scispace - formally typeset
Search or ask a question
Book ChapterDOI

A comparative study of image processing methods for the assessment of the red blood cells deformability in a microfluidic device

18 Oct 2017-Vol. 27, Iss: 1, pp 923-929
TL;DR: In this article, the authors compare different image processing methods for assessing the RBCs deformability in a continuous flow, measured in a polydimethylsiloxane (PDMS) microchannel composed by 15 μm spacing inner pillars.
Abstract: Red blood cells (RBCs) deformability is a high relevant mechanical property, whose variations are associated with some diseases, such as diabetes and malaria. Therefore, the present study aims to compare different image processing methods for assessing the RBCs deformability in a continuous flow, measured in a polydimethylsiloxane (PDMS) microchannel composed by 15 μm spacing inner pillars. The images were acquired with a high speed camera and analyzed with ImageJ software for tracking and measuring the RBCs deformation index (DI). Additionally, to understand the performance of the software, it was performed a comparison between different image processing tools provided by ImageJ and the best methods for the deformation measurements were selected, considering the measured RBCs number and their DIs. The results show that those image methods significantly affect the number of measured RBCs and their DIs and, therefore, the studies focused on the deformability measurements need to take into account the effect of the image processing methods for avoiding loss of relevant information in the images.

Summary (2 min read)

1 Introduction

  • Due to its complexity, many analytical techniques require the previous separation and sorting of blood cells, which consumes time and resources [1, 2].
  • The use of specific geometries, such as micropillars arrays, micro-weirs or membranes with holes, inside the microfluidic devices, could be helpful for cell separation.
  • In these geometries, the size of the geometries determines the cells deformation [4].
  • For the best of the authors knowledge, the use of such structures was never used for separation and, at the same time, for measuring the deformation index of RBCs.

2.1 Microchannel Fabrication and Geometry

  • The PDMS microfluidic device used in this work was fabricated by using a soft lithography technique.
  • More detail about the fabrication technique can be found elsewhere [7, 8].
  • Figure 1 shows the main dimensions of the microfluidic device used in this study.

2.2 Working Fluid and Experimental Set-up

  • Human blood of a healthy donor was collected into 2.7 mL tubes (S-Monovette®, Sarstedt) containing ethylenediaminetetraacetic acid (EDTA).
  • Plasma and the buffy coat were removed and the RBCs were re-suspended and washed once in physiological salt solution (PSS) with 0.9% NaCL (B. Braun Medical, Germany).
  • The working fluid used was Dextran 40 (Sigma-Aldrich, USA) solution containing 5% of haematocrit (Hct).
  • At the same time, the images of the flowing cells at the established flow rate were captured by the high speed camera at a frame rate of 2000 frames/s and a shutter speed ratio of 1/75000, which minimized the dragging of the cells normally seen at the high flow rates in study.

2.3 Image Analysis

  • The experimental images recorded in each test were transferred to a computer, processed and analyzed by an image handling software, ImageJ (1.46r, NIH, USA) [9].
  • The differences between these methods are presented in Table 1.
  • All the performed analysis was carried out with the same segment of the captured video, containing 102 frames.
  • It was created an averaged image from the original stack of images, by averaging each pixel over the sequence of static images and then subtracted the from the images stack.
  • Then, a threshold was applied using the otsu algorithm and, finally, the RBCs were manually measured.

3 Results and Discussion

  • After applying the referred methods, the RBCs were manually measured and their DI was calculated.
  • Figure 4 presents the comparison of the three applied methods.
  • Therefore, considering both the shape of the cells (Fig. 3) and the measured DIs (Fig. 4), it can be concluded that, even with a minor measured RBCs number, the method 2 is a more reliable representation of the RBCs deformation behavior through micropillars, and consequently, it is the most suitable for processing the images and for improving their quality.
  • The used threshold in methods 1 and 3 had influence the RBCs choice, i.e., this threshold implied the selection of the interior region of the RBCs instead of the exterior membrane (as occurred in method 2).
  • The contrast option did not have a considerable influence on the results, but improved the image quality.

4 Conclusion

  • The presented results show that the chosen image processing methods significantly affect the number of measured RBCs and their DIs and, therefore, the studies focused on the deformability measurements need to take into account the effect of those methods for avoiding loss of relevant information in the images.
  • Considering that, the method 2 was superior, once the outputted results besides being in conformity with other studies, in which the DI value was measured, are a better representation of the RBCs structure.
  • The pre-processing of microscale images for further extraction of information is a great challenge, even when good quality videos are acquired.
  • It is essential to ensure that the data presented in the frames remain almost intact.
  • This work was supported by FCT with the reference project UID/EEA/ 04436/2013, by FEDER funds through the COMPETE 2020 – Programa Operacional Competitividade e Internacionalização (POCI) with the reference project POCI-01- 0145-FEDER- 006941.

Did you find this useful? Give us your feedback

Content maybe subject to copyright    Report

A Comparative Study of Image Processing
Methods for the Assessment of the Red Blood
Cells Deformability in a Microuidic Device
Vera Faustino
1,2,3
, Susana O. Catarino
1
, Diana Pinho
2,4
,
Graça Minas
1
, and Rui Lima
2,3,4(&)
1
Microelectromechanical Systems Research Unit (CMEMS-UMinho),
University of Minho, Campus de Azurém, 4800-058 Guimarães, Portugal
2
Department of Chemical Engineering, Engineering Faculty,
University of Porto, Transport Phenomena Research Center,
R. Dr. Roberto Frias, 4200-465 Porto, Portugal
rl@dem.uminho.pt
3
MEtRiCS, Department of Mechanical Engineering, Minho University,
Campus de Azurém, 4800-058 Guimarães, Portugal
4
Polytechnic Institute of Bragança, ESTiG/IPB, C. Sta. Apolonia,
5301-857 Bragança, Portugal
Abstract. Red blood cells (RBCs) deformability is a high relevant mechanical
property, whose variations are associated with some diseases, such as diabetes
and malaria. Therefore, the present study aims to compare different image
processing methods for assessing the RBCs deformability in a continuous ow,
measured in a polydimethylsiloxane (PDMS) microchannel composed by 15 lm
spacing inner pillars. The images were acquired with a high speed camera and
analyzed with ImageJ software for tracking and measuring the RBCs defor-
mation index (DI). Additionally, to understand the performance of the software,
it was performed a comparison between different image processing tools pro-
vided by ImageJ and the best methods for the deformation measurements were
selected, considering the measured RBCs number and their DIs. The results
show that those image methods signicantly affect the number of measured
RBCs and their DIs and, therefore, the studies focused on the deformability
measurements need to take into account the effect of the image processing
methods for avoiding loss of relevant information in the images.
Keywords: Deformation index
Red blood cells Deformability
Microuidics Image processing
1 Introduction
Blood is a complex uid full of valuable information. Due to its complexity, many
analytical techniques require the previous separation and sorting of blood cells, which
consumes time and resources [1, 2]. Microuidic devices have arisen to overcome
some limitations of the existing techniques and make it possible the use of small
samples and less biochemical labels that may change cells properties [3].
© Springer International Publishing AG 2018
J.M.R.S. Tavares and R.M. Natal Jorge (eds.), VipIMAGE 2017,
Lecture Notes in Computational Vision and Biomechanics 27,
DOI 10.1007/978-3-319-68195-5_100

The use of specic geometries, such as micropillars arrays, micro-weirs or mem-
branes with holes, inside the microuidic devices, could be helpful for cell separation.
In these geometries, the size of the geometries deter mines the cells deformation [4].
Many literature studies use these structures (micropillars arrays and micro-weir lters),
which create a cross-ow ltration to overcome some issues related with clogging and
jamming [5, 6]. However, for the best of the authors knowledge , the use of such
structures was never used for separation and, at the same time, for measuring the
deformation index of RBCs. So, in this study, it is proposed a PDMS microuidic
device with 19 µm height, and with a sequence of 15 µm separation pillars with
45 50 µm, where was measured the DI of RBCs. To choose the best method that can
improves the images quality and measuring the RBCs deformation index through the
pillars, different image processing methods were compared using the ImageJ software.
2 Materials and Methods
This section presents the geometry, experimental setup and materials used in the
experimental procedures for measuring the RBCs deformability.
2.1 Microchannel Fabrication and Geometry
The PDMS microuidic device used in this work was fabricated by using a soft
lithography technique. More detail about the fabrication technique can be found
elsewhere [7, 8]. Figure 1 shows the main dimensions of the microuidic device used
in this study.
Fig. 1. Schematic representation and main dimensions of microuidic device.
924 V. Faustino et al.

2.2 Working Fluid and Experimental Set-up
Human blood of a healthy donor was collected into 2.7 mL tubes (S-Monovette®,
Sarstedt) containing ethylenediaminetetraacetic acid (EDTA). The whole blood was
centrifuge at 2500 rpm for 10 min at 20 °C. Plasma and the buffy coat were removed
and the RBCs were re-suspended and was hed once in physiological salt solution
(PSS) with 0.9% NaCL (B. Braun Medical, Germany). The working uid used was
Dextran 40 (Sigma-Aldrich, USA) solution containing 5% of haematocrit (Hct).
The high-speed video microscopy system used in the present study consisted of an
inverted microscope (IX71, Olympus) combined with a high-speed camera (Fastcam
SA3, Photron, USA), as shown in Fig. 2. The PDMS microchannel was placed and
xed in the microscope and the ow rate of the working uid was kept constant at
50 µL/min using a syringe pump (PHD Ultra, Harvard Apparatus, USA) with a 5 mL
syringe (Terumo, Japan). At the same time, the images of the owing cells at the
established ow rate were captured by the high speed camera at a frame rate of 2000
frames/s and a shutter speed ratio of 1/75000, which minimized the dragging of the
cells normally seen at the high ow rates in study. Al l the experimental assays were
performed at room temperature (T = 22 ± C).
2.3 Image Analysis
The experimental images recorded in each test were transferred to a computer, pro-
cessed and analyzed by an image handling software, ImageJ (1.46r, NIH, USA) [9].
Then, it was used three different methods for comparing the different methods provided
by the ImageJ, in the RBCs measuring and in the assessment of their deformability
aiming the improvement of the individual RBCs imaging. The differences between
these methods are presented in Table 1. All the performed analysis was carried out with
the same segment of the captured video, containing 102 fram es. The rst step of the
pre-processing image was equally applied to the three methods. In this step, it was
Fig. 2. Experimental set-up.
A Comparative Study of Image Processing Methods 925

created an averaged image (background) from the original stack of images, by aver-
aging each pixel over the sequence of static images and then subtracted the background
from the images stack. With this step, all the static objects, dusts and the microchannel
walls are eliminated and only the RBCs, which move through the microchannel, remain
visible. The following steps consisted in subtr acting the median of the images stack
from the original stack images. Then, a threshold was applied using the otsu algorithm
and, nally, the RBCs were manually measured. The differences between the methods
are focused in the subtraction options, the threshold values and the contrast options of
each method (see Table 1).
Figure 3 presents examples of RBCs that were obtained after all the processing
steps.
Table 1. Differences between the three used methods, in ImageJ software, focused on
subtraction, threshold values and contrast options of each method.
Method Subtraction Threshold Contrast options
1 Subtract median Stacks from images stacks, with 8bit Min: 14
Max: 255
Not used
2 Subtract median Stacks from images stacks, with 32bit oat
result
Min: 41
Max: 7
Not used
3 Subtract median Stacks from images stacks, with 32bit oat
result
Min: 1.71
Max: 134
Min: 37.27
Max: 103.23
Fig. 3. Representation of the RBCs (zoom of one RBC) obtained after image processing using
the different ImageJ methods (a) method 1; (b) method 2; (c) method 3. The measurements were
made between pillars.
926 V. Faustino et al.

For each cell, the DI was calculated using the equation DI = (X Y)/(X + Y),
where X and Y are the major and the minor axis length of each RBC, respectively
(Fig. 2).
3 Results and Discussion
After applying the referred methods, the RBCs were manually measured and their DI
was calculated.
Figure 4 presents the comparison of the three applied methods. Methods 1 and 3
were the ones that allowed the measurement of more RBCs (21 and 25, respectively),
and had higher DI values. However, taking into account the theoretically previewed
values, the method 2 corresponds to a best representation of a real RBC [10, 11], also
seen by Fig. 3(b).
Therefore, considering both the shape of the cells (Fig. 3) and the measured DIs
(Fig. 4), it can be concluded that, even with a minor measured RBCs number, the
method 2 is a more reliable representation of the RBCs deformation behavior through
micropillars, and consequently, it is the most suitable for processing the images and for
improving their quality.
The used threshold in methods 1 and 3 had inuence the RBCs choice, i.e., this
threshold implied the selec tion of the interior region of the RBCs instead of the exterior
membrane (as occurred in method 2). This reason explains the higher DI and the larger
obtained number of RBCs.
Fig. 4. Averaged DI values of the RBCs for the methods 1, 2 and 3, with standard deviation
error (0.0511; 0.0640 and 0.0541, respectively) using ImageJ. The numbers in the data columns
represent the total RBCs number evaluated by each method over 102 frames.
A Comparative Study of Image Processing Methods 927

References
More filters
Journal ArticleDOI
TL;DR: A wide range of microfluidic technologies have been developed to separate and sort cells by taking advantage of differences in their intrinsic biophysical properties, including size, electrical polarizability, and hydrodynamic properties.
Abstract: Cell separation and sorting are essential steps in cell biology research and in many diagnostic and therapeutic methods. Recently, there has been interest in methods which avoid the use of biochemical labels; numerous intrinsic biomarkers have been explored to identify cells including size, electrical polarizability, and hydrodynamic properties. This review highlights microfluidic techniques used for label-free discrimination and fractionation of cell populations. Microfluidic systems have been adopted to precisely handle single cells and interface with other tools for biochemical analysis. We analyzed many of these techniques, detailing their mode of separation, while concentrating on recent developments and evaluating their prospects for application. Furthermore, this was done from a perspective where inertial effects are considered important and general performance metrics were proposed which would ease comparison of reported technologies. Lastly, we assess the current state of these technologies and suggest directions which may make them more accessible.

889 citations


"A comparative study of image proces..." refers background in this paper

  • ...Microfluidic devices have arisen to overcome some limitations of the existing techniques and make it possible the use of small samples and less biochemical labels that may change cells properties [3]....

    [...]

Journal ArticleDOI
TL;DR: Based on the experimental results, the crossflow microfilter is superior and can be integrated with downstream components for on-chip genomic analysis.
Abstract: This paper reports on the comparison analysis of four main types of silicon-based microfilter for isolation of white blood cells (WBCs) from red blood cells (RBCs) in a given whole blood. The microfilter designs, namely, weir, pillar, crossflow, and membrane, all impose the same cut-off size of 3.5 mum to selectively trap WBCs. Using human whole blood, the microfilters have been characterized and compared for their blood handling capacity, WBCs trapping efficiency and RBCs passing efficiency. Based on the experimental results, the crossflow microfilter is superior and can be integrated with downstream components for on-chip genomic analysis.

267 citations


"A comparative study of image proces..." refers background in this paper

  • ...Many literature studies use these structures (micropillars arrays and micro-weir filters), which create a cross-flow filtration to overcome some issues related with clogging and jamming [5, 6]....

    [...]

Journal ArticleDOI
TL;DR: In this article, two kinds of microfluidic chips based on the cross-flow filtration principle were designed and fabricated by microelectromechanical system (MEMS) technology, in which parallel micropillar arrays and parallel microweirs were used to separate cells via their different sizes.
Abstract: Whole blood is a mixture of various cells, such as red blood cell (RBC), white blood cell (WBC) and so on Separation and collection of WBC and RBC, starting from a sample of whole blood, are the required steps for the subsequent clinica and basic research assays We created two kinds of microfluidic chips based on the crossflow filtration principle which can be more effective than conventional types in the area of avoiding clogging or jamming Pillar-type and weir-type filtration microchips were designed and fabricated by microelectromechanical system (MEMS) technology, in which parallel micropillar-array and parallel microweirs were used to separate cells via their different sizes After separation, WBC and RBC were collected, respectively Cell concentration and the length of separation channels were investigated and optimized Under the optimal condition, more than 95% RBC can be removed from the initial whole blood, while 274% WBC can be obtained Isolation efficiency of WBC by using the crossflow filtration microchip is approximately twice as high as that of the dead-end filtration microchip Furthermore, plasma, WBC and RBC can be simultaneously separated and collected at different outlet ports with multilevel filtration barriers

243 citations


"A comparative study of image proces..." refers background in this paper

  • ...In these geometries, the size of the geometries determines the cells deformation [4]....

    [...]

Journal ArticleDOI
TL;DR: In this review, a selection of the most recent lithographic and non-lithographic low-cost techniques to fabricate microfluidic structures, focused on the features and limitations of each technique.

235 citations


"A comparative study of image proces..." refers methods in this paper

  • ...More detail about the fabrication technique can be found elsewhere [7, 8]....

    [...]

Frequently Asked Questions (2)
Q1. What have the authors contributed in "A comparative study of image processing methods for the assessment of the red blood cells deformability in a microfluidic device" ?

Therefore, the present study aims to compare different image processing methods for assessing the RBCs deformability in a continuous flow, measured in a polydimethylsiloxane ( PDMS ) microchannel composed by 15 lm spacing inner pillars. Additionally, to understand the performance of the software, it was performed a comparison between different image processing tools provided by ImageJ and the best methods for the deformation measurements were selected, considering the measured RBCs number and their DIs. 

Therefore, further studies on the comparison between image processing methods for microfluidics could be a key factor for improving the results, increasing their reliability.