.

The pre-guideline disagreements were likely high due to limited training, lack of a guideline, and the trainees’ inexperience. The improvements in the post-guideline ICCs can be attributed to the delineation of mimicking scenarios, as well as specific criteria that need to be met for a leukocyte to be counted as adherent or rolling. Rolling leukocytes were particularly challenging to identify pre-guideline. Leukocytes could merely be rotating within the bloodstream instead of rolling along the vessel wall. Based on observations in animal studies,^{14,17,46,20,31,36,41–45} leukocytes are much more likely to be rolling if they are seen near the periphery of the blood flow and moving at a slower speed than the rest of the cells. Including these two observations as part of the rolling leukocyte criteria allowed the raters to separate rolling leukocytes from mimicking scenarios more easily.

Although the guideline improved ICCs for adherent and rolling leukocytes, the ICCs did not improve past 0.791 and 0.626, respectively. After discussing each disagreement that occurred post-guideline, the raters concluded that most discrepancies were caused by differing interpretations of the written guideline. The raters did not have access to the illustrative videos (S1–S8) that accompany the guideline, because these videos were selected only after finishing all analyses. The scenarios resulting in the most discrepancies between the two raters include stationary leukocytes fading in and out of the frame. Additionally, the degree of fluctuating pixel brightness on the surface of a cell was often insufficient to count a leukocyte as adherent. We commonly observed flowing or sliding leukocytes that could be mistaken for rolling leukocytes. We envision that the inter-rater agreement will be higher for future raters who have access to the sample videos, in addition to the written guideline. The illustrative videos help ensure that the interpretation of the written guideline is similar among raters. Future studies will need to validate the written guideline that is accompanied by illustrative videos.

Because both raters were involved in the guideline development, the reliability of these cutaneous leukocyte counts among other raters needs to be investigated. Future studies will explore the imaging of cutaneous leukocytes via reflectance confocal videomicroscopy as noninvasive biomarkers in a variety of diseases with inflammatory components, such as psoriasis, cutaneous graft-versus-host disease, sepsis, and cardiovascular disease. ^1–7,52

PERSPECTIVES

We present a guideline that reliably identifies cutaneous adherent and rolling leukocytes via in vivo reflectance confocal videomicroscopy. The guideline will enable future studies to reliably test the diagnostic and prognostic potential of these biomarkers in dermatological, inflammatory, and immune diseases.

LIST OF ABBREVIATIONS

FOV

field of view

FPS

frames per second

FDA

Food and Drug Administration

GVHD

graft-versus-host disease

ICC

intraclass correlation coefficient

HCT

hematopoietic cell transplantation

RCM

reflectance confocal microscopy

Data availability:

Datasets related to this article can be found by request to corresponding author and will be posted at http://vdtrc.org.