Immune cells continuously travel throughout our body as a means of immune surveillance. Moving within the bloodstream allows for their fast transport to even distant sites but requires extravasation once they have reached their target organ. Immune cell extravasation across the vascular wall is a multistep process regulated by the sequential interaction of different signaling and adhesion molecules on the endothelium and the immune cells (Nishihara et al., 2020; Marchetti and Engelhardt, 2020). These molecular interactions mediate distinct sequential steps, namely tethering and rolling to reduce travel speed, shear-resistant arrest, polarization, and crawling of the immune cell on the luminal surface of the endothelium and, finally, immune cell diapedesis across the endothelial layer (Marchetti and Engelhardt, 2020; Nourshargh and Alon, 2014; Nourshargh et al., 2010).

The precise molecular mechanisms mediating the multistep immune cell extravasation in each organ depend not only on the immune cell subset but also on the specific characteristics of the vascular bed. For example, in the central nervous system (CNS), the endothelial blood-brain barrier (BBB) establishes a tight barrier that strictly controls the transport of molecules across the BBB, ensuring tissue homeostasis required for neuronal function (Marchetti and Engelhardt, 2020; Engelhardt et al., 2017). The BBB similarly controls immune cell trafficking into the CNS. Thus, accounting for these special barrier properties, unique characteristics of the multistep T-cell migration across the BBB have been described. For instance, T cells crawl for very long distances against the direction of blood flow on the surface of the BBB endothelium in search of permissive locations for diapedesis (Bartholomäus et al., 2009; Lyck and Engelhardt, 2012; Castro Dias et al., 2021). Research on T-cell interaction with the BBB has already been successfully translated into therapies in the clinic (Steinman and Zamvil, 2005; Singer, 2017).

Exploring the entire multistep extravasation of immune cells across the BBB has been significantly advanced by making use of in vitro BBB models maintaining their barrier properties and placing them into microfluidic devices (Lyck et al., 2023). Combined with microscopic setups that allow for in vitro live-cell imaging (Figure 1A and B) of the immune cell interaction with the brain endothelial monolayer under physiological flow over time, the molecular mechanisms mediating the sequential interaction of T cells during extravasation across the BBB have been delineated (Marchetti and Engelhardt, 2020). The live-cell imaging is largely performed in the phase-contrast imaging modality (Figure 1C), which does not require establishing fluorescent labels of the imaged cells and avoids potential phototoxicity of the fluorescent imaging modality. These studies have shown that upon their arrest, T cells polarize and either crawl at speeds between 3 and 10 µm/min over the brain endothelial monolayer or probe the endothelial monolayer by remaining rather stationary and sending cellular protrusions into the endothelial monolayer (Figure 1D; Steiner et al., 2010; Rudolph et al., 2016). Both behaviors can lead to diapedesis of the T cells across the brain endothelial monolayers. This process lasts at least 3–5 min, with some immune cells observed to protrude and retract several times prior to finalizing a prolonged diapedesis process to the abluminal side of the brain endothelial monolayer (Castro Dias et al., 2021; Abadier et al., 2015). Finally, T cells that have successfully migrated across the brain endothelial monolayer usually continue to migrate underneath the endothelial monolayer (Shulman et al., 2009; Arts et al., 2021).

Figure 1

Download asset Open asset

The data analysis of these ‘in vitro flow assays’ requires time-consuming offline frame-by-frame analysis of the imaging data by individual experimenters, in which the dynamic interactions of each individual immune cell has to be followed over the entire time of the assay manually and assigned to specific categories. Such analysis is tedious, and accurately assigning the different T-cell behaviors requires experience. Thus, this manual analysis is prone to inevitable errors and subjective judgments of the various experiments. Furthermore, the time-consuming manual T-cell tracking limits the number of events that can be studied and, thus, the statistical power of the analysis.

Automation of the analysis of the recorded multistep T-cell extravasation across the BBB in the microfluidic device would thus be highly desirable. It is, however, hampered as these assays are usually performed with unlabeled cells and imaged by phase contrast (Figure 1C), which poses a challenge due to the similar grayscales and morphology of the immune cells interacting with the brain endothelial cells. Further challenges include rapid cell movement under flow. When perfused on top of the endothelial cells in the presence of shear flow, at first T cells instantly appear within the field of view (FoV). Later on they are often either suddenly displaced over a certain distance or completely detached and washed away (Figure 1D). Proper analysis of these events is mandatory for reliable T-cell tracking but also with respect to the study of the overall avidity of the dynamic T-cell interaction steps with the underlying brain endothelium. Thus, it was compulsory to establish a tracking solution that accounts for the effect of the flow on the migrating cells and the distinct migration regimes.

Segmenting of cells imaged in phase contrast migrating on top of cellular monolayer of similar appearance is a challenging task. Furthermore, the analysis of T-cell interactions requires detection and quantification of the transmigration. While recent advances in computer vision established several frameworks for detection (He et al., 2017; Wang et al., 2022; Carion et al., 2020; Schmidt et al., 2018) or segmentation (Ronneberger et al., 2015; Stringer et al., 2021), no tool is readily available to complete both tasks simultaneously.

There is a large number of powerful cell tracking solutions developed that focus on different aspects of cell migration and are embedded in different frameworks or solutions. The general-purpose Trackpy (Allan et al., 2021) allows for simple particle tracking and is not suitable to track cells migrating under flow. It is used for the tracking of cells imaged in phase contrast in the Usiigaci framework (Tsai et al., 2019). The widely used TrackMate (Ershov et al., 2022) plugin for ImageJ (Rueden et al., 2017) allows for tracking of particles performing either the Brownian motion or a linear motion, but not the combination of two. Instant appearance or disappearance as well as the resolving tracks of under-segmented cells cannot be easily established in this framework. Similarly, CellTraxx (Holme et al., 2023) performs matching of cells in the adjacent frames, prohibiting its use for tracking of under-segmented, intersecting cells or under flow. The Bayesian Tracker (btrack) framework (Ulicna et al., 2021) is using spatial information as well as appearance information for track linking focusing on the cell divisions required for cell lineage tracing.

Here, we introduce the developed under-flow migration tracker (^UFMTrack) framework that systematically addresses the hurdles mentioned above, allowing it to perform automated tracking and analysis of cell-cell interactions imaged in phase contrast. We also show a successful implementation of ^UFMTrack to build an analysis pipeline for T-cell interactions with brain microvascular endothelial cells in vitro under physiological flow. ^UFMTrack reaches 90% T-cell tracking efficiency, performing comparably to manual analysis while eliminating the experimenter’s bias and improving the accuracy of T-cell tracking. Therefore, it enables significant savings in the labor force and time for data analysis.

To design and develop the automated T-cell under-flow migration analysis framework (^UFMTrack framework) presented here, we made use of in vitro imaging datasets following T-cell migration across primary mouse brain microvascular endothelial cells (pMBMECs) under physiological flow in vitro. The framework combines three components: T-cell segmentation and transmigration detection, T-cell tracking under flow, and analysis of each of the multistep T-cell migration cascade steps. Segmentation and transmigration detection of the T cells, migrating on the pMBMECs, is performed with a 2D+T U-Net-like convolutional neural network (Ronneberger et al., 2015). The T-cell under-flow tracking algorithm was formulated as a constrained optimization problem.

Next, we describe the methods and algorithms employed to develop ^UFMTrack. Links to the code and the datasets used for model training can be found in the Data availability section.

T-cell segmentation

Reliable cell segmentation is crucial for reliable cell tracking. In the phase-contrast imaging modality, it was impossible to achieve reliable differentiation between T cells and endothelial cells of the pMBMEC monolayer based on pixel intensity. For detection of T-cell transmigration across pMBMEC monolayers (diapedesis), sufficiently reliable cell segmentation of the transmigrated T cells was also required. To achieve this, we have designed 2D and 2D+T U-Net-like (Ronneberger et al., 2015) fully convolutional neural network-based models for multitask learning. The models were trained to predict three maps: cell probability, the probability that the T cell is below the pMBMEC monolayer, and cell centroids. The models performed predictions based on the grayscale of the respective phase-contrast images in the case of the 2D model or sequences of 5 time frames in the case of the 2D+T model. The models were implemented in TensorFlow (Abadi et al., 2016). The training was performed on a dataset corresponding to 154 Megapixels of raw image data using the annotation mask for T cells (‘T-cell mask’) and the mask of the transmigrated part of the T cells (‘transmigration mask’), the centroids map, and the weight map (Figure 2). For validation, 37 Megapixels of raw imaging data were used. The models have approximately 3 and 8 M parameters correspondingly, and the model architectures are summarized in Appendix 1—table 1 and Appendix 1—table 2. Details on models’ implementation, training, and image processing can be found in the ‘Segmentation models’ section in Appendix 1.

Figure 2

Download asset Open asset

In the T-cell mask prediction task, the 2D+T model outperformed the 2D model by a notable 9% according to the average precision (AP) metric (Table 1). At the same time, for the transmigration mask, which is much more difficult to infer, the 2D model performance reached only 54%, which is insufficient for reliable detection of T cells migrating across the pMBMEC monolayer. In this task, the 2D+T model outperformed the 2D model by 32% AP. We observed that the 2D+T model was sensitive to frame misalignment, leading to false positive detection of transmigrated T cells. Thus, to generate the preliminary T-cell masks used for frame alignment and histogram normalization, we employed the 2D model. Afterward, for the T-cell segmentation and transmigration detection, we employed the 2D+T model, followed by a watershed algorithm based on the inferred T-cell mask and the cell centroids as seed points. A stitched phase-contrast image sequence can be seen in Video 1 and overlayed with the segmented cells and highlighted transmigration mask in Video 2.

Table 1

	Performance, AP
	2D	2D+T
T-cell mask	86.12%	95.26%
Transmigration mask	50.56%	82.74%

Video 1

Download asset

Video 2

Download asset

After segmentation, we suppressed the noise by discarding small objects with an area below 30 pixels (the mean area of a T cell is about 400 pixels). For each T cell, we evaluated its coordinates as geometric mean coordinates, the angle of the longer axis, and the elongation $ϵ=\sqrt{\frac{{\sum }_i{\left(x_i^{`}-x_0^{`}\right)}^2}{{\sum }_i{\left(y_i-y_0^{`}\right)}^2}}$ , where ${x_i^{`},y_i}^{`}$ are projections of the ith T-cell mask pixel coordinates on the orthogonal long and short axes of the T cell, and ${x_0^{`},y}_0^{`}$ are the corresponding projections of the geometrical center of the T cell. The summation is performed over the area covered by the mask of the corresponding T cell. Additionally, we evaluated the mean T-cell probability ${\displaystyle \overline{p_{cell}}=\frac{1}{n_{pix}}\sum _i{p}_{cell,i}}$, mean transmigration probability ${\displaystyle \overline{p_{tr}}=\frac{1}{n_{pix}}\sum _i{p}_{tr,i}}$, and cell transmigration coefficient as $t_c=\frac{1}{n_{pix}}{\sum }_i\frac{min(p_{tr,i},p_{cell,i})}{p_{cell,i}}$ , where $n_{pix}$ is number of cell pixels and $p_{cell,i}$, $p_{tr,i}$ are the T-cell and transmigration probabilities for ith pixel in the predicted T-cell mask.

T-cell tracking

The presence of flow causes several types of discontinuous events that were managed with a specialized tracking algorithm. These are: (1) the sudden appearance of T cells in the FoV, (2) the sudden detachment of a T cell followed by its disappearance from FoV, and (3) the displacement over significant distances (up to hundreds of micrometers) of T cells that do not adhere firmly to the pMBMECs. The primary inspiration for our approach was the conservation tracking algorithm (Schiegg et al., 2013). We could consistently reconstruct the T-cell tracks by performing global optimization constrained to controlled probabilities of T-cell appearance, disappearance, and displacement due to the flow at every timepoint. The standard tracking algorithms like tracking by association is either unable to detect track links with high displacement and thus reconstructs several disconnected segments of a cell track, or leads to a high number of false track links. Our procedure favors the reconstruction of long T-cell tracks while at the same time allowing for tracking of the T cells which detach or are displaced by the flow (‘accelerated T-cell movement’) over a longer distance (8 µm) with speed significantly higher than their crawling speed.

The tracking consists of four main steps: linking, searching for track candidates, global track consistency resolving, and resolving the track intersections. The inputs for the tracking are the segmented cells and the cell centroids (Figure 3A). The centroids and the cell proximity are used to quantify under-segmentation by introducing the node multiplicity (Figure 3B, see section T-cell tracking in Appendix 1 for details). At the linking step, we identified all possible connections of T cells between the time frames (Figure 3C).

Figure 3 with 1 supplement see all

Download asset Open asset

Next, during the search for track candidates, we aimed to find continuous track segments of T cells or under-segmented groups of T cells crawling on top of or below the pMBMEC monolayer without accelerated movement segments on the T-cell track characterized by rapid T-cell displacements. The whole dataset of T cells across all timepoints was represented as a graph. Each vertex corresponds to a T cell (multiplicity m=1) or a group of potentially under-segmented T cells (m>1). The vertices are connected according to links obtained at the linking step. Track segments were found by performing global optimization to find consistent connectivity of vertices across the timepoints (Figure 3D) by employing an approach inspired by the conservation tracking algorithm summarized in Schiegg et al., 2013. Optimization was performed using the CP-SAT constrained optimization procedure using the open-source OR-Tools library (Perron and Furnon, 2021).

Next, resolving global track consistency was performed. In this step, the scope of T-cell tracking is shifted from individual nodes (representing T cells at particular time frames as well as groups of under-segmented T cells) to the track segments (unambiguous sequences of nodes) and vertices at the endpoints of the segments. These are track start and end points, points of merging and separation of track segments in case of under-segmentation, as well as ambiguous points on a track. The latter was identified by sudden T-cell displacement, a hallmark of detaching and reattaching T cells and T-cells transitioning from properly segmented to under-segmented T cells or vice versa (Figure 3E). This was followed by the search for potential missing track segments due to accelerated T-cell movement under flow and missing links in the track crossing points. Both are characterized by significant displacement lengths such that they were not detected during the linking step. Thus, we will refer to both as ‘jumps’ in this section. We have also eliminated short and therefore unreliable segments, as well as segments whose multiplicity was found to be $M=0$ (Figure 3F). Afterward, we separated the segments into two categories, namely segments with multiplicity $M=1$, i.e., tracks of isolated T cells, and segments with multiplicity ${\displaystyle M>1}$, i.e., tracks of under-segmented groups of T cells, where tracks of several T cells intersected. Finally, we extended the tracks of isolated T cells into the intersection if the branching vertex had a multiplicity of $n$ and was splitting in exactly $n$ segments with multiplicity $M=1$ (Figure 3G).

Lastly, we resolved the track intersections to obtain reliable T-cell tracks, i.e., identifying track segments corresponding to the same T cell before and after the under-segmented track region (Figure 3H).

Detailed information on the developed under-flow T-cell tracking algorithm is given in the T-cell tracking section in Appendix 1.

T-cell migration analysis

Next, we performed the T-cell migration analysis based on the reconstructed T-cell tracks. We selected tracks inside the fiducial area of the FoV, namely coordinates of the T cell at all timepoints along the track were located at least 25 µm away from the bounding box enclosing all segmented T cells. Next, tracks of T cells touching another T cell at the end of the assay acquisition were excluded since T cells directly adjacent to each other can hide the start of T-cell transmigration across the pMBMEC monolayer and thus compromise correct detection and quantification of this step. Additionally, only tracks with T cells were assigned in at least 6 time frames during the physiological flow phase. We also require T cells to be assigned for at least 75% of time frames along the track. Under-segmented parts of T-cell tracks were not considered. Examples of selected tracks can be seen in Video 3.

Video 3

Download asset

After exclusion of the tracks of the cellular debris, which were misclassified as T cells during the segmentation step, using a dedicated ‘not-a-T-cell’ classifier, we performed identification of transmigration (Figure 3—figure supplement 1), probing and crawling migration regimes, as well as accelerated movement along the T-cell track. Detailed information on the analysis of the T-cell tracks is given in the ‘T-cell migration analysis’ section in Appendix 1.

Finally, for each track, we evaluated motility parameters for each of the following migration regimes: probing before the transmigration, crawling before the transmigration, all crawling above pMBMECs monolayer including T-cell crawling segments after the first transmigration attempts, all crawling below pMBMECs monolayer, whole T-cell track excluding accelerated movement and tracking inefficiency regions, as well as whole T-cell track. Specifically, we evaluated the following T-cell motility parameters: duration of each migration regime, the total vector and absolute displacements, the migration path length, the average migration speed (displacement over time), average crawling speed (path length over time), and finally the mean and standard deviation of the instantaneous speed. We evaluated migration time, displacement, and average speed for the accelerated movement regime.

Analysis of trafficking datasets

Having developed a full pipeline based on the ^UFMTrack framework for automated analysis of the multistep T-cell migration cascade across the BBB model, we next aimed to compare the results of the automated analysis with previous studies, assess the capacity of the framework to gain novel insight into T-cell migration under flow, and evaluate its performance as compared to manual analysis.

CD4⁺ T-cell analysis

To this end, we first analyzed a total of 18 imaging datasets each corresponding to 32 min recording dedicated to understanding the multistep migration of CD4⁺ T cells across non-stimulated (n=6), TNF-stimulated (n=5), and IL1-β-stimulated (n=7) pMBMEC monolayers under physiological flow with the developed pipeline.

In these in vitro live-cell imaging datasets, phase-contrast imaging was performed as described in the ‘Data acquisition’ subsection in the Materials and methods section. The T-cell accumulation phase corresponding to the shear stress of 0.1 dynes/cm² lasted for 32 time frames (5 min), followed by conditions of physiological flow (shear stress 1.5 dynes/cm²) for the subsequent 160 time frames (27 min). The shear stress is controlled by an automated syringe pump. After increasing to physiological flow rates (Steiner et al., 2010), a significant number of T cells detached from the pMBMEC monolayers, with higher numbers of T cells detaching from non-stimulated pMBMEC monolayers when compared to those stimulated with pro-inflammatory cytokines as detected by the distribution of the track ending times between 5 and 30 min (Figure 4A).

Figure 4

Download asset Open asset

Analyzing these datasets with the established ^UFMTrack framework, the T-cell behavior type for each detected T-cell track and the aggregated T-cell behavior statistic were obtained for the stimulated and non-stimulated pMBMEC monolayers (Figure 4B). The total number of analyzed T-cell tracks is n=971, n=780, and n=1218 for the non-stimulated, TNF-stimulated, and IL1-β-stimulated pMBMEC monolayers correspondingly. The data obtained with ^UFMTrack were in accordance with our previous observations obtained by manual frame-by-frame analysis (Steiner et al., 2010; Abadier et al., 2015; Haghayegh Jahromi et al., 2019). We further obtained the data for T-cell migration speed, T-cell displacement, and T-cell path lengths for the crawling T cells, as these are the primary cell motility parameters. As shown in Figure 4C–H, we observed statistically significant differences in the mean T-cell motility parameters depending on the pMBMECs stimulation condition. Furthermore, we observed differences in the shape of the distribution of T-cell crawling speeds, consistent with our previous reports of underlying differences in the mechanisms mediating T-cell crawling on non-stimulated vs cytokine-stimulated pMBMECs (Haghayegh Jahromi et al., 2019). While not statistically significant for the IL1-b and with a low significance for the TNF stimulations, an increased variance of the T-cell instantaneous speed on stimulated compared to non-stimulated pMBMECs is observed (Figure 4G). This suggests that T-cell crawling is often interrupted by T-cell recognizing specific cues on the stimulated endothelium. The T-cell meandering index distribution was high on non-stimulated and lower on stimulated pMBMECs (Figure 4H), underscoring that cytokine stimulation enhances directed T-cell movement on the pMBMEC monolayer.

Since the analysis was performed automatically with ^UFMTrack, it allowed for deeper insights into the T-cell migration behavior on the pMBMEC monolayers than obtained by manual frame-by-frame analysis. Specifically, the detection of accelerated movement by ^UFMTrack enables the researcher to quantify the kinetics of individual T-cell detachment from the endothelium rather than simply quantifying a bulk T-cell detachment rate. This detachment kinetics is reflected in the average speed experienced during accelerated movement, which is significantly lower for the IL1-b condition (Figure 4I), suggesting the need to break more bonds with the endothelium when pro-inflammatory cytokines are present.

As our ^UFMTrack workflow achieves sufficient segmentation efficiency to detect T cells below the pMBMEC monolayer and T-cell transmigration across the pMBMEC monolayer, it also enables investigation of T-cell movement after the transmigration step. In Figure 4J, we show the distribution of the migration speed for the transmigrated CD4⁺ T cells. Interestingly, cytokine stimulation of pMBMECs, although applied from the luminal side, also affected T-cell movement at the abluminal side of the pMBMEC monolayer. While, in the microfluidic device used in the present study, the migration of T cells below the pMBMEC monolayer may not be biologically relevant, this analysis option will be valuable for future studies involving multilayer in vitro BBB models that include the vascular basement membrane in addition to pericytes and astrocytes to mimic the entire neurovascular unit (Castro Dias et al., 2021; Mossu et al., 2019).

Importantly, our ^UFMTrack automatically detects and characterizes unusual events, such as reverse T-cell transmigration, that are easily overlooked with manual counting. We do not present the statistics here, as only a few such events were observed. However, when applied to the multilayer BBBs in forthcoming in vitro models, systematic detection and analysis of these rare events will be important to understand T-cell migration in immune surveillance.

CD8⁺ T-cell analysis

As we have previously shown that the multistep T-cell extravasation across pMBMEC monolayers differs between CD4⁺ and CD8⁺ T cells (Rudolph et al., 2016), we next analyzed two datasets studying the multistep migration of CD8⁺ T cells across non-stimulated and TNF/interferon-gamma (TNF/IFN-γ)-stimulated pMBMEC monolayers under physiological flow over 161 time frames (27 min). The CD8⁺ T cells were slightly different in size and appearance when compared to the CD4⁺ T cells used for training the segmentation model. To benchmark our established ^UFMTrack pipeline in this more difficult configuration, the datasets were first manually analyzed by four experimenters: one advanced experimenter with 4 years of experience (AdEx) and three inexperienced experimenters who received comparable 2 hr introduction and training (Ex1-Ex3). The analyses were performed on the subset of the acquisition, 161 time frames long starting from time frame 31.

CD8⁺ T-cell analysis

Manual cell analysis and tracking were performed as described in the ‘Materials and methods’ section separately for each of the 8 tiles of the tiled acquisition. Next, all crawling CD8⁺ T cells that did not perform transmigration were manually tracked for the time span after the flow was increased to the physiological level using the manual tracking in ImageJ.

First, we compared the CD8⁺ T-cell behavior statistics obtained manually and by our automated analysis pipeline (Figure 5A–C). The results obtained by the automated analysis were in full agreement with the manual analysis performed by the experienced experimenter. At the same time, the data highlight the variability of the results from the inexperienced experimenters, confirming the potential for significant experimenter bias.

Figure 5 with 1 supplement see all

Download asset Open asset

We also show the detection efficiency of T cells by the ^UFMTrack. To achieve this goal, we combined the numbers of CD8⁺ T cells detected in each tile by manual analysis. Due to the overlap between the individual tiles, some CD8⁺ T cells are seen more than once. To obtain an estimate of the total CD8⁺ T-cell count detected manually, we scaled the number of CD8⁺ T cells accordingly to the number of detected unique T-cell tracks (see next paragraph). By this approach, we found T-cell detection efficiency to be above 90% (Figure 5D). The increased T-cell density can explain lower efficiency in the non-stimulated condition. These findings set the appropriate T-cell density for automated analysis of their migration behavior on pMBMEC monolayers to 200–300 cells per FoV of the size of 3.8 mm² or 50–80 cells/mm².

Next, we compared the T-cells tracks as obtained manually by three experimenters as well as automatically by our ^UFMTrack. To this end, we first matched the tracks in adjacent tiles to obtain the tracks on the whole imaging area, avoiding multiple counts of the same track. This was achieved by pattern matching with initial offsets between tiles obtained by an automatic frame alignment procedure performed as part of the automatic analysis. We considered T-cell tracks observed in different tiles to be tracks of the same T cell if ${\displaystyle f_{in}>\frac{1}{2}{f}_{out}-0.8}$, where ${\displaystyle f_{in}}$ is the fraction of timepoints along a track at which the distance between the T cells was below 17 µm and $f_{out}$ was the fraction of timepoints along a track at which distance between cells was above 25 µm. We used the same approach to match the T-cell tracks obtained by each experimenter and the automated ^UFMTrack. To compare performance in an objective manner, we excluded manually obtained tracks that lay outside of the fiducial volume of the automated analysis, as well as T cells that were touching another T cell at the end of the acquisition, as those were also excluded from the automated analysis. We then took as 100% the sum of all T-cell tracks detected manually. We evaluated the fraction of all T-cell tracks detected by each experimenter as well as automatically by our ^UFMTrack.

In Figure 5E, we show that our pipeline achieves a comparable T-cell tracking efficiency when compared to the manual analysis. While its performance was not superior to that of the experimenter with 4 years of expertise in the analysis of the under-flow datasets, it does perform better than less experienced experimenters.

We also compared the T-cell motility parameters obtained manually and automatically for the non-stimulated and stimulated pMBMECs (Figure 5F–I). Data obtained for the CD8⁺ T-cell migration speed on pMBMECs was comparable between all experimenters and the automated analysis. In contrast, results obtained for the T-cell crawling speed were significantly different between the automated approach and manual analysis. Taking a closer look at the T-cell tracks (Figure 5—figure supplement 1), we readily observed that the manual analysis contains significant errors in the assignment of the T-cell position. This leads to a jittery pattern in the T-cell migration tracks and overestimates the T-cell path and, thus, T-cell crawling speed. Employing our novel automated analysis pipeline eliminated this systematic error in the measurements.

To finally investigate the potential benefit in the time required for automated vs manual data analysis, we examined the time required for a given experimenter to analyze such datasets manually using the Clockify time tracker (Clockify, 2022). In Figure 5J, we show the average time investment needed to perform a complete T-cell behavior analysis and T-cell tracking in each imaging tile, as well as on average. Clearly, the experienced experimenter outperforms the inexperienced experimenters by a factor of 3. On average, a researcher would thus spend 8.1 hr analyzing a dataset with 300T cells when only the tracks of crawling T cells (50%, i.e. 150 cells) are analyzed. If all T-cell tracks were to be analyzed, this time would further increase to 12.9 hr. Given that on average 10 datasets can be produced per day, manual analysis becomes a bottleneck, leading to delays in exhaustive data analysis and thus ultimately in research progress. The ^UFMTrack framework reduces analysis time by a factor of 3 when analyzing only crawling cells and 5 if all cell tracks are to be analyzed (Figure 5K). With an average analysis time of 2.3 hr, the 10 experiments carried out in 1 day can be thoroughly analyzed within 1 day of machine time. This enables the scalability of flow-based immune cell migration experiments while simultaneously lifting the burden of tedious and time-consuming manual analysis from the researchers.

Applicability of ^UFMTrack to different models of immune cell-endothelial cell interactions

Next we have assessed the broader applicability of the ^UFMTrack framework to different cell typesand BBB models often employed for the in vitro under-flow cell trafficking assays. To this end, we have performed the automated analysis of the multistep T-cell migration cascade in a set of datasets (one to eight) across different cell types and assays. Specifically, we attempted to verify the feasibility of analysis of datasets where the cells forming the monolayer or the migrating cells have noticeably different appearances, cells from other species are used, specifically human-derived cells, and datasets with a different imaging modality. These include human peripheral blood mononuclear cells (hPBMCs), human T cells, bone marrow-derived macrophages (BMDM) interacting with a monolayer of mouse- and human-derived brain microvascular endothelial cells as well as with immobilized recombinant BBB adhesion molecules (see ‘Materials and methods’ section for details). To assess the performance of the model in each condition, we have performed a visual inspection of the cell masks, transmigration masks, and cell tracks for few datasets in each assay type. We then analyzed cell behavior distribution or motility parameters relevant in each case.

hPBMC on HBMEC monolayer

The interaction of hPBMCs with the BBB is altered under different disease conditions. We have therefore analyzed a dataset of human PBMCs on human brain microvascular endothelial cells (HBMEC) to quantify the cell behavior statistic. While the endothelial cells look significantly different than the pMBMECs, the segmentation and transmigration detection quality was sufficient to perform the cell migration analysis. This was also confirmed by the visual inspection of the cell and transmigration masks and cell tracking results (Figure 6A). We then quantified the PBMC behavior distribution and the crawling and migration speed of the crawling PBMC (Figure 6B and C).

Figure 6 with 1 supplement see all

Download asset Open asset

BMDM on the pMBMEC monolayer

Invasion of macrophages into the CNS contributes to many neurological disorders. Therefore, we have explored if ^UFMTrack also allows us to analyze the interaction of BMDMs with the BBB under physiological flow in vitro. To this end, we have analyzed two datasets of mouse BMDMs interacting with pMBMEC monolayers (Figure 6D). Despite the different behavior and appearance of the macrophages as compared to the T cells, the segmentation quality was sufficient to perform the cell tracking and analyze cell motility parameters (Figure 6E and F), as confirmed by visual inspection of the macrophages masks and cell tracking results (Figure 6D). Nonetheless, due to the different appearance of the flattening crawling cells and the transmigrated cells, the transmigration detection trained on the mouse T cells does not perform sufficiently well for the robust detection of transmigrating macrophages.

Human T cells on EECM-BMEC monolayer

Next, we analyzed eight datasets of human T cells interacting with the human induced pluripotent stem cell (hiPSC) derived extended endothelial cell culture method brain microvascular endothelial cell (EECM-BMEC)-like cell monolayers (Figure 6G). EECM-BMECs establish barrier properties comparable to primary human BMECs and display a mature immune phenotype allowing to perform studies for the first time in a fully autologous manner of immune cell interactions with patient-derived in vitro models of the BBB (Nishihara et al., 2021). Despite the cells being of different species, the T cells were successfully segmented, and their transmigration was detected, as confirmed by visual inspection of the T-cell and transmigration masks and cell tracking results (Figure 6G).

We have confirmed the differences in migration statistics (Figure 6J and K), migration and crawling speed of the crawling T cells (Figure 6H), as well as the previously reported difference in relative (normalized to track path length) displacement (Figure 6I) of the T cells treated with anti-α4-integrin+anti-β2-integrin blocking antibodies (bab) and control (ctrl) condition (Soldati et al., 2023).

Human T cells on immobilized recombinant BBB adhesion molecules

Finally, we have analyzed datasets acquired in a different imaging modality. Seven datasets of human T cells interacting with immobilized recombinant BBB adhesion molecules, namely ICAM-1, VCAM-1, and their combination (Figure 6l). In this case, the fluorescently labeled cells were imaged with epi-fluorescent microscopy. In the absence of a monolayer, transmigration detection is not necessary. Thus, we performed cell segmentation directly on the fluorescent signal (see Materials and methods for details) using the same segmentation strategy as for the T-cell masks. The tracking was then performed the same way as for other experiments. The T cells were successfully segmented and tracked, as confirmed by visual inspection. We compared T-cell behavior and motility parameters for three conditions of the recombinant adhesion molecules: ICAM-1, VCAM-1, and ICAM1+VCAM1 (Figure 6M and O). Consistent with previous observations (Steiner et al., 2010; Soldati et al., 2023) where the analysis was performed using the ImageJ (Rueden et al., 2017) either manually or with the TrackMate (Ershov et al., 2022) plugin, we observed that in the absence of ICAM-1, T cells were moving in the direction of flow (Figure 6N). At the same time, ^UFMTrack allows for capturing the rapid accelerated motions of the cells and thus reconstructing longer continuous T cell tracks. This enables capturing the migrations properties of the consecutive steps of the T-cell migration cascade on the cell-by-cell basis.

In this study, we have developed the ^UFMTrack framework. It consists of independent modules and allows for segmentation (Figure 2), tracking (Figure 3), and motility analysis of immune cells, migrating on, across, and below (Figure 1D) the monolayer of brain microvascular endothelial cells under physiological flow in vitro. The developed method relies exclusively on phase-contrast imaging data (Figure 1B). Therefore, it does not require establishing fluorescent labels of the migrating immune cell population to be studied, avoids potential phototoxicity to be considered for fluorescent imaging modalities (Jae Oh et al., 1999; Saetzler et al., 1997; Purschke et al., 2010), and is thus the preferred choice for analyzing the trafficking of sensitive cell types. T-cell segmentation and prediction of the transmigrated T-cell areas are performed using a custom 2D+T U-Net-like convolutional neural network. It enables reliable segmentation of T cells both above and below the pMBMEC monolayer. The existing particle and cell tracking toolkits consider migration of one cell type and thus are not suitable for the detection of distinct migration regimes and cell interactions (Allan et al., 2021; Tsai et al., 2019; Ershov et al., 2022; Ulicna et al., 2021). Furthermore, to our knowledge, none of the existing algorithms consider migration under the flow causing rapid cell displacement. The tracking of T-cell interactions with the pMBMECs during all migration regimes under physiological flow required designing a new tracking algorithm considering rapid T-cell appearance, disappearance, and displacement in the FoV caused by the flow. We have also developed approaches that resolve track intersections, i.e., identifying track segments corresponding to the same T cell before and after under-segmented track regions, which are inevitable during T-cell migration. By establishing the detection of T-cell crawling, probing, transmigration, and accelerated movement combined with reliable T-cell tracking, we have enabled the in-depth analysis of distinct migration regimes on a cell-by-cell basis. By reducing the dataset analysis time by a factor of 5, ^UFMTrack allows for performing a thorough analysis of 10 experiments that a researcher can carry out in 1 day within 1 day of machine time. We have demonstrated that the automated analysis performs on par with manual analysis while improving accuracy and eliminating experimenter bias, and enables scalability of flow-based immune cell migration experiments by reducing the analysis cost and lifting the burden of time-consuming manual analysis.

In this work, we have demonstrated the applicability of the developed framework to the analysis of the multistep extravasation of CD4⁺ and CD8⁺ T cells across non-stimulated or cytokine-stimulated pMBMEC monolayers under physiological flow. We have also demonstrated that the developed framework allows for automated analysis of T-cell behavior statistics and motility parameters of distinct T-cell migration regimes. Results of the automated analysis of CD4⁺ T-cell behavioral statistics performed with ^UFMTrack are in agreement with previous studies using manual data analysis (Figure 4B; Abadier et al., 2015). The automated analysis of datasets of CD8⁺ T-cell migration has shown comparable performance with manual analysis performed by one experimenter with 4 years of analysis experience and three less-experienced experimenters. At the same time, the variance of the results obtained manually showed significant experimenter bias that the automated analysis eliminated (Figure 5F). Additionally, the automated analysis allowed for in-depth analysis of all T-cell migration categories and precise evaluation of T-cell motility parameters, which was not achieved by manual T-cell tracking, even by the most experienced user.

We have further demonstrated the applicability of the ^UFMTrack framework to the analysis of different cell types (BMDM, PBMC), species (human-derived T cells, HBMEC, and EECM-BMEC), as well as tracking and analysis of cell migration imaged using the fluorescent imaging modality. In all cases, the cells could be successfully segmented and tracked, and the cell migration analysis was performed on a cell-by-cell basis. The main challenge for the system is reliable segmentation of the T cells and the transmigrated T cells. In the case of HBMEC the monolayer has a dotted pattern unlike the rather smooth pMBMEC. This leads to more false positives in the T-cell segmentation at the level of individual frames, which nonetheless could be efficiently filtered at the level of cell tracking. In the case of the BMDMs, since they have significantly different appearances during transmigration from the T cells, reliable transmigration detection was not obtained. This prevents the system from quantifying the transmigration statistics of the BMDMs, yet the cell detection and thus motility parameter evaluation of BMDMs was successful (Figure 6D–F). For all the other analyzed cell types, both the cell segmentation and transmigration detection model trained on the annotated CD4⁺ T cells also produced transmigration maps of sufficient quality for reliable transmigration detection. The framework was able to successfully detect the cells of sizes that differ by almost a factor of two (Figure 6—figure supplement 1). This demonstrates that, simply by discriminating cells by size, it is possible to extend ^UFMTrack to study the interaction of several types of immune cells migrating on top of a cellular monolayer under flow and imaged by phase contrast.

Quantifying the fraction of transmigrated T cells is crucial for studying the molecular mechanisms governing the infiltration of autoaggressive T cells across the BBB into the CNS parenchyma and the immune surveillance. The role of specific endothelial or T-cell adhesion molecules in influencing the dynamic interactions of T cells with pMBMECs can, e.g., be probed by quantifying the ratio of T-cell crawling behavior to T-cell probing (Figure 6M; Abadier et al., 2015; Wimmer et al., 2019; Aydin et al., 2023). The analysis procedure established in the ^UFMTrack framework is flexible and can be easily extended to evaluate further properties of the migrating cells. For example, the number of times a T cell interrupts its crawling regime by switching to short probing behavior on the endothelium can be evaluated, thus providing information on the distribution of ‘hot spots’ on the endothelium. Additionally, experiment scalability and the analysis on a cell-by-cell basis enable the search for distinct populations of T cells, e.g., according to the distribution of probing to crawling behavioral ratios. The strength of T-cell adhesion to the endothelium can be probed with the developed framework using the measure of T-cell detachment rate and the distribution of the previously overlooked T-cell accelerated movement occurrences and accelerated movement speed. Finally, the possibility of quantifying the motility parameters of the transmigrated T cells enables future studies involving multilayer in vitro BBB models that include the vascular basement membrane in addition to mural cells such as pericytes and astrocytes mimicking the entire neurovascular unit.

While in this work we have focused primarily on the analysis of mouse T cells interacting with pMBMEC under flow or other immune cells interacting with the endothelial monolayer, the methods we present establish a foundation for a broad range of studies involving in vitro under-flow studies of immune cell trafficking. The tracking algorithm and motility analysis developed in the ^UFMTrack framework are also directly applicable to the analysis of immune cell interactions with recombinant BBB adhesion molecules under flow. In this case, either phase-contrast or epi-fluorescent imaging can be employed when studying fluorescently labeled immune cells, as demonstrated. The T-cell segmentation based on deep neural networks can be applied to studies of the trafficking of other immune cell subsets on and across the pMBMEC monolayer. We demonstrated the successful application of the ^UFMTrack to cells with significantly different appearances, such as the HBMEC cellular monolayers.

We make ^UFMTrack available to the community to build upon it for addressing their specific research questions. The modularity of the framework simplifies the adaptation to analysis of different cell types. The application to immune cell trafficking across other endothelial monolayers, including those from different species or vascular beds as well as lymphatic endothelial cells where the cell appearance is drastically different from the ones demonstrated here, is possible but requires fine-tuning of the trained segmentation model. While training the models presented here required a large dataset of annotated T-cell masks, and transmigration masks, this can be largely avoided for future development. Future models can be developed more efficiently by leveraging our existing annotated dataset of T-cell migration on the pMBMECs while employing transfer learning approaches in a multitask framework with weak supervision and fluorescent labels as auxiliary learning targets to adapt the model for the segmentation of cells with different appearances. This can be further improved by adopting self-supervised contrastive or masked learning methods, which have demonstrated significant advancements in model pretraining in recent years (Caron et al., 2021; Archit et al., 2023; Kraus et al., 2024; Xie et al., 2023). This is the subject of our future studies. The framework could be employed in combination of other methods, e.g., by incorporating fluorescently labeled endothelial adherens junctions one could differentiate the trans- from the paracellular transmigration in the migrating cells. In this case the transmigration locations detected by the ^UFMTrack framework could be automatically extracted for analysis. One could employ the tracking under flow implemented in the ^UFMTrack framework for tracking the immune cells migrating on the vascular wall in vivo to capture the accelerated movement of cells. By sharing the data and open-source code of the ^UFMTrack framework, i.e., the training data used for the segmentation model training, trained models, as well as the model architecture, and full under-flow T-cell tracking and migration analysis pipeline, we hope to encourage the community to pursue these developments to advance the field.

One current limitation of the method is the performance reduction of the T-cell tracking when the density of migrating T cells significantly increases (>250 cells/dataset). Thus, the recommended T-cell concentration is about 100–150 k/ml. However, to avoid nonphysiological interactions between migrating T cells and T-cell clumping, the T-cell density should be kept at moderate levels anyway. Additionally, analyzing data from 8 stitched FoVs allows for obtaining comparable statistics. The current analysis of the behavioral statistics quantifies the fraction of different T-cell migration regimes with respect to the number of adhered cells instead of the total T-cell count passing through the microfluidic device. While the number of fast-moving T cells during the accumulation phase cannot be directly counted using the imaging modality employed here, it can be estimated indirectly from the imaging data with additional calibration experiments and a dedicated machine learning model. Another limitation is that currently, we do not consider dividing immune cells. While cell division events happen rarely (<1% of T cells) and are not primary events for the study of immune cell interaction with the BBB model, the modular architecture of our framework will facilitate future extension to detect cell division.

By enabling experiment scalability, unbiased analysis with advanced accuracy, and an in-depth analysis of large datasets of T-cell dynamics under flow, the computational and analytical framework presented here contributes to the 3R principle when studying the interaction of cells derived from animal models by reducing the number of animals to be sacrificed. ^UFMTrack can be employed for fundamental research of the molecular mechanisms governing immune cell trafficking across a range of vascular beds, screening of pharmaceutical treatments, as well as for personalized medicine based on the evaluation of treatment efficacy on patient-derived T-cell migration behavior on patient-derived endothelial monolayers. One can furthermore envisage that ^UFMTrack can be suitable to study cancer cell metastasis across different endothelial cell monolayers. Eventually, the developed framework can be extended to real-time operation during image acquisition. Combined with transgenic photo-convertible immune cells allowing for the photoconversion of immune cells according to their behavior will allow for subsequent fluorescent cell sorting and scRNA-Seq analysis. Such advanced studies are needed to reveal the role of genetic differences governing T-cell migration regimes.

The source code for the T-cell segmentation models, data preprocessing, training, performance evaluation, and inference scripts are available on GitHub (Vladymyrov, 2025a). This repository also contains win64 binaries used for the watershed-based segmentation of the predicted T cell probability maps and transmigration probability maps. The source code for the under-flow T-cell tracking and migration analysis, along with the scripts used for performance evaluation of the framework, is available on GitHub (Vladymyrov, 2025b). The training phase-contrast data with manual annotations, the reference datasets for histogram normalization, and trained models are available on Zenodo: [https://doi.org/10.5281/zenodo.7489557]. The datasets used for evaluation of our framework are available on Zenodo: [https://doi.org/10.5281/zenodo.7489972, https://doi.org/10.5281/zenodo.7489984, https://doi.org/10.5281/zenodo.7489996, https://doi.org/10.5281/zenodo.7490012, https://doi.org/10.5281/zenodo.7490020, https://doi.org/10.5281/zenodo.7490029].

1. 1. Vladymyrov M
   2. Marchetti L
   3. Aydin S
   4. Soldati S
   5. Mossu A
   6. Pal A
   7. Gueissaz L
   8. Ariga A
   9. Engelhardt B

   (2022) Zenodo

   UFMTrack T-cell segmentation models and training data.

   https://doi.org/10.5281/zenodo.7489557
2. 1. Vladymyrov M
   2. Marchetti L
   3. Aydin S
   4. Soldati S
   5. Mossu A
   6. Pal A
   7. Gueissaz L
   8. Ariga A
   9. Engelhardt B

   (2022) Zenodo

   UFMTrack phase contrast imaging dataset of CD4 T cell interaction with pMBMEC, part 1.

   https://doi.org/10.5281/zenodo.7489972
3. 1. Vladymyrov M
   2. Marchetti L
   3. Aydin S
   4. Soldati S
   5. Mossu A
   6. Pal A
   7. Gueissaz L
   8. Ariga A
   9. Engelhardt B

   (2022) Zenodo

   UFMTrack phase contrast imaging dataset of CD4 T cell interaction with pMBMEC, part 2.

   https://doi.org/10.5281/zenodo.7489984
4. 1. Vladymyrov M
   2. Marchetti L
   3. Aydin S
   4. Soldati S
   5. Mossu A
   6. Pal A
   7. Gueissaz L
   8. Ariga A
   9. Engelhardt B

   (2022) Zenodo

   UFMTrack phase contrast imaging dataset of CD8 T cell interaction with pMBMEC.

   https://doi.org/10.5281/zenodo.7489996
5. 1. Vladymyrov M
   2. Marchetti L
   3. Aydin S
   4. Soldati S
   5. Mossu A
   6. Pal A
   7. Gueissaz L
   8. Ariga A
   9. Engelhardt B

   (2022) Zenodo

   T-cell segmentation, tracking and migration analysis of the CD4 T cells with UFMTrack, part 1.

   https://doi.org/10.5281/zenodo.7490012
6. 1. Vladymyrov M
   2. Marchetti L
   3. Aydin S
   4. Soldati S
   5. Mossu A
   6. Pal A
   7. Gueissaz L
   8. Ariga A
   9. Engelhardt B

   (2022) Zenodo

   T-cell segmentation, tracking and migration analysis of the CD4 T cells with UFMTrack, part 2.

   https://doi.org/10.5281/zenodo.7490020
7. 1. Vladymyrov M
   2. Marchetti L
   3. Aydin S
   4. Soldati S
   5. Mossu A
   6. Pal A
   7. Gueissaz L
   8. Ariga A
   9. Engelhardt B

   (2022) Zenodo

   T-cell segmentation, tracking and migration analysis of the CD8 T cells with UFMTrack.

   https://doi.org/10.5281/zenodo.7490029

1. Preprint

   1. Abadi M
   2. Barham P
   3. Chen J
   4. Chen Z
   5. Davis A
   6. Dean J

   (2016) TensorFlow: a system for large-scale machine learning

   arXiv.

   https://doi.org/10.48550/arXiv.1605.08695

   • Google Scholar
2. 1. Abadier M
   2. Haghayegh Jahromi N
   3. Cardoso Alves L
   4. Boscacci R
   5. Vestweber D
   6. Barnum S
   7. Deutsch U
   8. Engelhardt B
   9. Lyck R

   (2015) Cell surface levels of endothelial ICAM-1 influence the transcellular or paracellular T-cell diapedesis across the blood-brain barrier

   European Journal of Immunology 45:1043–1058.

   https://doi.org/10.1002/eji.201445125

   • PubMed
   • Google Scholar
3. Software

   1. Allan DB
   2. Caswell T
   3. Keim NC
   4. Wel CM
   5. Trackpy VRW

   (2021) Trackpy, version 0.5.0

   GitHub.

   https://github.com/soft-matter/trackpy
4. Preprint

   1. Archit A
   2. Nair S
   3. Khalid N
   4. Hilt P
   5. Rajashekar V
   6. Freitag M
   7. Gupta S
   8. Dengel A
   9. Ahmed S
   10. Pape C

   (2023) Segment anything for microscopy

   bioRxiv.

   https://doi.org/10.1101/2023.08.21.554208

   • Google Scholar
5. 1. Arts JJ
   2. Mahlandt EK
   3. Grönloh ML
   4. Schimmel L
   5. Noordstra I
   6. Gordon E
   7. van Steen AC
   8. Tol S
   9. Walzog B
   10. van Rijssel J
   11. Nolte MA
   12. Postma M
   13. Khuon S
   14. Heddleston JM
   15. Wait E
   16. Chew TL
   17. Winter M
   18. Montanez E
   19. Goedhart J
   20. van Buul JD

   (2021) Endothelial junctional membrane protrusions serve as hotspots for neutrophil transmigration

   eLife 10:e66074.

   https://doi.org/10.7554/eLife.66074

   • PubMed
   • Google Scholar
6. 1. Aydin S
   2. Pareja J
   3. Schallenberg VM
   4. Klopstein A
   5. Gruber T
   6. Page N
   7. Bouillet E
   8. Blanchard N
   9. Liblau R
   10. Körbelin J
   11. Schwaninger M
   12. Johnson AJ
   13. Schenk M
   14. Deutsch U
   15. Merkler D
   16. Engelhardt B

   (2023) Antigen recognition detains CD8⁺ T cells at the blood-brain barrier and contributes to its breakdown

   Nature Communications 14:3106.

   https://doi.org/10.1038/s41467-023-38703-2

   • PubMed
   • Google Scholar
7. 1. Bartholomäus I
   2. Kawakami N
   3. Odoardi F
   4. Schläger C
   5. Miljkovic D
   6. Ellwart JW
   7. Klinkert WEF
   8. Flügel-Koch C
   9. Issekutz TB
   10. Wekerle H
   11. Flügel A

   (2009) Effector T cell interactions with meningeal vascular structures in nascent autoimmune CNS lesions

   Nature 462:94–98.

   https://doi.org/10.1038/nature08478

   • PubMed
   • Google Scholar
8. Conference

   1. Carion N
   2. Massa F
   3. Synnaeve G
   4. Usunier N
   5. Kirillov A
   6. Zagoruyko S

   (2020) End‐to‐End Object Detection with Transformers

   Lecture Notes in Computer Science.

   https://doi.org/10.1007/978-3-030-58452-8_13

   • Google Scholar
9. Conference

   1. Caron M
   2. Touvron H
   3. Misra I
   4. Jegou H
   5. Mairal J
   6. Bojanowski P
   7. Joulin A

   (2021) Emerging properties in self-supervised vision transformers

   2021 IEEE/CVF International Conference on Computer Vision (ICCV).

   https://doi.org/10.1109/ICCV48922.2021.00951

   • Google Scholar
10. 1. Castro Dias M
    2. Odriozola Quesada A
    3. Soldati S
    4. Bösch F
    5. Gruber I
    6. Hildbrand T
    7. Sönmez D
    8. Khire T
    9. Witz G
    10. McGrath JL
    11. Piontek J
    12. Kondoh M
    13. Deutsch U
    14. Zuber B
    15. Engelhardt B

    (2021) Brain endothelial tricellular junctions as novel sites for T cell diapedesis across the blood-brain barrier

    Journal of Cell Science 134:jcs253880.

    https://doi.org/10.1242/jcs.253880

    • PubMed
    • Google Scholar
11. Website

    1. Clockify

    (2022) ClockifyTM ‐ 100% FREE time tracking software

    Accessed October 23, 2022.

    https://clockify.me/
12. 1. Coisne C
    2. Lyck R
    3. Engelhardt B

    (2013) Live cell imaging techniques to study T cell trafficking across the blood-brain barrier in vitro and in vivo

    Fluids and Barriers of the CNS 10:e7.

    https://doi.org/10.1186/2045-8118-10-7

    • PubMed
    • Google Scholar
13. 1. Engelhardt B
    2. Vajkoczy P
    3. Weller RO

    (2017) The movers and shapers in immune privilege of the CNS

    Nature Immunology 18:123–131.

    https://doi.org/10.1038/ni.3666

    • PubMed
    • Google Scholar
14. 1. Engen SA
    2. Valen Rukke H
    3. Becattini S
    4. Jarrossay D
    5. Blix IJ
    6. Petersen FC
    7. Sallusto F
    8. Schenck K

    (2014) The oral commensal Streptococcus mitis shows a mixed memory Th cell signature that is similar to and cross-reactive with Streptococcus pneumoniae

    PLOS ONE 9:e104306.

    https://doi.org/10.1371/journal.pone.0104306

    • PubMed
    • Google Scholar
15. 1. Ershov D
    2. Phan MS
    3. Pylvänäinen JW
    4. Rigaud SU
    5. Le Blanc L
    6. Charles-Orszag A
    7. Conway JRW
    8. Laine RF
    9. Roy NH
    10. Bonazzi D
    11. Duménil G
    12. Jacquemet G
    13. Tinevez JY

    (2022) TrackMate 7: integrating state-of-the-art segmentation algorithms into tracking pipelines

    Nature Methods 19:829–832.

    https://doi.org/10.1038/s41592-022-01507-1

    • PubMed
    • Google Scholar
16. Book

    1. Gonzalez RC
    2. Woods RE

    (2008)

    Digital Image Processing

    Upper Saddle River, N.J: Prentice Hall.

    • Google Scholar
17. 1. Haghayegh Jahromi N
    2. Marchetti L
    3. Moalli F
    4. Duc D
    5. Basso C
    6. Tardent H
    7. Kaba E
    8. Deutsch U
    9. Pot C
    10. Sallusto F
    11. Stein JV
    12. Engelhardt B

    (2019) Intercellular adhesion molecule-1 (ICAM-1) and ICAM-2 differentially contribute to peripheral activation and CNS entry of autoaggressive Th1 and Th17 cells in experimental autoimmune encephalomyelitis

    Frontiers in Immunology 10:3056.

    https://doi.org/10.3389/fimmu.2019.03056

    • PubMed
    • Google Scholar
18. Book

    1. He K
    2. Gkioxari G
    3. Dollár P
    4. Girshick R

    (2017) Mask R‐CNN

    IEEE.

    https://doi.org/10.1109/ICCV.2017.322

    • Google Scholar
19. 1. Holme B
    2. Bjørnerud B
    3. Pedersen NM
    4. de la Ballina LR
    5. Wesche J
    6. Haugsten EM

    (2023) Automated tracking of cell migration in phase contrast images with CellTraxx

    Scientific Reports 13:22982.

    https://doi.org/10.1038/s41598-023-50227-9

    • PubMed
    • Google Scholar
20. 1. Jae Oh D
    2. Min Lee G
    3. Francis K
    4. Palsson BO

    (1999) Phototoxicity of the fluorescent membrane dyes PKH2 and PKH26 on the human hematopoietic KG1a progenitor cell line

    Cytometry 36:312–318.

    https://doi.org/10.1002/(sici)1097-0320(19990801)36:4<312::aid-cyto5>3.0.co;2-v

    • PubMed
    • Google Scholar
21. Conference

    1. Kingma DP
    2. Ba J

    (2014)

    Adam: a method for stochastic optimization

    3rd International Conference on Learning Representations, ICLR 2015 - Conference Track Proceedings.

    • Google Scholar
22. Preprint

    1. Kraus O
    2. Kenyon-Dean K
    3. Saberian S
    4. Fallah M
    5. McLean P
    6. Leung J

    (2024) Masked autoencoders for microscopy are scalable learners of cellular biology

    arXiv.

    https://doi.org/10.48550/arXiv.2404.10242

    • Google Scholar
23. 1. Lyck R
    2. Engelhardt B

    (2012) Going against the tide--how encephalitogenic T cells breach the blood-brain barrier

    Journal of Vascular Research 49:497–509.

    https://doi.org/10.1159/000341232

    • PubMed
    • Google Scholar
24. 1. Lyck R
    2. Nishihara H
    3. Aydin S
    4. Soldati S
    5. Engelhardt B

    (2023) Modeling brain vasculature immune interactions in vitro

    Cold Spring Harbor Perspectives in Medicine 13:a041185.

    https://doi.org/10.1101/cshperspect.a041185

    • PubMed
    • Google Scholar
25. 1. Marchetti L
    2. Engelhardt B

    (2020) Immune cell trafficking across the blood-brain barrier in the absence and presence of neuroinflammation

    Vascular Biology 2:H1–H18.

    https://doi.org/10.1530/VB-19-0033

    • PubMed
    • Google Scholar
26. 1. Mossu A
    2. Rosito M
    3. Khire T
    4. Li Chung H
    5. Nishihara H
    6. Gruber I
    7. Luke E
    8. Dehouck L
    9. Sallusto F
    10. Gosselet F
    11. McGrath JL
    12. Engelhardt B

    (2019) A silicon nanomembrane platform for the visualization of immune cell trafficking across the human blood-brain barrier under flow

    Journal of Cerebral Blood Flow and Metabolism 39:395–410.

    https://doi.org/10.1177/0271678X18820584

    • PubMed
    • Google Scholar
27. 1. Nishihara H
    2. Soldati S
    3. Mossu A
    4. Rosito M
    5. Rudolph H
    6. Muller WA
    7. Latorre D
    8. Sallusto F
    9. Sospedra M
    10. Martin R
    11. Ishikawa H
    12. Tenenbaum T
    13. Schroten H
    14. Gosselet F
    15. Engelhardt B

    (2020) Human CD4⁺ T cell subsets differ in their abilities to cross endothelial and epithelial brain barriers in vitro

    Fluids and Barriers of the CNS 17:3.

    https://doi.org/10.1186/s12987-019-0165-2

    • PubMed
    • Google Scholar
28. 1. Nishihara H
    2. Gastfriend BD
    3. Kasap P
    4. Palecek SP
    5. Shusta EV
    6. Engelhardt B

    (2021) Differentiation of human pluripotent stem cells to brain microvascular endothelial cell-like cells suitable to study immune cell interactions

    STAR Protocols 2:100563.

    https://doi.org/10.1016/j.xpro.2021.100563

    • PubMed
    • Google Scholar
29. 1. Nourshargh S
    2. Hordijk PL
    3. Sixt M

    (2010) Breaching multiple barriers: leukocyte motility through venular walls and the interstitium

    Nature Reviews. Molecular Cell Biology 11:366–378.

    https://doi.org/10.1038/nrm2889

    • PubMed
    • Google Scholar
30. 1. Nourshargh S
    2. Alon R

    (2014) Leukocyte migration into inflamed tissues

    Immunity 41:694–707.

    https://doi.org/10.1016/j.immuni.2014.10.008

    • PubMed
    • Google Scholar
31. 1. Pedregosa F
    2. Varoquaux G
    3. Gramfort A
    4. Michel V
    5. Thirion B
    6. Grisel O

    (2011)

    Scikit-learn: machine learning in python

    Journal of Machine Learning Research 12:2825–2830.

    • Google Scholar
32. Website

    1. Perron L
    2. Furnon V

    (2021) OR‐Tools

    Accessed March 19, 2025.

    https://developers.google.com/optimization/
33. 1. Purschke M
    2. Rubio N
    3. Held KD
    4. Redmond RW

    (2010) Phototoxicity of Hoechst 33342 in time-lapse fluorescence microscopy

    Photochemical & Photobiological Sciences 9:1634–1639.

    https://doi.org/10.1039/c0pp00234h

    • PubMed
    • Google Scholar
34. Preprint

    1. Ronneberger O
    2. Fischer P
    3. Brox T

    (2015) U‐Net: convolutional networks for biomedical image segmentation

    arXiv.

    https://doi.org/10.48550/arXiv.1505.04597

    • Google Scholar
35. 1. Rudolph H
    2. Klopstein A
    3. Gruber I
    4. Blatti C
    5. Lyck R
    6. Engelhardt B

    (2016) Postarrest stalling rather than crawling favors CD8(+) over CD4(+) T-cell migration across the blood-brain barrier under flow in vitro

    European Journal of Immunology 46:2187–2203.

    https://doi.org/10.1002/eji.201546251

    • PubMed
    • Google Scholar
36. 1. Rueden CT
    2. Schindelin J
    3. Hiner MC
    4. DeZonia BE
    5. Walter AE
    6. Arena ET
    7. Eliceiri KW

    (2017) ImageJ2: ImageJ for the next generation of scientific image data

    BMC Bioinformatics 18:529.

    https://doi.org/10.1186/s12859-017-1934-z

    • PubMed
    • Google Scholar
37. 1. Saetzler RK
    2. Jallo J
    3. Lehr HA
    4. Philips CM
    5. Vasthare U
    6. Arfors KE
    7. Tuma RF

    (1997) Intravital fluorescence microscopy: impact of light-induced phototoxicity on adhesion of fluorescently labeled leukocytes

    Journal of Histochemistry & Cytochemistry 45:505–513.

    https://doi.org/10.1177/002215549704500403

    • Google Scholar
38. 1. Sallusto F
    2. Lenig D
    3. Mackay CR
    4. Lanzavecchia A

    (1998) Flexible programs of chemokine receptor expression on human polarized T helper 1 and 2 lymphocytes

    The Journal of Experimental Medicine 187:875–883.

    https://doi.org/10.1084/jem.187.6.875

    • PubMed
    • Google Scholar
39. Book

    1. Schiegg M
    2. Hanslovsky P
    3. Kausler BX
    4. Hufnagel L
    5. Hamprecht FA

    (2013) Conservation Tracking

    IEEE Xplore.

    https://doi.org/10.1109/ICCV.2013.364

    • Google Scholar
40. Preprint

    1. Schmidt U
    2. Weigert M
    3. Broaddus C
    4. Myers G

    (2018) Cell detection with star‐convex polygons

    arXiv.

    https://doi.org/10.1007/978-3-030-00934-2_30

    • Google Scholar
41. 1. Shulman Z
    2. Shinder V
    3. Klein E
    4. Grabovsky V
    5. Yeger O
    6. Geron E
    7. Montresor A
    8. Bolomini-Vittori M
    9. Feigelson SW
    10. Kirchhausen T
    11. Laudanna C
    12. Shakhar G
    13. Alon R

    (2009) Lymphocyte crawling and transendothelial migration require chemokine triggering of high-affinity LFA-1 integrin

    Immunity 30:384–396.

    https://doi.org/10.1016/j.immuni.2008.12.020

    • PubMed
    • Google Scholar
42. 1. Singer BA

    (2017) The role of natalizumab in the treatment of multiple sclerosis: benefits and risks

    Therapeutic Advances in Neurological Disorders 10:327–336.

    https://doi.org/10.1177/1756285617716002

    • PubMed
    • Google Scholar
43. 1. Soldati S
    2. Bär A
    3. Vladymyrov M
    4. Glavin D
    5. McGrath JL
    6. Gosselet F
    7. Nishihara H
    8. Goelz S
    9. Engelhardt B

    (2023) High levels of endothelial ICAM-1 prohibit natalizumab mediated abrogation of CD4⁺ T cell arrest on the inflamed BBB under flow in vitro

    Journal of Neuroinflammation 20:123.

    https://doi.org/10.1186/s12974-023-02797-8

    • PubMed
    • Google Scholar
44. 1. Steiner O
    2. Coisne C
    3. Cecchelli R
    4. Boscacci R
    5. Deutsch U
    6. Engelhardt B
    7. Lyck R

    (2010) Differential roles for endothelial ICAM-1, ICAM-2, and VCAM-1 in shear-resistant T cell arrest, polarization, and directed crawling on blood-brain barrier endothelium

    Journal of Immunology 185:4846–4855.

    https://doi.org/10.4049/jimmunol.0903732

    • PubMed
    • Google Scholar
45. 1. Steinman L
    2. Zamvil SS

    (2005) Virtues and pitfalls of EAE for the development of therapies for multiple sclerosis

    Trends in Immunology 26:565–571.

    https://doi.org/10.1016/j.it.2005.08.014

    • PubMed
    • Google Scholar
46. 1. Stringer C
    2. Wang T
    3. Michaelos M
    4. Pachitariu M

    (2021) Cellpose: a generalist algorithm for cellular segmentation

    Nature Methods 18:100–106.

    https://doi.org/10.1038/s41592-020-01018-x

    • PubMed
    • Google Scholar
47. 1. Tsai HF
    2. Gajda J
    3. Sloan TFW
    4. Rares A
    5. Shen AQ

    (2019) Usiigaci: instance-aware cell tracking in stain-free phase contrast microscopy enabled by machine learning

    SoftwareX 9:230–237.

    https://doi.org/10.1016/j.softx.2019.02.007

    • Google Scholar
48. 1. Ulicna K
    2. Vallardi G
    3. Charras G
    4. Lowe AR

    (2021) Automated deep lineage tree analysis using a bayesian single cell tracking approach

    Frontiers in Computer Science 3:e559.

    https://doi.org/10.3389/fcomp.2021.734559

    • Google Scholar
49. 1. Vladymyrov M
    2. Haghayegh Jahromi N
    3. Kaba E
    4. Engelhardt B
    5. Ariga A

    (2020) VivoFollow 2: distortion-free multiphoton intravital imaging

    Frontiers in Physics 7:eo222.

    https://doi.org/10.3389/fphy.2019.00222

    • Google Scholar
50. Software

    1. Vladymyrov M

    (2025a) UFMSegm, version 960bbe9

    GitHub.

    https://github.com/neworldemancer/UFMSegm
51. Software

    1. Vladymyrov M

    (2025b) UFMTrack, version e9d94c1

    GitHub.

    https://github.com/neworldemancer/UFMTrack
52. Preprint

    1. Wang CY
    2. Bochkovskiy A
    3. Liao HYM

    (2022) YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors

    arXiv.

    https://doi.org/10.48550/arXiv.2207.02696

    • Google Scholar
53. 1. Wimmer I
    2. Tietz S
    3. Nishihara H
    4. Deutsch U
    5. Sallusto F
    6. Gosselet F
    7. Lyck R
    8. Muller WA
    9. Lassmann H
    10. Engelhardt B

    (2019) PECAM-1 stabilizes blood-brain barrier integrity and favors paracellular T-cell diapedesis across the blood-brain barrier during neuroinflammation

    Frontiers in Immunology 10:711.

    https://doi.org/10.3389/fimmu.2019.00711

    • PubMed
    • Google Scholar
54. Conference

    1. Xie R
    2. Pang K
    3. Bader GD
    4. Wang B

    (2023) MAESTER: masked autoencoder guided segmentation at pixel resolution for accurate, self-supervised subcellular structure recognition

    2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 3292–3301.

    https://doi.org/10.1109/CVPR52729.2023.00321

    • Google Scholar
55. 1. Zielinski CE
    2. Mele F
    3. Aschenbrenner D
    4. Jarrossay D
    5. Ronchi F
    6. Gattorno M
    7. Monticelli S
    8. Lanzavecchia A
    9. Sallusto F

    (2012) Pathogen-induced human TH17 cells produce IFN-γ or IL-10 and are regulated by IL-1β

    Nature 484:514–518.

    https://doi.org/10.1038/nature10957

    • PubMed
    • Google Scholar

Author details

1. Mykhailo Vladymyrov

   1. Theodor Kocher Institute, Bern, Switzerland
   2. Data Science Lab, Bern, Switzerland

   Contribution

   Conceptualization, Data curation, Software, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing – original draft, Project administration, Writing – review and editing

   For correspondence

   mykhailo.vladymyrov@unibe.ch

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-5720-9740

2. Luca Marchetti

   Theodor Kocher Institute, Bern, Switzerland

   Contribution

   Data curation, Investigation, Methodology, Writing – review and editing

   Competing interests

   No competing interests declared

3. Sidar Aydin

   Theodor Kocher Institute, Bern, Switzerland

   Contribution

   Data curation, Formal analysis, Methodology, Writing – review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-7791-7931

4. Sasha GN Soldati

   Theodor Kocher Institute, Bern, Switzerland

   Contribution

   Data curation, Formal analysis, Methodology, Writing – review and editing

   Competing interests

   No competing interests declared

5. Adrien Mossu

   Theodor Kocher Institute, Bern, Switzerland

   Contribution

   Data curation

   Competing interests

   No competing interests declared

6. Arindam Pal

   Theodor Kocher Institute, Bern, Switzerland

   Contribution

   Formal analysis

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-3784-7868

7. Laurent Gueissaz

   Theodor Kocher Institute, Bern, Switzerland

   Contribution

   Formal analysis

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0009-0008-8695-7992

8. Akitaka Ariga

   Laboratory for High Energy Physics, University of Bern, Bern, Switzerland

   Contribution

   Conceptualization, Resources, Supervision, Funding acquisition, Methodology, Project administration

   Competing interests

   No competing interests declared

9. Britta Engelhardt

   Theodor Kocher Institute, Bern, Switzerland

   Contribution

   Conceptualization, Resources, Supervision, Funding acquisition, Writing – original draft, Project administration, Writing – review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-3059-9846

• 515

  views

• 37

  downloads

• 0

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.