IntechOpen
IntechOpen

Open access peer-reviewed article

A Simple Method for Visualization and Quantitative Evaluation of Single and Collective Cell Migration in Bright Field Microscopy Images

Yoko Kato

This Article is part of Precision Medicine Section

Article metrics overview

258 Article Downloads

View Full Metrics

Article Type: Research Paper

Date of acceptance: August 2024

Date of publication: August 2024

DoI: 10.5772/dmht.20240001

copyright: ©2024 The Author(s), Licensee IntechOpen, License: CC BY 4.0

Download for free

Table of contents

Introduction
A method for migration detection and its application
Results
Discussion
Conclusion
Conflict of interest

Abstract

Cell migration is observed in various cases such as embryonic and lesion developments. The migration directly influences the phenomena around the migration path. Bright field microscopy, generally used for cell observation, is effective in tracking cell movement, but the detection of the cell outline via image processing methods partially fails. In this study, a simple method, utilizing the intensity fluctuation of the image caused by the passage of a cell as a parameter for evaluation of the cell movement, has been proposed to visualize the region where the cell passed and quantitatively evaluate its frequency by the fluctuation pattern. The map that depicts the activity of cell movement is created, and the geometrical characteristics of each region and fluctuation degree are quantitatively shown. Because detection of the cell outline is not necessary, this method is applicable to collective cells as well as single cells. When this method was applied to the images of hemocytes in Halocynthia roretzi (Ascidiacea), which were composed of single and collective cells and showed complex patterns in movement, the map and quantitative parameters for the fluctuation were successfully obtained. The method can be improved by the calibration of intensity distribution and applied to cells in various species.

Keywords

  • collective cell

  • single cell

  • cell motility

  • bright field microscopy

  • visualization

  • quantitative parameter

Author information

Introduction

Image processing methods for detecting cellular regions in a microgram, such as bright field and fluorescent microscopy images, are essential for cell investigation. Therefore, various methods for segmentation and boundary detection have been proposed [1–3]. Recently, segmentation challenges have been carried out [4]. Deep learning has been helpful for developing this method; its limited application and large training datasets would be a constraint [1]. In the meantime, a segmentation method based on deep learning with pretrained models, Cellpose, has been developed [5, 6]. Adequate preparation for pretrained models leads to continual improvement as well as user-friendly operations. The images of cells in major cell lines and general observation methods or assays are applicable, but those of novel cell types and evaluation methods would be hardly analyzed. While fluorescent labeling, visualizing specific substances, is useful to observe the cellular phenomena properly, the labeling shows disadvantages such as phototoxicity [2, 7]. Bright field microscopy does not cause such damages so that it is widely used from the ordinary observation processes in cell collection and culture to the evaluation of cell migration. Cell migration in the wound healing process, observed in non-labeling and bright field microscopy, has been evaluated by the detection of cellular regions [3, 8–10]: The boundary between cellular and empty regions, which corresponds to the wound outline, has been detected by threshold setting based on the fast discrete curvelet transform [8] and the standard deviation of the intensity in each divided region [9], and a support vector machine based on intensity characteristics [10]. Although the size, shape, and position of the cellular region are detected by these methods [8–10], the behavior of each cell is not observed. Furthermore, CellTraxx [11], based on ImageJ macro, has successfully detected each cell of the major cell lines in phase-contrast microscopy images. While the accuracy of the segmentation, which directly influences the analyses in cell morphology, should be accomplished in the first place, the image characteristics, such as a pattern of the intensity distribution, are influenced by cell conditions so that the analyses could be useful for the investigation of cells.

Cell migration is observed in various phenomena such as embryonic and lesion development [12]. The direction of migration is controlled by chemotaxis, haptotaxis, and durotaxis [13, 14]. Besides, computational models related to the extracellular matrix [15] and competition between chemotaxis and durotaxis [16] have been proposed. Collective cell migration is also observed in various phenomena as cell migration is. In addition to the taxis and mechanical environment, the hierarchy and interaction between the cells should be considered [17–26]. Considering that single and collective cell migration occurs in various places and cell types, cell types related to migration will be found more. A general method to evaluate non-labeling, bright field microscopy images of single and collective cell migration would promote the finding.

Considering open circulatory systems, blood cells migrate in the blood vessel and tissue. Halocynthia roretzi (Ascidiacea) (Figure 1) has an open circulatory system [27]. H. roretzi is entirely covered with its tunic, whose thickness is kept constant with continuous removal and proliferation [28]. The tunic has blood vessels, nerve fibers, and various types of cells [29] so that it is different from a shell [27]. As the hemocytes in ascidians are the pivot of their defense systems [30], the secretions from the various types of hemocytes in H. roretzi have been found [31–38]. While the categories of the hemocytes in H. roretzi have been reported previously [39, 40], the movement of the hemocyte has been also observed [41, 42]. However, the patterns of the movement have been neither quantitatively analyzed nor visualized. The interaction among the hemocytes, which would influence the defense system, has been barely investigated.

Figure 1.

The sample of H. roretzi. The tunic covers the entire body.

Cellulose I𝛽, predominantly found in higher plants, is in the tunic of H. roretzi with high purity and crystallinity [43, 44]. The elasticity of its whisker is high (143 GPa) [45]. Various sulfated polysaccharides, including the hydrophilic substance such as  sulfated chitin in the tunic [46, 47], have been found [48–52]. Moreover, many kinds of carotenoids, which are  antioxidant compounds related to light, have been reported [53–64]. In the meantime, proteins that directly influence the mechanical characteristics of the tunic, including pseudokeratin [65], collagen [66], elastic fiber [66], and 𝛼-smooth muscle actin [66] have been found. Considering the contractility of 𝛼-smooth muscle actin [66], the nervous system [29, 66], and hydrophilicity [46, 47], the active contraction with changes in water content, caused by various stimuli [66–70], is agreeable. The hemocytes in the tunic were obtained by centrifugation [70], and actively moved [71]. Part of the hemocytes behaved as one group [70]. The standard deviation of the intensity at each pixel through a time series was used for tracking cell movements [71] but not for evaluation. Considering that the mechanical environment in the tunic would be maintained [72], the hemocytes could move freely as well as widely throughout the tunic so that its condition could be maintained properly.

In this study, an image processing method to evaluate single and collective cell migration in bright field microscopy images has been developed. When a cell migrates from one place to another, the intensity along the path fluctuates between the background and the intracellular region. In this method, the fluctuation is used as a parameter to evaluate cell migration. The fluctuation in intensity is generally related to the migration so that this method is not limited to a specific cell line. Furthermore, a foreign body, one of the artifacts, can be removed because of slight fluctuation. Training and parameter modulations are not necessary; just the images that show cell or collective migration are needed. The method is directly influenced by inhomogeneity in intensity so that the result shows a general tendency in each migration, such as the direction and space size in migration rather than a velocity or acceleration map. The method was applied to the hemocytes of H. roretzi because they would behave as a single cell or collective cell throughout the tunic under a stable mechanical condition.

A method for migration detection and its application

A method for migration detection

The images of bright field microscopy are used. When a cell passes through the region, the intensity of each pixel in the region fluctuates between the cell and background. That is, the region related to cell migrations shows fluctuation in intensity. The more frequently the cell passes through the pixel, the larger the increase in unbiased variance at the pixel. In the meantime, the fluctuation in intensity at the static cells and the background is low. The average and unbiased variance of the intensity in each pixel is calculated through a time series in order to categorize the regions of the static cell, migrating cell, and the background. The pixel whose unbiased variance is small corresponds to a static cell or the background. In order to identify the static cell and background, the average in each pixel is calculated. The pixel that the cell continuously passes through shows a small fluctuation. In such a case, a region of high unbiased variance area appears with a void. However, the static and this “void” are easily identified because the void is surrounded by a large region of high unbiased variance. Such a large region has not been found around the static cell even when the cell vibrates. In the following description, the region that the cells pass through and shows high unbiased variance is called the “dynamic region” and that for the static cell and the “void” on the migration track is the “static region”.

Just after the extraction, the hemocyte was spindle-shaped and actively moving. However, on the second day, part of the cells became rounded, especially those at the second passage, and moved only slightly. Because this method is developed for single and collective cell migration, these rounded cells were removed from the data analysis (5). However, the rounded cells at the second passage were difficult to remove from the analysis; so the images of these two samples were not used in the data analysis.

All the image analyses were implemented in programming language C while the free software for image processing and analysis (ImageJ, 1.53k, https://imagej.net/ij/index.html) was used to extract the raw data from each image and display its result.

Each process of this method is as described in the following:

  1. Prepare bright field microscopy images:

    The range of intensity decided by the image type is checked (Xmax, Xmin). For example, the maximum and minimum intensities in the 8-bit image are 0 and 255, respectively.

  2. Enhance the contrast in each image:

    Gray scale transformation: The maximum and minimum intensities (xmax, xmin) in the image are linearly changed to those in the image type.

    (x and x′, intensities before and after this process, respectively).

  3. Calculate the average and unbiased variance at each pixel:

    If the region of the cell is darker than the background, the inverted average is used for detecting the static region, instead of the average.

    • Average (Aij)

    • Inverted average

    • Unbiased variance (Sij)

      (i and j, x and y coordinates; t, slice number; tn, the total slice number; xij(t), the intensity at (i, j) of slice t).

  4. Identify the static and dynamic regions:

    • Binarization: The threshold is decided by the method previously proposed [73].

    • Region selection: The region including the third quartile in intensity is selected.

      After the aforementioned processes, when the area of the dynamic region is smaller than that of the cell, or apparently stable, the region is not evaluated.

  5. Analyze the unbiased variance at each pixel in the dynamic regions:

    The parameters for the shape and intensity in each region are evaluated as follows.

    • Shape: the area (A) and lengths from the centroid (the average (Lave) and standard deviation (Lsd))

    • Intensity: the average (Ave), standard deviation (SD), and length-weighted averages (, Wave2) (n, the area in each region; , the position vector from the centroid for each pixel in the region)

  6. Enhance the contrast in the images for the average and unbiased variance:

    Gray scale transformation: The intensity in the static region and dynamic region, whose intensity is more than 0, is linearly transformed as the smallest and largest intensities in these regions are changed to Xmin + 10 and Xmax, respectively.

  7. Visualize the static and dynamic regions:

    • Individual display (static and dynamic regions are displayed separately): pseudocolor

    • Complex display (static and dynamic regions appear in the same image): color (red, static region; green, dynamic region).

Cells and images for the application

In order to check the abilities of the method, it was applied to bright field microscopy images of H. roretzi hemocytes. The samples of H. roretzi (n = 2) were obtained from Marutaki Suisan (Miyagi, Japan). The tunic was separated from the inner organs by tweezers and trimming blades (feather trimming blade; Feather Safety Razor, Co. Ltd., Osaka, Japan) and put in artificial seawater (Reef Crystals, Aquarium Systems, Sarrebourg, France) at a temperature of approximately 10 °C overnight. The tunic was not mixed with that of another sample in order to avoid damage to the hemocytes. The tunic was cut into small pieces by the trimming blades. The pieces of the tunic with artificial seawater in a tube was centrifuged (1000 g (g, gravity), 7 min) to obtain hemocytes [32, 37]. The cells were seeded in eight dishes containing the medium (10% fetal bovine serum (F2442, Merck (Sigma-Aldrich), Germany) and 1% penicillin–streptomycin (P4333, Merck (Sigma-Aldrich), Germany) in the seawater filtered by a 0.22 μm pore size, PVDF membrane (S2GVU02RE, Merck (Millipore), Germany)) at a temperature of approximately 10 °C overnight. The cells were observed in bright field microscopy (LSM 5 LIVE, Zeiss, Germany). The image whose resolution was 0.656 μm/pixel was taken every 2 s for about 2 min. The cells in four dishes were also observed the next day (medium exchange, three dishes; subculture, one dish). The method was applied to the images of 13 samples.

Statistical analyses

The linear approximation between the two parameters was carried out by the least squares method. The correlation coefficient in each line was statistically evaluated by the F-test. The significant level was 0.05. The statistical analyses were carried out by spreadsheet software (Excel, Office 2021, Microsoft).

Results

Figure 2.

Enhancement in contrast of the image where the group of hemocytes moved. Right, the original image; left, the image with enhanced contrast. Scale bar, 20 μm.

Figure 3.

Enhancement in contrast of the image where the hemocytes moved separately. Right, the original image; left, the image with enhanced contrast. Scale bar, 20 μm.

Image processing was carried out successfully for all the images. The resultant images were obtained from all the sample images; the images for the subculture showed many rounded cells beside spindle-shaped cells. The rounded cell moved only slightly while the spindle-shaped cell moved actively. Figures 2 and 3 show part of the images, which is mainly composed of collective and individual behaviors, respectively. The numbers of dynamic regions were 12 and 27, respectively. Comparing the original image and the modified image whose contrasts were enhanced by process (2) in Section 2.1, the boundary between the cell and background became clear for both of the single and collective cells. Figures 4 and 5 show the images for the static region (a), dynamic region (b), and merger of these two regions (c) corresponding to Figures 2 and 3, respectively. In order to compare these results with the original images, (d) images are shown in Figures 4 and 5. The static and dynamic regions, whose intensity was shown by pseudocolor, are helpful to examine which region the cell actively passed through or stayed. Also, the image merging the static and dynamic regions is available to check the interaction between these regions. In all the images, the visualization for the static and dynamic regions has been successfully carried out. However, the region where the cell continuously passes through shows little change in intensity so that the unbiased variance is close to zero and the average is almost the same as that of the static cell. That is, such a region shows no difference from that of a static cell. When the static cell shows vibration, the unbiased variance in its surroundings becomes high. The difference between the static cell with vibration and the migrating cell is the size of the region with high unbiased variance surrounding the lower one. The region of the static cell is smaller than that of the migrating cell. As Figures 4(c) and 5(c) show, the difference is easily recognized. The area ratio between the neighboring regions with higher and lower unbiased variances would be useful to decide whether the region corresponds to the migrating cell.

Figure 4.

Visualization in migration. These images correspond to those in Figure 2: (a) static region; (b) dynamic region; (c) static region (Red) and dynamic region (Green); (d) original.

Figure 5.

Visualization in migration. These images correspond to those in Figure 3: (a) static region; (b) dynamic region; (c) static region (Red) and dynamic region (Green); (d) original.

Figure 6.

Fluctuation patterns in intensity. The modulated average (red) and unbiased variance (green) are shown: (a) and (b), corresponding to Figures 4(c) and 5(c), respectively; (c) the sample with many stable cells. The boxes show the typical migration patterns: a1, a cell with a linear movement; a2, a large collective cell; b1, a cell with turning; b2, a small collective cell; c1, a complex of individual cell movements; c2, a stable cell with vibration; c3, a collective cell with size between a2 and b2. The differences between the average and variance in the boxes along the two lines, passing around the center of the void (a1) or the red region (others) and parallel to the x and y axes, are shown as graphs: 1, 2, and 3, blue, orange, and green; solid and dotted lines, parallel to the x and y axes.

Comparing the static and dynamic regions, the images of the average and unbiased variance modulated in the same range, from 10 to 255, and the difference between these parameters are shown in Figure 6. Figures 6(a) and 6(b) are the same as Figures 4(c) and 5(c), respectively. Figure 6(c) is the result of the sample, which had many stable cells. The boxes in the images, from a1 to c3, in Figure 6, are the typical migration patterns: a1, a cell with a linear path; a2, a large collective cell; b1, a cell with turning; b2, a small collective cell; c1, a complex of individual cell movement; c2, a stable cell with vibration; c3, a collective cell with size between a2 and b2. The differences between the average and variance along the two lines, which pass around the center of the void in a1 and the red region in other boxes, and run parallel to the x and y axes, respectively, are shown by the average of that along these two lines and their adjacent lines. While the difference was symmetric in the stable cell (c2), that in the single or collective cell migration in other boxes was asymmetric. Except the stable cell, part of the migrating cells are bright red, which indicates that the intensity is kept as cellular regions. In such a case, the frequency of passing through the pixel would be greater than that of photographing or the cell would stay for a while to change the direction of movement. Part of the region a1 is low in both of the average and variance. The homogeneous intensity in the cell, whose intensity was close to that of the background, would continuously pass through the region. As these results show, it is possible to evaluate the characteristics of the migration in each region manually. The evaluation is important to investigate the migration. However, considering that the manual evaluation for all the regions is not practical because of the number of cells in each image, a method to carry out this process automatically is necessary. The aforementioned characteristics in each typical migration are helpful for the development of the method.

This method does not show the velocity and acceleration for each migration, but the results about the relative level of migration in the same image sequence. In order to evaluate the migration pattern in the image totally, the variance in the same image sequence was evaluated. Figures 7 and 8 show the results of collective cell in Figure 2. Figure 7 shows the relationship between the parameters related to shape and the unbiased variance of intensity. On the assumption that the region is an ellipse whose major and minor axes are Lave + Lsd and Lave − Lsd, respectively, the area (Aest) was estimated by the following equation:

As Figure 7(a) shows, the area in each region was estimated well by the length from the centroid. Also, the average and standard deviation of the length from the centroid was well correlated in Figure 7(b). These results indicate that the region which the cell passed through could be approximated as an ellipse so that the direction of migration could approximate to that of the major axis. Considering that the size of the collective cell was varied and part of the regions had a void, these results also show that shape approximation by an ellipse would be robust in the evaluation of the migration pattern. In the meantime, the average and standard deviation of the variance, which correspond to the frequency of passage and its range, were poorly correlated with each other (Figure 7(c)) and not influenced by the area size related to the region of migration (Figure 7(d)). These parameters, which show the pattern of passage, would not be influenced by each other, but individually determined. That is, even if the cell travels in the broader region, the frequency of passing through each pixel and its range would not be influenced. While these two parameters are related to the velocity of the migration, they are directly influenced by the intensity distribution pattern of the cellular region. Hence, the average and standard deviation would be associated with the velocity, but partially. In order to convert these parameters to velocity, how the distribution pattern of the intensity in the cellular region would influence the parameters should be investigated. Considering that the larger area size of migration is composed of the larger number of cells and the sample of this result is mainly composed of collective cells, the result would indicate interactions between the cells, whose patterns would not show similarity in collective cells with different sizes. Also, lack of similarity would influence the average and standard deviation of the variance.

Figure 7.

Quantitative analysis of collective cell migration in shape and unbiased variance. The results correspond to those in Figure 2: (a) the relationship between the area (A) and the estimated area based on Equation (7) (estimation); (b) the relationship between the average and standard deviation of the lengths from the centroid (Lave and Lsd); (c) the relationship between the average and standard deviation of the unbiased variance (Ave and SD); (d) the relationship between the area (A) and the average and standard deviation of the unbiased variance (Ave and SD).

In Figure 8, the magnitude of the vector Wave1 was poorly correlated with the area and the variance of intensity. The magnitude of the vector Wave1 shows the characteristics of the distribution in the variance and region size with direction. If each pixel shows the unbiased variance symmetrically around the centroid, Wave1 becomes 0. Hence, the distribution pattern was not symmetric around the centroid, and would be hardly influenced by the region size and the frequency of passing through the region. Also, the angle between the vector Wave1 and the x coordinate was not related to the parameters of the area and unbiased variance. Because the area size is related to the number of cells, which corresponds to the size of the collective cell, the spatial distribution of the variance would not be influenced by the size of the collective cell. Considering that the average and standard deviation of the variance would be hardly influenced by the size of the collective cell as Figure 7(d) shows, this result would be agreeable. In the meantime, Wave2 correlated well with the average and standard deviation of the variance of intensity while the area indicated poor correlation. As the influence of the difference in area size was not found in the result of Wave1, Wave2 would not be changed by the difference in the size of the collective cell. While Wave1 shows the distribution pattern with the distance and direction from the center of the area, Wave2 evaluates that just by the distance. Hence, this result shows that the migration pattern would be influenced by the distance from the center of the migration, not by the difference in the size.

Figure 8.

Quantitative analysis of collective cell migration in and Wave2. The results correspond to those in Figure 2: (a) the relationship between the magnitude of () and the area (A); (b) the relationship between the magnitude of () and the average and standard deviation of the unbiased variance (Ave and SD); (c) the relationship between Wave2 and the area (A); (d) the relationship between Wave2 and the average and standard deviation of the unbiased variance (Ave and SD).

Figures 9 and 10 show the results of cell behaviors in Figure 3. Figures 9(a) and 9(b) show that each area could be approximated as an ellipse. This result agrees with that of Figures 7(a) and 7(b). Considering that Figure 9 is the result for the sample, which had more single cells than that shown by Figure 7, the approximation would be reasonable for both of the categories. Also, Figure 9(d) shows that the average and standard deviation of the variance were poorly correlated with the area as Figure 7(d) shows. The result would be influenced by the mixture of single and collective cells. However, the average and standard deviation of the variance in Figure 9(c) were well correlated with each other while those in Figure 7(c) did not. That is, as the velocity of the migration is increased, the range of velocity is increased. The result shows that the modulation of velocity in single cells would be different from that in collective cells. In the meantime, Figure 10 shows that all the results related to Wave1 and Wave2 agreed with that of Figure 8. The result shows that the distribution pattern of the variance around the center of the area would be kept in single and collective cells.

Figure 9.

Quantitative analysis of collective cell migration in shape and unbiased variance. The results correspond to those in Figure 3: (a) the relationship between the area (A) and the estimated area based on Equation (7) (estimation); (b) the relationship between the average and standard deviation of the lengths from the centroid (Lave and Lsd); (c) the relationship between the average and standard deviation of the unbiased variance (Ave and SD); (d) the relationship between the area (A) and the average and standard deviation of the unbiased variance (Ave and SD).

Figure 10.

Quantitative analysis of collective cell migration in and Wave2. The results correspond to those in Figure 3: (a) the relationship between the magnitude of () and the area (A); (b) the relationship between the magnitude of () and the average and standard deviation of the unbiased variance (Ave and SD); (c) the relationship between Wave2 and the area (A); (d) the relationship between Wave2 and the average and standard deviation of the unbiased variance (Ave and SD).

The data analysis was carried out for all the sample images except two, containing a lot of rounded hemocytes. Although the number of samples was small to describe the representative characteristics of the hemocytes, the analysis described the characteristics in migration quantitatively in each cell sample. The proposed method provided the results, which quantitatively show both the similar and different characteristics in behaviors of single and collective cells.

Discussion

In this study, a method to evaluate and visualize single and collective cell migration by intensity fluctuation in bright field microscopy images was developed. To check whether the method is useful to evaluate migration, images of the hemocytes of H. roretzi, which are part of the open circulatory system and the defense system and move throughout the tunic, were used. The hemocytes behaved as a collective cell as well as a single cell so that the proposed method was applied to both cases. As a result, the pattern of migration was successfully visualized as well as quantitatively examined. In the static cells, high variance was symmetrically distributed while in moving cells, it was not. Also, the region where the cell continuously passes through showed low unbiased variance. In such a case, the region with low unbiased variance was much smaller than its surroundings with high unbiased variance. When the sizes of the neighboring regions with the low and high unbiased variances are compared, it can be identified whether the region shows cell migration. While the spindle-shaped hemocyte of H. roretzi moved actively, the rounded hemocyte moved only slightly. The circularity of the region will be also helpful to identify the migrating and static cells. The distribution of the unbiased variance of intensity in each image was used as a parameter to show the relative degree of cell migration; so the speed of cell migration is not clear. In order to transform the unbiased variance to the velocity, the image for calibration in each observation of cell migration is necessary. Moreover, the distribution pattern of intensity in the cellular region directly influences the variance so that a compensation method for the pattern is necessary to carry out an accurate evaluation of the migration. When the variance is converted to velocity, the distribution pattern of intensity and a calibration image are required. Images, which have cellular regions with various intensity and migration patterns, could be created so that similar to that in experiments, they would be helpful to calculate velocities in each migration. The calibration method will be developed in a future study. In the meantime, the evaluation of each area for migration is important to understand cellular behaviors in detail, but this manually carried out process is time-consuming. To develop a method for automatically evaluating each area, successful detection of the centers of a void and static region is necessary. A program that is effective when the number of these centers is large will be developed in a future study. The intensity characteristics of a static cell, and single and collective cells with movement, are useful as the dataset for training in machine learning to categorize areas into static cell, single cell, and collective cell. If the areas in the image are categorized automatically, the ratio of cell type in the image can be easily calculated. Since the ratio is useful as the parameter to represent the environment of the cells, the combination of the proposed method and machine learning will be carried out in the future.

The results of H. roretzi through the application of the proposed method have indicated similarity and difference in the behaviors of single and collective cells quantitatively as well as qualitatively. These results would be helpful to understand the behavior of the hemocytes in vivo, which is hardly observed.

The component related to cell movement has been frequently observed. For example, in fluorescent microscopy, the actin filament (F-actin), one of the most important proteins in cell migration, is often observed. The changes in the distribution of F-actin during migration can be detected by the behaviors of the labeled cell. Furthermore, the shape of nuclei can be determined by phase-contrast microscopy. The combination of these factors and behaviors shown by the proposed cells makes it possible to estimate the mechanical factors related to each migration. A computational model based on these mechanical factors would be useful to predict the behaviors of cells, such as the behavior in vivo, precisely. Hence, the proposed method will be useful for investigations in complex physiological systems of various species.

Conclusion

In this study, a method to evaluate single and collective cell migration in bright field microscopy images was developed. The qualitative and quantitative evaluations for the migration of hemocytes in H. roretzi were successfully performed, using the method. While the pattern of intensity distribution in the cellular region directly influenced the evaluations, it could be improved by a synthetic image for calibration and in combination with machine learning. The proposed method is effective for predicting cell migration in vivo so that it will be useful to investigate complex physiological systems in various species.

Conflict of interest

The author declares no conflict of interest.

References

  1. 1.
    Gogoberidze N, Cimini BA. Defining the boundaries: challenges and advances in identifying cells in microscopy images. Curr Opin Biotechnol. 2024;85: 103055. doi: 10.1016/j.copbio.2023.103055.
  2. 2.
    Vicar T, Balvan J, Jaros J, Jug F, Kolar R, Masarik M, Cell segmentation methods for label-free contrast microscopy: review and comprehensive comparison. BMC Bioinformatics. 2019;20(1):360, doi: 10.1186/s12859-019-2880-8.
  3. 3.
    Zaritsky A, Manor N, Wolf L, Ben-Jacob E, Tsarfaty I. Benchmark for multi-cellular segmentation of bright field microscopy images. BMC Bioinformatics. 2013;14: 319. doi: 10.1186/1471-2105-14-319.
  4. 4.
    Ma J, Xie R, Ayyadhury S, Ge C, Gupta A, Gupta R, The multimodality cell segmentation challenge: toward universal solutions. Nat Methods. 2024;21: 1103–1113. doi: 10.1038/s41592-024-02233-6.
  5. 5.
    Stringer C, Wang T, Michaelos M, Pachitariu M. Cellpose: a generalist algorithm for cellular segmentation. Nat Methods. 2021;18: 100–106. doi: 10.1038/s41592-020-01018-x.
  6. 6.
    Pachitariu M, Stringer C. Cellpose 2.0: how to train your own model. Nat Methods. 2022;19(12):1634–1641. doi: 10.1038/s41592-022-01663-4.
  7. 7.
    Wang Z, Millet L, Chan V, Ding H, Gillette MU, Bashir R, Label-free intracellular transport measured by spatial light interference microscopy. J Biomed Opt. 2011;16(2):026019. doi: 10.1117/1.3549204.
  8. 8.
    Gebäck T, Schulz MMP, Koumoutsakos P, Detmar M. TScratch: a novel and simple software tool for automated analysis of monolayer wound healing assays. Biotechniques. 2009;46(4):265–274. doi: 10.2144/000113083.
  9. 9.
    Topman G, Sharabani-Yosef O, Gefen A. A standardized objective method for continuously measuring the kinematics of cultures covering a mechanically damaged site. Med Eng Phys. 2012;34(2):225–232. doi: 10.1016/j.medengphy.2011.07.014.
  10. 10.
    Zaritsky A, Natan S, Horev J, Hecht I, Wolf L, Ben-Jacob E, Cell motility dynamics: a novel segmentation algorithm to quantify multi-cellular bright field microscopy images. PLoS One. 2011;6(11):e27593. doi: 10.1371/journal.pone.0027593.
  11. 11.
    Home B, Bjørnerud , Pedersen NM, de la Ballina LR, Wesche J, Haugsten EM. Automated tracking of cell migration in phase contrast images with CellTraxx. Sci Rep. 2023;13: 22982. doi: 10.1038/s41598-023-50227-9.
  12. 12.
    Trepat X, Chen Z, Jacobson K. Cell migration. Compr Physiol. 2012;2(4):2369–2392. doi: 10.1002/cphy.c110012.
  13. 13.
    SenGupta S, Parent CA, Bear JE. The principles of directed cell migration. Nat Rev Mol Cell Biol. 2021;22(8):529–547. doi: 10.1038/s41580-021-00366-6.
  14. 14.
    Espina JA, Marchant CL, Barriga EH. Durotaxis: the mechanical control of directed cell migration. FEBS J. 2022;289(10):2736–2754. doi: 10.1111/febs.15862.
  15. 15.
    Maxian O, Mogilner A, Strychaiski W. Computational estimates of mechanical constrains on cell migration through the extracellular matrix. PLoS Comput Biol. 2020;16(8):e1008160. doi: 10.1371/journal.pcbi.1008160.
  16. 16.
    Betorz J, Bokil GR, Deshpande SM, Kulkarni S, Araya DR, Venturini C, A computational model for early cell spreading, migration, and competing taxis. J Mech Phys Solids. 2023;179: 105390. doi: 10.1016/j.jmps.2023.105390.
  17. 17.
    Friedl P, Gilmour D. Collective cell migration in morphogenesis, regeneration and cancer. Nat Rev Mol Cell Biol. 2009;10(7):445–457. doi: 10.1038/nrm2720.
  18. 18.
    Haeger A, Wolf K, Zegers MM, Friedl P. Collective cell migration: guidance principles and hierarchies. Trends Cell Biol. 2015;25(9):556–566. doi: 10.1016/j.tcb.2015.06.003.
  19. 19.
    Mayor R, Etienne-Manneville S. The front and rear of collective cell migration. Nat Rev Mol Cell Biol. 2016;17(2):97–109. doi: 10.1038/nrm.2015.14.
  20. 20.
    Hakim V, Silberzan P. Collective cell migration: a physics perspective. Rep Prog Phys. 2017;80(7):076601, doi: 10.1088/1361-6633/aa65ef.
  21. 21.
    Ladoux B, Mège RM. Mechanobiology of collective cell behaviours. Nat Rev Mol Cell Biol. 2017;18(12):743–757. doi: 10.1038/nrm.2017.98.
  22. 22.
    George M, Bullo F, Campàs O. Connecting individual to collective cell migration. Sci Rep. 2017;7(1):9720. doi: 10.1038/s41598-017-10069-8.
  23. 23.
    Jain S, Ladoux B, Mège RM. Mechanical plasticity in collective cell migration. Curr Opin Cell Biol. 2021;72: 54–62. doi: 10.1016/j.ceb.2021.04.006.
  24. 24.
    Clark AG, Maitra A, Jacques C, Bergert M, Pérez-González C, Simon A, Self-generated gradients steer collective migration on viscoelastic collagen networks. Nat Mater. 2022;21(10):1200–1210. doi: 10.1038/s41563-022-01259-5.
  25. 25.
    Pajic-Lijakovic I, Milivojevic M. Physics of collective cell migration. Eur Biophys J. 2023;52(8):625–640. doi: 10.1007/s00249-023-01681-w.
  26. 26.
    Hartmann J, Mayor R. Self-organized collective cell behaviors as design principles for synthetic developmental biology. Semin Cell Dev Biol. 2023;141: 63–73. doi: 10.1016/j.semcdb.2022.04.009.
  27. 27.
    Goodbody I. The physiology of ascidians. In: Russel FS, Yonge M, editors. Advances in marine biology. vol. 12, Amsterdam: Elsevier; 1974. p. 1–149. doi: 10.1016/S0065-2881(08)60457-5.
  28. 28.
    Berill NJ. The tunicata with an account of British species. Pisces Conservation Ltd; 1950.
  29. 29.
    Das SM. On the structure and function of the ascidian test. J Morphol. 1936;59(3):589–600. doi: 10.1002/jmor.1050590308.
  30. 30.
    Wright RK. Urochordates. In: Ratcliffe NA, Rowley AF, editors. Invertebrate blood cells. New York: Academic Press; 1981. p. 565–626.
  31. 31.
    Yokosawa H, Sawada H, Abe Y, Numakunai T, Ishii S. Galactose-specific lectin in the hemolymph of solitary ascidian, Halocynthia roretzi: Isolation and characterization. Biochem Biophys Res Commun. 1982;107(2):451–457. doi: 10.1016/0006-291X(82)91512-1.
  32. 32.
    Azumi K, Yokosawa H, Ishii S. Halocyamines: Novel antimicrobial tetrapeptide-like substances isolated from the hemocytes of the solitary ascidian Halocynthia roretzi. Biochemistry. 1990;29(1):159–165. doi: 10.1021/bi00453a021.
  33. 33.
    Yokosawa H, Harada K, Igarashi K, Abe Y, Takahashi K, Ishii S. Galactose-specific lectin in the hemolymph of solitary ascidian, Halocynthia roretzi. Molecular, binding and functional properties. Biochim Biophys Acta. 1986;870(2):242–247. doi: 10.1016/0167-4838(86)90228-1.
  34. 34.
    Harada-Azumi K, Yokosawa H, Ishii S. N-acetyl-galactosamine-specific lectin, a novel lectin in the hemolymph of the ascidian Halocynthia roretzi: Isolation, characterization and comparison with galactose-specific lectin. Comp Biochem Physiol B Biochem Mol Biol. 1987;88B(1):375–381. doi: 10.1016/0305-0491(87)90130-1.
  35. 35.
    Azumi K, Yoshimizu S, Suzuki S, Ezura Y, Yokosawa H. Inhibitory effect of halocyamine, an antimicrobial substance from ascidian hemocytes, on the growth of fish viruses and marine bacteria. Experientia. 1990;46(10):1066–1068. doi: 10.1007/BF01940675.
  36. 36.
    Azumi K, Yokosawa H, Ishii S. Lipopolysaccharide induces release of a metallo-protease from hemocytes of the ascidian, Halocynthia roretzi. Dev Comp Immunol. 1991;15(1-2):1–7. doi: 10.1016/0145-305X(91)90041-V.
  37. 37.
    Azumi K, Ozeki S, Yokosawa H, Ishii S. A novel lipopolysaccharide-binding hemagglutinin isolated from hemocytes of the solitary ascidian, Halocynthia roretzi: It can agglutinate bacteria. Dev Comp Immunol. 1991;15(1-2):9–16. doi: 10.1002/jez.1402650312.
  38. 38.
    Azumi K, Satoh N, Yokosawa H. Functional and structural characterization of hemocytes of the solitary ascidian, Halocynthia roretzi. J Exp Zool. 1993;265: 309–316. doi: 10.1002/jez.1402650312.
  39. 39.
    Fuke MT. Studies on the coelomic cells of some Japanese ascidians. Bull Mar Biol St Asamushi Tohoku Univ. 1979;16(3):143–159.
  40. 40.
    Donaghy L, Hong HK, Park KI, Nobuhisa K, Youn SH, Kang CK, Flow cytometric characterization of hemocytes of the solitary ascidian, Halocynthia roretzi. Fish Shellfish Immun. 2017;66: 289–299. doi: 10.1016/j.fsi.2017.05.009.
  41. 41.
    Dan-Sohkawa M, Morimoto M, Mishima H, H Kaneko. Characterization of coelomocytes of the ascidian Halocynthia roretzi based on phase-contrast, time-lapse, video and scanning electron microscopic observations. Zoolog Sci. 1995;12: 289–301. doi: 10.2108/zsj.12.289.
  42. 42.
    Dan-Sohkawa M, Morimoto M, Kaneko H. In vitro reactions of coelomocytes against sheep red blood cells in the solitary ascidian Halocynthia roretzi. Zoolog Sci. 1995;12(4):411–417. doi: 10.2108/zsj.12.411.
  43. 43.
    Nishiyama Y, Langan P, Chanzy H. Crystal structure and hydrogen-bonding system in cellulose I𝛽 from synchrotron X-ray and neutron fiber diffraction. J Am Chem Soc. 2002;124(31):9074–9082. doi: 10.1021/ja0257319.
  44. 44.
    Zhao Y, Li J. Excellent chemical and material cellulose from tunicates: Diversity in cellulose production yield and chemical and morphological structures from different tunicate species. Cellulose. 2014;21(5):3427–3441. doi: 10.1007/s10570-014-0348-6.
  45. 45.
    Štrucova A, Davies GR, Eichhorn SJ. Elastic modulus and stress-transfer properties of tunicate cellulose whiskers. Biomacromolecules. 2005;6(2):1055–1061. doi: 10.1021/bm049291k.
  46. 46.
    Anno K, Otsuka K, Seno N. A chitin sulfate-like polysaccharide from the test of the tunicate Halocynthia roretzi. Biochim Biophys Acta. 1974;362(1):215–219. doi: 10.1016/0304-4165(74)90043-9.
  47. 47.
    Wagner GP. Evolution and multi-functionality of the chitin system. EXS. 1994;69: 559–577. doi: 10.1007/978-3-0348-7527-1_33.
  48. 48.
    Okiei W, Nishimura S, Somorin O, Nishi N, Tokura S. Inhibitory action of sulphated chitin derivatives on the hydrolytic activity of thrombin. In: Muzzarelli R, Jeuniaux C, Gooday GW, editors. Chitin in nature and technology. New York: Plenum Press; 1986. p. 453–460. doi: 10.1007/978-1-4613-2167-5_53.
  49. 49.
    Jayakumar R, Nwe N, Tokura S, Tamura H. Sulfated chitin and chitosan as novel biomaterials. Int J Biol Macromol. 2007;40(3):175–181. doi: 10.1016/j.ijbiomac.2006.06.021.
  50. 50.
    Farinha I, Freitas F. Chemically modified chitin, chitosan, and chitinous polymers as biomaterials. In: Gopi S, Thomas S, Pius A, editors. Handbook of chitin and chitosan (Chitin- and chitosan-based polymer materials for various applications). vol. 3, Amsterdam: Elsevier; 2020. p. 43–69. doi: 10.1016/B978-0-12-817966-6.00002-9.
  51. 51.
    Toda N, Horikawa T, Anno K, Seno N. Galactan sulfate from the test of tunicates. Carbohyd Res. 1978;62(2):389–392. doi: 10.1016/S0008-6215(00)80889-7.
  52. 52.
    Stelling MP, de Bento AA, Caloba P, Vilanova E, Pavão MSG. Methods for isolation and characterization of sulphated glycosaminoglycans from marine invertebrates. In: Vigetti D, Theocharis AD, editors. The extracellular matrix - Methods and protocols - (Methods in molecular biology). Totowa, NJ: Humana Press; 2019 doi: 10.1007/978-1-4939-9133-4.
  53. 53.
    Galanakis CM, editor. London, UK: Carotenoids: properties, processing and applications. Academic Press; 2020 doi: 10.1016/C2018-0-00646-5.
  54. 54.
    Matsuno T, Ookubo M. A new carotenoid, halocynthiaxanthin from the seq squirt, Halocynthia roretzi. Tetrahedron Lett. 1981;22(46):4659–4660. doi: 10.1016/S0040-4039(01)83006-9.
  55. 55.
    Matsuno T, Ookubo M, Nishizawa T, Shimizu I. Carotenoids of sea squirts. i. new marine carotenoids, halocynthiaxanthin and mytiloxanthinone from Halocynthia roretzi. Chem Pharm Bull. 1984;32(11):4309–4315. doi: 10.1248/cpb.32.4309.
  56. 56.
    Matsuno T. Aquatic animal carotenoids. Fisheries Sci. 2001;67: 771–783. doi: 10.1046/j.1444-2906.2001.00323.x.
  57. 57.
    Konishi I, Hosokawa M, Sashima T, Kobayashi H, Miyashita K. Halocynthiaxanthin and fucoxanthinol isolated from Halocynthia roretzi induce apoptosis in human leukemia, breast and colon cancer cells. Comp Biochem Phys C. 2006;142: 53–59. doi: 10.1016/j.cbpc.2005.10.005.
  58. 58.
    Maoka T. Recent progress in structural studies of carotenoids in animals and plants. Arch Biochem Biophys. 2009;483(2):191–195. doi: 10.1016/j.abb.2008.10.019.
  59. 59.
    Ito M, Yamano Y, Tode C, Wada A. Carotenoid synthesis: Retrospect and recent progress. Arch Biochem Biophys. 2009;483: 243–228. doi: 10.1016/j.abb.2008.11.021.
  60. 60.
    Maoka T. Carotenoids in marine animals. Mar Drugs. 2011;9: 278–293. doi: 10.3390/md9020278.
  61. 61.
    Rohmaha Z, Jeong U, Ticar BF, Kim J, Lee J, Kang SJ, Effect of ascidian (Halocynthia roretzi, Drasche 1884) tunics carotenoids on enhancing growth and muscle coloring of sea-reared rainbow trout (Oncorhynchus mykiss, Walbaum 1792). Aquacult Rep. 2016;4: 22–29. doi: 10.1016/j.aqrep.2016.05.003.
  62. 62.
    Maoka T. Carotenoids as natural functional pigments. J Nat Med. 2020;74: 1–16. doi: 10.1007/s11418-019-01364-x.
  63. 63.
    Ikeda C, Manabe Y, Tomonaga N, Wada T, Maoka T, Sugawara T. Evaluation of intestinal absorption of dietary halocynthiaxanthin, a carotenoid from the sea squirt Halocynthia roretzi. Mar Drugs. 2020;18: 588. doi: 10.3390/md18120588.
  64. 64.
    Maoka T, Tode C. A novel carotenoid with a unique 2,6-cyclo-𝜓-end group, roretziaxanthin, from the sea squirt Halocynthia roretzi. Mar Drugs. 2022;20(12):732. doi: 10.3390/md20120732.
  65. 65.
    Tsuchiya Y, Suzuki Y. Biochemical studies of the ascidian, Cynthia roretzi V. Drasche-V (The presence of pseudokeratin in test). Nippon Suisan Gakkaishi. 1962;28(2):222–226. doi: 10.2331/suisan.28.222.
  66. 66.
    Kato Y. Active movement of the tunic in Halocynthia roretzi. J Biomech Sci Eng. 2010;5(2):163–174. doi: 10.1299/jbse.5.163.
  67. 67.
    Kato Y. Mechanical senses and the tunic structure in Halocynthia roretzi. In: Proceedings of 2011 International Offshore and Polar Engineering Conference, 19–23 June 2011; Maui, California. Cupertino, CA: ISOPE; 2011. p. 250–253.
  68. 68.
    Kato Y. The role of protein as a deformation controller in cellulose tissue. In: Proceedings of ASME 2012 International Mechanical Engineering Congress and Exposition. New York: ASME; 2012. p. 607–613. doi: 10.1115/IMECE2012-89313.
  69. 69.
    Kato Y. Deformation control and mass transfer in the tunic of Halocynthia roretzi. Open Chem J. 2018;5: 1–17. doi: 10.2174/1874842201805010001.
  70. 70.
    Kato Y. Active deformation in the tunic of Halocynthia roretzi: How the tissue composed of cellulose responds to stimuli and deforms. In: Hossain A, editor. Plant stress physiology. London, UK: IntechOpen; 2020 doi: 10.5772/intechopen.93192.
  71. 71.
    Kato Y. Maintenance of the cell motility in the tunic of Halocynthia roretzi. In: Kongoli F, Marquis F, Chen P, Prikhna T, Chikhradze N, editors. Sustainable Industrial Processing Summit SIPS2018 Volume 6. New and Advanced Materials and Technologies. Mont-Royal, Canada: FLOGEN Star Outreach; 2018. p. 207–214.
  72. 72.
    Kato Y. Active contraction in the stable mechanical environment of the tunic of the ascidian, Halocynthia roretzi, a polysaccharide-based tissue with blood circulatory system. Polymers. 2023;15(1):4329. doi: 10.3390/polym15214329.
  73. 73.
    Ridler TW, Calvard S. Picture thresholding using an iterative selection method. IEEE T Syst Man Cyb. 1978;8(8):630–632. doi: 10.1109/TSMC.1978.4310039.

Written by

Yoko Kato

Article Type: Research Paper

•

Date of acceptance: August 2024

Date of publication: August 2024

•

DOI: 10.5772/dmht.20240001

Copyright: The Author(s), Licensee IntechOpen, License: CC BY 4.0

Download for free

© The Author(s) 2024. Licensee IntechOpen. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.

Impact of this article

258

Downloads

275

Views

1

Altmetric Score

Share this article

Join us today!

Submit your Article