Вы находитесь на странице: 1из 6

Towards Quantitative, Low-Cost, Webcam-Based Microscopy for Medical Diagnostics

Marten Darmawan, Amir Faisal, Gea O.F. Parikesit

Microfluidics Laboratory, Department of Engineering Physics, Faculty of Engineering, Gadjah Mada University,
Yogyakarta 55281, Indonesia

marten_darmawan@yahoo.com, amirf415al@yahoo.com, geaofp@yahoo.com

Abstract

Digital web cameras (popularly known as webcams) have recently gained a significant increase of relevance in
the field of optical microscopy, in particular to allow for quick and do-it-yourself methods in developing low-
cost and portable microscopes suitable for medical diagnostic applications in low-resource areas. Unfortunately,
these methods were published without any systematic explanation and quantitative assessment of the imaging
performances. In this paper we reproduce these do-it-yourself methods, discuss the optical considerations that
are relevant for them, and quantitatively compare their imaging performances with a commercial digital
microscope in order to clarify both the advantages and disadvantages of using the webcam-based microscopes
for medical diagnostics purposes.

Keywords: microscopy, webcam, do-it-yourself, quantitative imaging, medical diagnostics.

I. Introduction for them, and quantitatively assess their imaging


The seminal works performed by Hooke performances. Further comparisons with
and van Leeuwenhoek [1] have arguably performances of a commercial digital microscope
transformed optical microscopes into standard were also performed in order to clarify the
scientific instrumentations, but it was the discovery advantages and disadvantages of using the
of digital sensor technologies that has allowed the “webcam-based microscopes” for medical
microscopes to function as quantitative diagnostics purposes.
measurement tools [2] versatile for various
applications, spanning from biomedical II. Methods
investigations [3] to engineering research [4]. In general, there are three different do-it-yourself
Pollak and Hutter reported their method [5] for methods that have been published: (i) the lens-less
mounting a digital web camera (popularly known as method, where the lens is simply removed from the
‘webcam’) on an optical microscope, effectively webcam [8]; (ii) the inverted-lens, where the lens is
providing a low-cost version of digital microscopy. inverted and repositioned in the webcam [9]; and
Other methods, using cellular phones, further (iii) the extra-lens method, where another lens is
demonstrate a level of portability that can allow added into the webcam system [10]. In this paper
digital microscopes to work efficiently in low- we focus only on the first two methods, since these
resource area [6,7]. are the most straightforward methods that do not
Recently some researchers have devised require components other than the ones existing
methods to develop “webcam-based microscopes”, within the webcam. To reproduce the do-it-yourself
where webcams were directly turned into digital methods, we use a Prolink webcam (type:
microscopes, hence circumventing the requirement PCC8020; sensor: CMOS; 1280 x 1024 pixels;
to possess an existing microscopy setup [8,9,10]. interface: USB 2.0; maximum frame rate: 30
Even though these quick and do-it-yourself frames per second; F# of the lens: 2.8; see
methods have successfully demonstrated both low- http://www.prolink2u.com/new/products/index.php
cost and portability, these methods were published ?cid=102). The Prolink webcam was chosen mainly
without a systematic explanation and quantitative due to its low cost (i.e. around US$ 30.00) and
assessment of the imaging performances. In this commercial availability at our geographical area.
paper1, we reproduce these do-it-yourself methods, Below we describe the configurations of the
discuss the optical considerations that are relevant reproduced do-it-yourself webcam-based
microscopes. Figure 1 shows the webcam in (a) its
1
A more comprehensive version of this manuscript has been original form, (b) after modification using the lens-
submitted for publication in the journal of Optical Engineering:
http://spie.org/x867.xml
less method [8], and (c-d) after further modification product specification) is 2.8; D (measured) is 3.7
using the inverted-lens method [9]. mm. After the lens was inverted and repositioned,
the webcam system was enclosed in a plastic jar
(from Skippy peanut butter,
http://www.skippypeanutbutter.com/index.aspx),
such that fine adjustments of the object plane
position can be achieved simply by screwing in or
out the threaded lid. For the illumination, we again
(a) (b) (c) use a LED lamp as the illumination source, located
on top of the object.
In order to comprehensively evaluate the
webcam-based microscopes, we also perform
measurements with a commercial digital
microscope. For this, we use a QX5 Computer
Microscope from Digiblue, with 640 x 480 pixels,
where we use the 60x and 200x magnification
together with the bottom illumination source
(http://store.digiblue.com/QX5_Computer_Microsc
ope_for_PC_p/db12013.htm). Note that such a
(d) microscope has been used for scientific research
before, particularly for low-cost gel electrophoresis
Figure 1. (a) The webcam in its original form; (b) analysis of DNA molecules [13].
After removal of lens and its holder; (c) After All these microscopes were tested using
repositioning of the lens and enclosing of the several periodic objects available at our laboratory.
system in a plastic jar; (d) Schematics of the lens The first object is a 1951 USAF Resolution Target
repositioning (not to scale). (Edmund Optics, Barrington, NJ, USA). This
resolution target, commonly used to determine the
The lens-less method is principally similar resolution of imaging systems, consists of arrays of
to the lens-less Optofluidic Microscope method periodic lines (i.e. opaque lines on translucent
reported by Heng et al. [11], where microscopic background) with varying spatial periodicity. The
objects were imaged directly into an array of digital second test object is a grooved microchannel (width
sensors, without any optical element in between. = 100 µ m, depth = 1 µ m, translucent grooves in
After the webcam’s body has been dismantled, the translucent channels), with grooves periodicity of 5,
lens and its support were removed from the setup. 10, and 15 µ m [14]. This test object is useful for
A thin glass coverslip (thickness of 170 µ m) was testing the contrast and resolution obtained
then put on the CMOS sensor, both to cover it from particularly when both the objects and backgrounds
dust and to place samples directly on top of the are translucent, which is relevant should the
sensor. Meanwhile, adhesive tapes were used to developed microscopes be used with unstained
cover the electronics around the CMOS sensor from biological cells. The digital images were captured
dusts and wet bio-samples. Due to the close and stored in our laptops, before being analyzed
distance between the sensor and the object, the using the software ImageJ
CMOS pixel arrays will capture the image with (http://rsbweb.nih.gov/ij/).
essentially no magnification. The imaging
resolution will therefore depend only on the pixel III. Result and Discussion
size. For the illumination source, we used a LED III.1. Resolution Target
lamp that is positioned above the object plane. A In this test we focus only on the inverted-
black tube, made from cardboard, was positioned lens microscope and the QX5 microscope, as the
around the webcam to prevent stray lights from the lens-less microscope effectively has no optical
ambience to reach our sensor. magnification (and its resolution is therefore
Meanwhile, the inverted-lens method is defined simply by their pixel size). Figure 2
actually an adaptation from the “reverse macro” presents the digital images of the 1951 USAF
technique used in photography [12], which is Resolution Target captured with the inverted-lens
popularly used as a low-cost alternative to produce microscope and the QX5 microscope. Comparisons
images with a large magnification and a short depth of these images indicate two important things. First,
of focus. Figures 1(c-d) illustrate the configuration the inverted-lens microscope’s field of view is
used in our reproduction of the inverted-lens comparable to the QX5 microscope with a
method. In order to obtain a focused image plane at magnification between 60x and 200x. Second, the
the sensor, the object plane need to be located 10.36 inverted-lens microscope can resolve a minimum
mm on top of the inverted lens (see f in Figure 1.e), spatial periodicity of 6.2 µ m (corresponding to
based on the following parameters: F# (given in the Group 7 and Element 3 in the Resolution Test),
which is superior to the resolutions obtained with resulting illumination is not ideally uniform, as the
the QX5 microscope, 12.4 µ m (200x light beams coming out of the LED lamp may
magnification). Note that we also performed the possess different directions of propagation. This, in
resolution test with QX5 microscope for 10x and turn, leads to shading at the microchannel
60x magnification, but these results (not shown sidewalls, as can be observed from the intensity
here) were inferior to the one obtained with the dips shown in Figures 5(a) and 5(b) that occur at
200x lens. the sidewalls locations.

(a) (b) (a)


Figure 2. Digital images of the 1951 USAF
Resolution Target captured with: (a) the inverted-
less microscope; (b) the QX5 microscope, with
magnifications of 200x.

III.2. Grooved Microchannel


The test reported in this section is similar,
but more challenging, than in the previous section,
since here both the object and the background are
translucent, which would generally result in low- (b)
contrast images. Figure 3 displays the obtained raw
images, indicating the locations of the grooves in
the channels. Meanwhile, Figures 4 and 5 show the
image intensity distributions along the channel
length and the channel width, respectively, and
Figure 6 displays the Fourier transformation of the
images shown in Figure 3.
The data shown in Figures 4 and 6 shows
that the best spatial resolution, quantitatively
represented by a periodicity of the digital image (c)
intensity, was obtained with the inverted-lens Figure 3. The digital raw images obtained with (a)
microscope. Both the lens-less microscope and the the lens-less microscope, (b) the inverted-lens
QX5 microscope fail to show any detectable microscope, and (c) the QX5 microscope (with 60x
periodicity of the grooves (see Figures 4(a) and magnification), for the grooves periodicity of (left)
4(c)), as also shown in the Fourier spectra (see 5 µ m, (middle) 10 µ m, and (right) 15 µ m.
Figures 6(a) and 6(c)). The inverted-lens
microscope, however, only succeeds to resolve the Im a g e inte nsity v a ria tio n a lo ng the cha n ne l le ng th
210
grooves periodicities of 15 µ m and 10 µ m, but
not 5 µ m (see Figures 4(b) and 6(b)). This agrees 205
well with the results in the previous section, where
the minimum spatial periodicity resolvable with the 200

inverted-lens microscope is 6.2 µ m. 195


Figures 4 and 5 indicate that the lowest
image contrast, quantitatively represented by the
Counts

190
intensity differences between the channel and its
surrounding sidewalls, was obtained with the QX5 185

microscope. This caveat was caused, ironically, by


180
the good illumination distribution of the QX5 5 micron
10 micron
microscope, where the light coming out from the 175 15 micron
LED lamp is passed through a diffuser to ensure
uniform illumination intensity across the whole 170
0 20 40 60 80 100 120 140 160 180 200
field of view, while the microscope object is placed Pixels

sufficiently close to the diffuser. As the webcam-


based microscopes do not use such a diffuser, their (a)
(a)
Im a g e inte nsity v a ria tio n a lo ng the cha nne l leng th
230
Ima ge intensity va ria tio n a lo ng the cha nnel w idth
5micron 210
220 10micron
15micron
200
210
190
200
180
190
Counts

Counts
170
180

160
170

160 150
5micron
10m icron
150 140 15m icron

140 130
0 50 100 150 200 250 300 350 400 450 500 0 20 40 60 80 100 120
Pixels Pixels

(b) (b)
Ima ge intensity va ria tio n a lo ng the cha nnel leng th Im a g e in te n sity v a ria tio n a lo ng th e ch a n ne l w id th
123 126
5micron
122 10micron
15micron 124
121
122
120

119 120
Counts

Counts

118
118
117

116
116
5micron
115 114 10 micron
15 micron
114
0 50 100 150
Pixels 112
0 10 20 30 40 50 60
Pixels

(c)
Figure 4. Image intensity distributions along the (c)
channel length (i.e. parallel to the grooves
periodicity), obtained with (a) the lens-less Figure 5. Image intensity distributions along the
microscope, (b) the inverted-lens microscope, and channel width (i.e. perpendicular to the
(c) the QX5 microscope (with 60x magnification). microchannel sidewalls), obtained with (a) the lens-
less microscope, (b) the inverted-lens microscope,
Im a g e inte nsity v a ria tio n a lo ng the cha nnel w id th and (c) the QX5 microscope (with 60x
190
magnification).

180

170
Counts

160

150
5micron
10micron
15micron
140

130
0 10 20 30 40 50 60
Pixels
IV. Conclusions
Based on the results described above, we
can conclude that our reproduction of the inverted-
lens microscope provides the most optimum
performance. It allows us to obtain the best spatial
resolution (6.2 µ m), which is twice better than the
(a) QX5 microscope (12.4 µ m, at 200x
magnification). Equipped with this good resolution,
the inverted-lens microscope was also able to
resolve translucent objects with a translucent
background, for up to a spatial periodicity of 10
µ m, which could be relevant for life sciences
investigations on unstained biological cells that
(b) further can be support medical diagnostics. There
are, however, two caveats in the inverted-lens
microscope. First, the illumination is not uniform
(yet), but this can be further engineered using a
light diffuser. The second caveat is the optical
aberrations (data not shown in this paper), which is
due to the inferior lens quality compared to the
(c) QX5 microscope. This problem, however, can be
Figure 6. The Fourier transformation of the raw solved using digital image processing [17-18]. Note
images obtained with (a) the lens-less microscope, that the inverted-lens setup would not allow us to
(b) the inverted-lens microscope, and (c) the QX5 obtain a Koehler-type illumination [19], mainly
microscope (with 60x magnification), for the because it only employs a single lens.
grooves periodicity of (left) 5 µ m, (middle) 10 The lens-less microscope is the easiest
µ m, and (right) 15 µ m. setup that can be developed, but it involves
essentially no magnification. This can also be
The intensity dips displayed by Figures advantageous, however, because it also means that
5(a) and 5(b) can also be used to quantitatively we obtain no optical aberration. The decision to
estimate the locations of the sidewalls, and choose between a lens-less microscope, an
eventually to also estimate the microchannel width. inverted-lens microscope, or a commercial digital
Table 1 shows the results of these estimations, microscope would therefore depend on the specific
using 1000 similar measurements along lines requirements of the applications. Another
perpendicular to the channel sidewalls, in which the consideration would be the cost and time available:
sidewalls locations are estimated by finding the the commercially-available QX5 microscope costs
center-of-mass of the dips [16]. The values in the about US$ 100 and is directly ready to use after
table, together with the actual width of the channels purchase, while the Prolink webcam costs only Rp.
(100 µ m), allow us to further estimate two 300,000.00 (i.e. around US$ 30,00) but requires
parameters of the webcam. First, the CMOS pixel several hours to develop.
size in the webcam can be estimated as 3.2 In conclusion, this paper reports on the
µ m/pixel (i.e. 100 µ m per 31.3 pixels). Second, reproduction of do-it-yourself methods to develop
the effective magnification obtained by inverting low-cost digital microscopy, discuss the optical
and repositioning the webcam lens can be estimated considerations that are relevant for these methods,
by comparing the dimensions (in pixels) in the and quantitatively compare their imaging
images obtained by the lens-less and the inverted- performances with a commercial digital microscope
lens microscopes; the estimated value of in order to clarify both the advantages and
magnification is 2.6x (i.e. 82.5/31.3). disadvantages of the webcam-based microscopes.
While the comparisons discussed in this paper has
Table 1. Quantitative estimation of the sidewalls been the result of measurements using a Prolink
locations and the microchannel width webcam-based microscope and a QX5 microscope,
a similar analysis method can obviously be
employed to compare the imaging performances of
any arbitrary pair of a webcam-based microscope
and a commercial digital microscope.

V. Acknowledgement
We gratefully acknowledge discussions
with Alex Gunawan, Aura Mimosa Nugrowati, and
Indraswari Kusumaningtyas.
References [16] "The Digital Signal Processing Handbook",
[1] “The discovery of microorganisms by Robert I.T. Young, J.J. Gerbrands, and L.J. van Vliet, CRC
Hooke and Antoni van Leeuwenhoek, Fellows of Press, Boca Raton, 1998, Chap. 51.
The Royal Society”, H. Gest, Notes Rec. R. Soc. [17] "An efficient algorithm for measurement and
Lond. (2004) 58:187-201. correction of chromatic aberrations in fluorescence
[2] “Nobel Prize for the Invention of the CCD microscopy", M. Kozubek and P. Matula, Journal
Sensor”, Martin Friedrich, Thomas Matzelle; of Microscopy (2001), 200:206-221.
Imaging and Microscopy (2009) 11:3. [18] "An embedded camera lens distortion
[3] “Multiphoton microscopy in life sciences”, K. correction method for mobile computing
Koenig, Journal of Microscopy (2001) 200:83-104. applications", W. Yu, IEEE Transactions on
[4] “Highly accurate non-contact characterization Consumer Electronics (2003), 49: 894-901.
of engineering surfaces using confocal [19] “Fundamentals of light microscopy and
microscopy”, H-J Jordan, M Wegner, and H electronic imaging“, D.B. Murphy, Wiley-Liss,
Tiziani, Measurement Science and Technology New York, 2001, Chap. 1.
(1998) 9:1142.
[5] “A Webcam as Recording Device for Light
Microscopes”, C. Pollak and H. Hutter, Journal of
Computer-Assisted Microscopy (1998) 10:179-183.
[6] “Lensfree microscopy on a cellphone”, D.
Tseng, O. Mudanyali, C. Oztoprak, S.O. Isikman, I.
Sencan, O. Yaglidere and A. Ozcan, Lab Chip
(2010), DOI: 10.1039/c003477k
[7] “Lensfree on-chip microscopy over a wide
field-of-view using pixel super-resolution”, W.
Bishara, T-W. Su, A.F. Coskun, and A. Ozcan,
Optics Express (2010) 18:11181.
[8] http://www.instructables.com/id/How-to-make-
a-USB-Microscope-for-under-15/ and
http://blog.makezine.com/archive/2008/06/weekend
_project_lensless.html, accessed on June 8th, 2010.
[9] http://tweaklabs.org/Webcam+Microscope,
accessed on June 8th, 2010.
[10]
http://blog.makezine.com/archive/2006/08/homema
de_microscope_using.html, accessed on June 8th,
2010.
[11] “Optofluidic microscopy—a method for
implementing a high resolution optical microscope
on a chip”, X. Heng, D. Erickson, L.R. Baugh, Z.
Yaqoob, P.W. Sternberg, D. Psaltis and C. Yang,
Lab Chip (2006) 6:1274-1276.
[12]
http://home.comcast.net/~flash19901/tt_on_macro_
photo.htm, accessed on June 9th, 2010.
[13] "An inexpensive microslab gel DNA
electrophoresis system with real-time fluorescence
detection", X. Chen and V.M. Ugaz,
Electrophoresis (2006) 27:387-393.
[14] “Observation of hydrophobic-like behavior in
topologically-patterned hydrophilic microfluidics“,
G.O.F. Parikesit, E.X. Vrouwe, M.T. Blom, J.
Westerweel; submitted to Physical Review Letters,
2010.
[15] “Fabrication of nanofluidic devices in glass
with polysilicon electrodes”, V.G. Kutchoukov, L.
Pakula, G.O.F. Parikesit, Y. Garini, L.K. Nanver,
A. Bossche; Sensors and Actuators A: Physical
(2005) 1213-1214:602-607.

Вам также может понравиться