Open Access
Add to BibSonomy
Abstract
In cardiovascular disorders, the study of thrombocytes, commonly known as platelets, is highly important since they are involved in blood clotting, essential in hemostasis, and they can in pathological situations affect the blood circulation. In this paper, single deposited platelets are measured using interferometric digital holographic microscopy. We have shown that the average optical height of platelets is significantly lower in healthy volunteers than in dialyzed patients, meaning a better spreading. It demonstrates the great interest for assessing this parameter in any patients, and therefore the high potential of analyzing single spread platelets using digital holographic microscopy in fundamental research as well as a diagnostic tool in routine laboratories, for usual blood tests.
© 2021 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Platelets are anucleate small blood cells, which are essential for primary hemostasis: they adhere to the subendothelial collagen exposed by vessel injury, become activated, aggregate, and form a thrombus to stop bleeding. Adhesion and aggregation are often considered as distinct processes through which platelets establish individual contacts with the extracellular matrix, or form clumps by sticking to each other, respectively.
It is well known that some specific adhesion receptors play an essential role in the adhesion process, such as α2β1 integrin and platelet glycoprotein (GP) VI that mediate interactions with collagen IV [1,2]. Numerous works studied the response of platelets, in terms of morphological changes, to a variety of agonists, and the general process of platelet spreading on different surfaces is also well documented [3–5].
However, how the platelets differentially respond to specific adhesive ligands remains unclear in patients suffering from different pathologies. In particular, patients with end-stage renal disease (ESRD) exhibit a dysfunction of hemostatic processes [6]. Paradoxically, bleeding diathesis and thrombotic tendencies are observed. In this context, changes in platelet behavior have been investigated [7,8]. Nonetheless, to the best of our knowledge, the spreading of platelets has never been studied in this clinical situation.
The commonly used technique to study the platelet spreading is the classical optical transmission microscopy, which provides two-dimensional (2D) information: the surface area covered by each observed platelet. However, a finer analysis of spread platelets requires three-dimensional (3D) information, in order to get the height profiles. Confocal microscopy could be used, as it gives 3D information about samples, but it would need a long time of analysis per platelet, making this technique unsuitable to undertake a statistical study with a large number of platelets.
Digital holographic microscopy (DHM) is well appropriate for such an analysis. Indeed, DHM, e.g., in off-axis configuration, enables the quantitative phase contrast with one single acquired image, called a hologram [9,10]. It provides the optical thickness measurement — hence 3D information — of the observed sample, up to nanometric accuracy [11]. DHM was used to image a variety of biological samples [11,12]. Illuminations with a high degree of coherence, such as lasers, are usually used in DHM, inducing some coherent noise, typically speckle, which alters the images. Nevertheless, decreasing the spatial coherence of the illumination in DHM allows to drastically reduce the coherent noise, resulting in high quality holograms [13–15].
In a previous paper, we used DHM to characterize platelet aggregates, generated from whole blood of healthy volunteers by the Impact-R platelet analyzer, through the quantitative phase contrast imaging [16]. The present paper is really distinct and addresses novel issues: single deposited platelets are analyzed and not platelet aggregates, which allows the study of another process (adhesion vs. aggregation, as explained above). The platelet deposition occurs in a static well, which improves the reproducibility. Moreover, we demonstrate it in a clinical application by comparing dialyzed patients and healthy volunteers.
For the analysis of platelet morphology, it was shown that results obtained with DHM were consistent with those from scanning electron microscopy, atomic force microscopy, and flow cytometry, hence demonstrating the high potential of DHM for this purpose [17]. The combination of DHM with tomography is also possible, by merging numerous acquired holograms with varying angles of observation, giving rise to a 3D map of the refractive index; this was considered for the measurement of platelet morphology, e.g., for the comparison of several agonists in the activation process [18], and in mouse models with genetic abnormalities [19]. Whatever the optical setup, including in high-throughput imaging, the analysis of platelet morphology may benefit from deep learning algorithms [20].
In the present study, specific experimental protocol and image processing, described in points 2 and 3.1, are developed in order to perform the comparative statistical 3D analysis, given in point 3.2, of individual spread platelets from healthy donors and dialyzed patients, on different coatings. The work also aims to highlight the benefits of getting 3D features with DHM instead of the 2D measurements in classical microscopy, when analyzing platelet spreading. The potential of the developed method, e.g., for routine blood test, is therefore discussed in the conclusions, in point 4.
2. Materials & methods
2.1 Sample preparation
The reproducibility is essential when platelet spreading and aggregation are studied. This is the reason why, in the method developed in this paper, the platelet spreading occurs during a sedimentation step: the well is static during the spreading process. This allows a high reproducibility.
Blood samples were collected from healthy volunteers (number of individuals n = 7) and patients in dialysis (n = 7). The procedure was approved by the hospital ethics committee of the Centre hospitalier universitaire (CHU) de Charleroi, Belgium (Comité d’Ethique I.S.P.P.C: OM008). The studies conform to the principles outlined in the Declaration of Helsinki. Venous blood was drawn into tubes with 3.2% sodium citrate solution, at pH 7.4. Platelet-rich plasmas (PRP) were obtained by the centrifugation of whole blood at 150 g during 10 min. These parameters are usual for a gentle centrifugation. For each donor, the platelet (plt) concentration of the PRP was adjusted at 105 plt/µL by diluting it with the platelet-poor plasma (PPP) from this donor.
Cell culture glass slides with ten wells were used (CELLview, Greiner, Ref. 543078, diameter in the bottom of 6.4 mm, cf. Figure 1). For every slide, two wells were coated with collagen IV (3 µg/well), two with laminin (0.5 µg/well), and two without any coating, used as control. These proteins were solubilized in Hank's balanced salt solution (HBSS). In order to coat a well, 75 µL of the suitable prepared solution were injected inside and remained during 90 min at 37°C. Then, wells were washed with HBSS, three times. This coating procedure corresponds to the manufacturer’s instructions.
Fig. 1. Experimental cells used for the study, with ten wells. (a) Initial configuration and (b) once the well walls are removed, for the analysis with the microscope.
For each donor, 100 µL of PRP were poured into each of the six wells described above. Platelet spreading was obtained after 30 min at 37°C in humid atmosphere. Working at 37°C is usual and relevant because the biochemical reactions occurring inside human cells are optimal at this temperature. Moreover, room temperature can fluctuate from one day to another day, which would induce some variability. Then, wells were washed three times with HBSS and platelets were fixed with 200 µL of paraformaldehyde (PAF) 2%. Fixation stops the spreading process and the biochemical reactions inside platelets. This allows to have got exactly the same experimental duration for all platelets. Working without fixation would induce some variability. Wells were finally washed three times with demineralized water and dried. The well walls were then removed [cf. Figure 1(b)] before microscopy analysis.
2.2 Optical setup and hologram acquisition
Digital holograms were acquired using the red, green, and blue (RGB) DHM depicted in Fig. 2, working with partially spatially coherent illumination. This microscope is fully described in [15,21]. The light source (SO) is composed by three laser diodes: a Schäfter & Kirchhoff red laser ($\lambda $ = 639 nm, where $\lambda $ is the wavelength, 35 mW), a Cobolt Samba green laser ($\lambda $ = 532 nm, 25 mW), and a Newport blue laser ($\lambda $ = 488 nm, 20 mW). Passing through the lens L1 (f = 20 mm, where f is the focal length of the lens), the light beam is focused near the rough surface of a rotating ground glass (RGG). The lens L2 (f = 20 mm) then collimates the beam, which is afterwards injected into a multimode light guide (LG, diameter = 365 µm, NA = 0.22, where NA is the numerical aperture) by the lens L3 (f = 20 mm). The lens L4 (f = 80 mm) collimates the beam at the output of the light guide. A graduated iris (GI), placed in the back focal plane of the lens L5 (f = 200 mm), and opened at a diameter of 8 mm, then filters the beam. The illumination after the lens L5 is hence spatially partially coherent, and the degree of spatial coherence is governed by the diameter of the iris GI [15,21]. Using such an illumination in DHM allows a drastic reduction of the usual coherent noise, therefore giving rise to high quality images, both for intensity and phase [15].
Fig. 2. Digital holographic microscope used for the analysis of platelets, working with a source of partial spatial coherence: BS1–BS2, beam splitters; C, optical path compensator; CCD, camera; G, diffraction grating; GI, graduated iris; L1–L10, lenses; LG, light guide; M1–M4, mirrors; ML1–ML2, microscope lenses; ND, neutral density filter; OS, optical stop; RA, rotation assembly; RGG, rotating ground glass; S, sample; SO, light source.
The light beam then enters into a Mach–Zehnder interferometer, made up of two beam splitters (BS1–BS2) and two mirrors (M2–M3). One arm of the interferometer contains the sample (S), whereas the other arm works as a reference. The reference beam is slanted with respect to the optical axis, thanks to a Ronchi grating (G) followed by an optical stop (OS) that only keeps a first order of diffraction. Such an off-axis configuration enables the extraction of the complex amplitude, i.e., both optical phase and intensity, from one single acquired image, using the Fourier method [9,21], described in point 2.3. With exposure times of typically 10 ms, the sample illumination can be seen as of low spatial coherence, but high temporal coherence. However, it is interesting to note that the optical setup used in this study is also efficient for DHM working with both spatially and temporally, low coherent illumination [21].
The sample slide is set in the microscope with the adhering platelets facing the microscope lens, i.e., in such a way that the illumination firstly passes through the slide, secondly through the platelets, and thirdly propagates by air, before reaching the microscope lens. Microscope lenses are Leica HCX PL Fluotar L 40${\times} $ with NA = 0.6. For each analyzed well, 40 distinct RGB holograms were recorded in the central area, using a Hamamatsu ORCA-3CCD camera (C7780) with a 1344 ${\times} $ 1024 pixels array, pixel size 6.45 µm ${\times} $ 6.45 µm. The motorized sample holder allows to get the same positions for every well. The sample is static during the acquisition.
2.3 Hologram processing
The successive steps for image processing are illustrated in Fig. 3. For each of the three RGB channels, the images of the complex amplitude of the optical field, 1024 ${\times} $ 1024 pixels corresponding to 183.5 µm ${\times} $ 183.5 µm, are extracted from the holograms using the Fourier method [9,21]: the discrete Fourier transform [Fig. 3(b)] of the hologram [Fig. 3(a)] is computed, whence the right side lobe is isolated and centered. Some punctual defects in the Fourier domain, e.g., due to multiple reflections, are removed using a multiplicative filter [Fig. 3(c)], made of truncated Gaussian surfaces, centered in the defects. The used filter is the same for all holograms in the present study. The filtered lobe [Fig. 3(d)] corresponds, in the Fourier domain, to the complex amplitude of the optical field of the object beam [9,21]; its inverse Fourier transform thus provides the complex amplitude in the recorded plane from which are extracted the intensity [modulus, i.e., square root of the intensity, in Fig. 3(e)] and the phase [Fig. 3(f)].
Fig. 3. Automatic extraction, correction, and refocusing of the complex amplitude images from digital holograms of platelets. Only the green channel is shown in this figure. (a) Recorded hologram; (b) discrete Fourier transform of (a); (c) multiplicative filter used in the Fourier domain for the correction of the right side lobe; (d) resulting filtered right side lobe; (e) modulus and (f) wrapped phase of the complex amplitude, directly extracted from the filtered side lobe; (g) modulus and (h) phase images of the computed blank, containing permanent defects and aberrations; (i) modulus and (j) phase images corresponding to (e) and (f), respectively, corrected using the blank; and (k) automatically refocused modulus and (l) phase images of the corrected complex amplitude images. The scale bar is worth the same in (b), (c), and (d).
Moreover, the correction of defects and aberrations in the extracted images is performed using a computed reference blank amplitude [Figs. 3(g) and (h)], following the method described in [22,23]. It is obtained by averaging all complex amplitude images, for every well. Prior to the averaging process, the phase of every image is shifted in order to have the same value of the phase background for all images. This is required to compensate the eventual phase fluctuations or the thickness changes of the glass slides. The computed blank amplitude then divides every image of interest, resulting in high quality images, regarding both the intensity [modulus in Fig. 3(i)] and the phase [Fig. 3(j)].
Fig. 4. Automatic segmentation of the images and detection of single platelets. (a) Phase image corresponding to Fig. 3(l), with a third-degree polynomial correction for flattening the background; (b) normalized correlation product, as defined in Eq. (1), of the complex amplitude of the image corresponding to (a) with the complex amplitude of the reference object shown in (c); (c), (d), and (e) phase images of the reference objects used for the detection process, with a field of view of 11.5 µm ${\times} $ 11.5 µm; (f) profile along the green bar in (b), showing the efficiency of the normalized correlation product for the platelet detection process; and (g) representation, in red, of the platelets detected and segmented by the method. Blue and pink arrows in (a) respectively point to the aggregated platelets and to those at the border of the image, respectively, both automatically rejected by the algorithm, as seen in (g).
Furthermore, the images are finely refocused [Figs. 3(k) and (l)], using the color refocusing criterion based on the amplitude analysis [24,25], with the fast algorithm that we developed in [25]. As a function of the reconstruction distance, the criterion is maximal in the best focal plane of the platelets. It is indeed crucial, in order to compare the observed platelets, to ensure that the sample is accurately focused in the analyzed plane. As we previously showed, the use of this RGB criterion, rather than a similar criterion in monochromatic conditions, allows an improved detection of the refocusing distance, with a better accuracy [25]. For all analyzed images, the refocusing distances in vacuum ranged from –5 to 15 µm, approximately.
A final correction is performed to further flatten the background of the phase images [Fig. 4(a)]. Such a fine correction is required due to some flatness defects of the glass slides. For every phase image, the polynomial surface (3rd degree) best approaching the background, in the sense of least squared optimization, is determined. This phase surface is subtracted, point by point, to the corrected and refocused phase image. In the final phase images, the background is set to 1, whereas the values range from 0 to 2$\mathit{\pi }$.
3. Results and discussion
3.1 Detection of platelets and measurement of the parameters
The accurate detection and segmentation of platelets in the phase images are required for a proper analysis per platelet. This process is illustrated in Fig. 4. It must be automated, as a huge number of platelets are measured. The average number of analyzed platelets per well is indeed circa 1500.
Since the setup was finely aligned with the green wavelength, this color provides the highest accuracy among the three channels. The following steps hence only consider the green channel. As explained in point 2.3, the use of RGB holograms in the first steps of the hologram processing allows a more accurate refocusing, which improves the accuracy of the subsequent phase measurements.
The segmentation itself is implemented in the phase images. Every image is partitioned into several areas using a threshold applied on the phase maps, fixed at 1.06 for all the experiments: when the phase is locally larger than 1.06, we consider that the corresponding pixel is within a platelet area, otherwise it is considered as belonging to the background. This process is used to create a partition mask. The method is actually robust and relatively not sensitive to the value of this threshold. The value of 1.06 was chosen to be as closest as possible to the background value, at 1.00, while being superior to the maximal fluctuations of the background, for all images. Since the images are of high quality and efficiently corrected, the phase background is very flat, in such a way that it does never fluctuate over 1.06 for all experiments. Moreover, the areas exhibiting a hole, i.e., surrounding a smaller area which is lower than the threshold, are processed to fill the holes in the partition mask. This allows to properly analyze the platelets forming a crown shape by spreading, as in Fig. 4(d). Furthermore, the areas in contact with the border of the images, such as indicated by pink arrows in Fig. 4(a), are deleted since they may correspond to incomplete platelets.
The detection of individual platelets is required in order to reject the areas with multiple platelets. Indeed, platelets may form aggregates, such as pointed by blue arrows in Fig. 4(a), which would induce a bias: the platelet parameters must be measured per platelet; one single platelet per segmented area is thus allowed.
The detection is operated by comparing every complex amplitude image to be analyzed with a set of typical shapes of platelets, shown in Fig. 4(c–e). The detection process is also illustrated in Fig. 4. The comparison between a complex amplitude image $f({\textbf r} )$, where ${\textbf r} = ({x,y} )$ is the position vector, and a reference complex amplitude image $g({\textbf r} )$ is performed by computing the normalized correlation product ${\mathrm{\tilde{\circledast }}_\mathrm{\Gamma }}$, which gives a dimensionless number, that we define with
For each detected platelet, several parameters are measured in the phase images: (1) the surface area S, (2) the optical volume V, and (3) the average optical height $\hat{h} = V/S$. The optical volume V is the integrated optical height, over the whole surface of the platelet. The optical height $h({\textbf r} )$ is related to the measured phase $\mathit{\varphi }({\textbf r} )$ through
3.2 Statistical analysis of platelet parameters
Figure 5 shows the effect of two coatings, collagen and laminin, on the morphology of adherent platelets, compared to an untreated control well. Healthy volunteers and dialyzed patients are separately analyzed. As explained in 2.1, samples are studied in duplicate, which means that two wells are processed for each coating, for every individual. The platelet morphology is studied using the three parameters S, V, and $\hat{h}$, defined in 3.1. For each parameter, the median values are assessed for every well, over all the detected platelets. The diagrams in Fig. 5 represent the statistical distribution of these median values. The p-value (p), computed in the sense of the Wilcoxon signed-rank test, allows to compare two coatings; a value $ p < 0.05$ is considered as statistically significant. This statistical test is well suited here since every point is for one single well.
Fig. 5. Comparison between platelet spreading in (a), (b), and (c) healthy volunteers ($n = 7$, in duplicate), and (d), (e), and (f) dialyzed patients ($n = 7$, in duplicate), on different coatings: control (CTRL), collagen, and laminin. The diagrams show the statistical distribution (median, interquartile range, 5th and 95th percentiles, and minimum and maximum), among the samples, of the median, computed for all platelets of each sample, of (a) and (d) the surface area S, (b) and (e) the optical volume V, and (c) and (f) the average optical height $\hat{h}$. The p-value between two coatings is computed in the sense of the Wilcoxon signed-rank test.
In healthy volunteers, from control to collagen coating, we observe that platelets behave to improve their spreading process. Platelets are characterized by an increase in their surface area S [Fig. 5(a), p = 0.03] and a decrease in their optical volume V [Fig. 5(b), p = 0.03]. If the average optical height, $\hat{h}$, is considered, the phenomenon is even more visible, with a decrease in $\hat{h}$ from control to collagen [Fig. 5(c), p = 0,004). The laminin coating does not show any significant difference with respect to the control in healthy volunteers.
Conversely, from control to collagen coating, the platelets of dialyzed patients present reduced surface areas S with respect to the volunteer subjects [Fig. 5(d), p = 0.013], whereas the optical volume V and the average optical height $\hat{h}$ do not show any significant change.
Therefore, the effect of collagen coating with respect to the control situation is explored: Fig. 6 compares the reactivity of platelets in healthy volunteers and dialyzed subjects. For every individual (in duplicate), the ratio between the median values for the collagen and the control conditions is assessed. The statistical distribution of these ratios is represented in Fig. 6. The p-value corresponds to the Mann–Whitney–Wilcoxon rank-sum test; $\; p < 0.05$ being considered as meaning a statistically significant difference between two groups. This statistical test is suitable because the ratios are computed between the median values of two distinct wells.
Fig. 6. Comparison between platelet spreading in healthy volunteers ($n = 7$, in duplicate) and dialyzed patients ($n = 7$, in duplicate): statistical distributions (median, interquartile range, 5th and 95th percentiles, and minimum and maximum) of the collagen/control ratio for (a) the surface area S, (b) the optical volume V, and (c) the average optical height $\hat{h}$. For each sample, the collagen/control ratio is defined as the ratio of the median of the parameter measured per platelet in the collagen-coated well to the median of the measured parameter in the corresponding control well. The p-value between healthy volunteers and dialyzed patients is computed in the sense of the Mann–Whitney–Wilcoxon rank-sum test.
We observe that for healthy subjects we have an improved spreading from control to collagen. Indeed, for the surface area S the ratios are mainly greater than 1; whereas for dialyzed patients, ratios are mainly lower than 1, meaning a lesser spreading [Fig. 6(a), p < 0.001]. The difference between healthy and dialyzed individuals is also significant for the average optical height $\hat{h}$ [Fig. 6(c), p = 0.04], but not for the optical volume V [Fig. 6(b)]. However, this can be explained by the fact that the optical volume does not obviously reflect the spreading efficiency, contrarily to the surface area and the average height. Nevertheless, it seems interesting to study the optical volume. Indeed, when a platelet spreads, it loses some materials (granules and water) in its environment, which could affect the optical volume. Studying the optical volume independently of the average height could therefore be of interest in some cases and, anyway, it provides some pieces of information.
4. Conclusions and perspectives
It is well known that the platelets of dialyzed patients exhibit phenotypic changes [27]; however, to our knowledge, this paper reports for the first time that a detailed statistical 3D analysis of platelet spreading has been carried out in dialyzed patients and that a significant difference is demonstrated in the platelet spreading between healthy volunteers and dialyzed patients. It is of high importance because it shows the potential of our method as an analysis tool in diagnosis.
Indeed, in this study, we have characterized the 3D morphology of single spread platelets using DHM, considering the effect of the coating, in both healthy volunteers and dialyzed patients. The automated method allows the easy measurement of a great number of platelets per individual and per coating condition (circa 1500 per well in this study). In the developed method, the reproducibility is particularly effective since the platelet spreading occurs after a sedimentation step, while the well is static. The method is confirmed by the clinical application, since the results show a significant difference between both groups.
DHM has the advantage of providing the parameters assessed by classical optical microscopy, i.e., the surface of each platelet, in addition to 3D measurement. The measured values of the surface areas of spread platelets in healthy individuals are in the same range of what it is reported in the literature [28,29].
For dialyzed patients, the findings presented in this study give the opportunity to undertake future research along with the implications of these results, and to go further in the understanding of the molecular mechanisms involved in platelet dysfunction, which is essential for improving the treatment.
The present study has also important perspectives in diagnostic procedures. From one hand, DHM could also allow to perform the automated classification of spread platelets with respect to their shape, e.g., crown shape vs. usual shape, and to compare their distribution between pathological and healthy cases. Computing the normalized correlation product of the complex images with two — or two sets of — reference images and keeping the maximal value would give an efficient tool for this sake. Further studies will potentially show the interest of such analyses in diagnosis.
On the other hand, clinical tests of platelet function are commonly performed on platelet-rich plasma or on whole blood, but in suspension. In practice, these tests are often carried out in the presence of agonists or molecules activating platelets such as adenosine diphosphate (ADP), collagen, epinephrine, arachidonic acid, ristocetin or thrombin receptor activator peptide (TRAP). To the best of our knowledge, no clinical laboratory test takes the platelet spreading process into account. As we have shown in this work, digital holographic microscopy reveals to be a powerful tool to assess the platelet spreading in 3D. Larger clinical studies in different pathological situations will show how this original approach could highlight alterations in platelet function that are still little understood.
Funding
Fonds de la Recherche Scientifique – FNRS; Centre hospitalier universitaire de Charleroi; Fonds de la Chirurgie Cardiaque; Fonds de la Recherche Médicale en Hainaut; Horizon 2020 Framework Programme (823712); European Space Agency (PRODEX Experiment Arrangement 90233).
Acknowledgments
Dr Jérôme Dohet-Eraly was Chargé de Recherches du Fonds de la Recherche Scientifique – FNRS (F.R.S.-FNRS, Belgium).
Disclosures
The authors declare no conflict of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. Z. M. Ruggeri and G. L. Mendolicchio, “Adhesion mechanisms in platelet function,” Circ. Res. 100(12), 1673–1685 (2007). [CrossRef]
2. K. R. Machlus and J. E. Italiano Jr, “The incredible journey: from megakaryocyte development to platelet formation,” J. Cell Biol. 201(6), 785–796 (2013). [CrossRef]
3. J. H. Hartwig, “Mechanisms of actin rearrangements mediating platelet activation,” J. Cell Biol. 118(6), 1421–1442 (1992). [CrossRef]
4. M. Jiroušková, J. K. Jaiswal, and B. S. Coller, “Ligand density dramatically affects integrin αIIbβ3-mediated platelet signaling and spreading,” Blood 109(12), 5260–5269 (2007). [CrossRef]
5. M. B. Horev, Y. Zabary, R. Zarka, S. Sorrentino, O. Medalia, A. Zaritsky, and B. Geiger, “Differential dynamics of early stages of platelet adhesion and spreading on collagen IV- and fibrinogen-coated surfaces,” F1000Research 9, 449 (2020). [CrossRef]
6. J. Lutz, J. Menke, D. Sollinger, H. Schinzel, and K. Thürmel, “Haemostasis in chronic kidney disease,” Nephrol. Dial. Transplant. 29(1), 29–40 (2014). [CrossRef]
7. J. Lutz and K. Jurk, “Platelets in advanced chronic kidney disease: two sides of the coin,” Semin. Thromb. Hemostasis 46(03), 342–356 (2020). [CrossRef]
8. P. Boccardo, G. Remuzzi, and M. Galbusera, “Platelet dysfunction in renal failure,” Semin. Thromb. Hemostasis 30(5), 579–589 (2004). [CrossRef]
9. M. Takeda, H. Ina, and S. Kobayashi, “Fourier-transform method of fringe-pattern analysis for computer-based topography and interferometry,” J. Opt. Soc. Am. 72(1), 156–160 (1982). [CrossRef]
10. Y. Rivenson, Y. Wu, and A. Ozcan, “Deep learning in holography and coherent imaging,” Light: Sci. Appl. 8, 85 (2019). [CrossRef]
11. F. Dubois, C. Yourassowsky, O. Monnom, J.-C. Legros, O. Debeir, P. Van Ham, R. Kiss, and C. Decaestecker, “Digital holographic microscopy for the three-dimensional dynamic analysis of in vitro cancer cell migration,” J. Biomed. Opt. 11(5), 054032 (2006). [CrossRef]
12. B. Kemper, A. Bauwens, D. Bettenworth, M. Götte, B. Greve, L. Kastl, S. Ketelhut, P. Lenz, S. Mues, J. Schnekenburger, and A. Vollmer, “Label-free quantitative in vitro live cell imaging with digital holographic microscopy,” in Label-Free Monitoring of Cells in vitro, Vol. 2 of Bioanalytical Reviews, J. Wegener, ed. (Springer, 2019), pp. 219–272.
13. F. Dubois, L. Joannes, and J.-C. Legros, “Improved three-dimensional imaging with a digital holography microscope with a source of partial spatial coherence,” Appl. Opt. 38(34), 7085–7094 (1999). [CrossRef]
14. F. Dubois, M.-L. Novella Requena, C. Minetti, O. Monnom, and E. Istasse, “Partial spatial coherence effects in digital holographic microscopy with a laser source,” Appl. Opt. 43(5), 1131–1139 (2004). [CrossRef]
15. J. Dohet-Eraly, C. Yourassowsky, A. El Mallahi, and F. Dubois, “Quantitative assessment of noise reduction with partial spatial coherence illumination in digital holographic microscopy,” Opt. Lett. 41(1), 111–114 (2016). [CrossRef]
16. K. Zouaoui Boudjeltia, D. Ribeiro de Sousa, P. Uzureau, C. Yourassowsky, D. Perez-Morga, G. Courbebaisse, B. Chopard, and F. Dubois, “Quantitative analysis of platelets aggregates in 3D by digital holographic microscopy,” Biomed. Opt. Express 6(9), 3556–3563 (2015). [CrossRef]
17. Y. Kitamura, K. Isobe, H. Kawabata, T. Tsujino, T. Watanabe, M. Nakamura, T. Toyoda, H. Okudera, K. Okuda, K. Nakata, and T. Kawase, “Quantitative evaluation of morphological changes in activated platelets in vitro using digital holographic microscopy,” Micron 113, 1–9 (2018). [CrossRef]
18. S. Lee, S. Jang, and Y. Park, “The study of morphology and dynamics of platelets in an activation process using 3-D quantitative phase imaging,” Circulation 140, A15342 (2019). [CrossRef]
19. T. A. Stanly, R. Suman, G. Fatima Rani, P. J. O’Toole, P. M. Kaye, and I. S. Hitchcock, “Quantitative optical diffraction tomography imaging of mouse platelets,” Front. Physiol. 11, 568087 (2020). [CrossRef]
20. Y. Zhou, A. Isozaki, A. Yasumoto, T.-H. Xiao, Y. Yatomi, C. Lei, and K. Goda, “Intelligent platelet morphometry,” Trends Biotechnol. 39(10), 978–989 (2021). [CrossRef]
21. F. Dubois and C. Yourassowsky, “Full off-axis red-green-blue digital holographic microscope with LED illumination,” Opt. Lett. 37(12), 2190–2192 (2012). [CrossRef]
22. C. Yourassowsky and F. Dubois, “High throughput holographic imaging-in-flow for the analysis of a wide plankton size range,” Opt. Express 22(6), 6661–6673 (2014). [CrossRef]
23. J. Dohet-Eraly, C. Yourassowsky, and F. Dubois, “Color imaging-in-flow by digital holographic microscopy with permanent defect and aberration corrections,” Opt. Lett. 39(20), 6070–6073 (2014). [CrossRef]
24. F. Dubois, C. Schockaert, N. Callens, and C. Yourassowsky, “Focus plane detection criteria in digital holography microscopy by amplitude analysis,” Opt. Express 14(13), 5895–5908 (2006). [CrossRef]
25. J. Dohet-Eraly, C. Yourassowsky, and F. Dubois, “Refocusing based on amplitude analysis in color digital holographic microscopy,” Opt. Lett. 39(5), 1109–1112 (2014). [CrossRef]
26. J. Dohet-Eraly, L. Méès, T. Olivier, F. Dubois, and C. Fournier, “Analysis of three-dimensional objects in quantitative phase contrast microscopy: a validity study of the planar approximation for spherical particles,” Proc. SPIE 11351, 113510Y (2020). [CrossRef]
27. M. Schoorl, M. P. C. Grooteman, P. C. M. Bartels, and M. J. Nubé, “Aspects of platelet disturbances in haemodialysis patients,” Clin. Kidney J. 6(3), 266–271 (2013). [CrossRef]
28. S. Lickert, S. Sorrentino, J.-D. Studt, O. Medalia, V. Vogel, and I. Schoen, “Morphometric analysis of spread platelets identifies integrin αIIbβ3-specific contractile phenotype,” Sci. Rep. 8, 5428 (2018). [CrossRef]
29. J. A. Pike, V. A. Simms, C. W. Smith, N. V. Morgan, A. O. Khan, N. S. Poulter, I. B. Styles, and S. G. Thomas, “An adaptable analysis workflow for characterization of platelet spreading and morphology,” Platelets 32(1), 54–58 (2021). [CrossRef]
