MOBILE PHONE BASED
CLINICAL MICROSCOPY FOR
GLOBAL HEALTH
APPLICATIONS
Breslauer, Maamari, Switz, Lam, Fletcher
Authors

David N. Breslauer


Robi N. Maamari


Biophysics Grad Student
Wilbur A. Lam


Bioengineering Grad Student
Neil A. Switz


Bioengineering Grad Student
Assistant Professor of Pediatrics & Bioengineering
Daniel A. Fletcher

Professor of Bioengineering
Introduction: Initial Concerns about
Feasibility & Impact

“…we have built a mobile phone-mounted light
microscope and demonstrated its potential for clinical
use by imaging P. falciparum-infected and sickle red
blood cells in brightfield and M. tuberculosis-infected
sputum samples in fluorescence with LED excitation.”
Marli: Do camera phones have high enough resolution to
take light microscope pictures?
 Marli: Does anyone actually have tuberculosis anymore?
 Marli: I know LEDs are bright, but are they THAT BRIGHT?

Introduction: Advantages of This
Method

LEDs are inexpensive
Rob: Not sure how well the device would cut down
equipment costs; you still have to buy their device, plus the
relevant dyes. The training costs are an obvious savings,
however.
 Cost breakdown in a few slides!


Mobile phone networks are prevalent; useful for image
transmission
Marli: So like picture messages?
 Marli: This means they don't have to pay to fly out
specialists for a consult.
 Are phone data plans cheaper/faster than wifi in other
countries?

Introduction: Advantages of This
Method Cont’d

Mobile phones today almost always have cameras
 Erin:
That phone doesn't look very smart, I bet we could
do even better with an iPhone (8MP vs 3.2MP).
 Sunitha:
Not sure if there were iPhones in 2009 :-). But the
gadget looks neat - like a prize-winning science-fair
project...

Digital image processing available on phones
 Rob:
Hadn't thought about this advantage. Interesting!
 More on this later… but keep in mind, how much did
they actually process the images using phones vs
computers?
Brightfield Imaging of Malaria and
Sickle Cell Anemia

“…malaria can be effectively diagnosed from emailed smear images.”
 Sunitha:
This proves that in a hospital with minimal lab
facility (like a microscope and internet connection), the
same can be achieved without having to use the special
hardware with the cell phones. The cell-phone
technology described here is probably suitable for
places with absolutely no lab or internet facility.
 How common is it for a hospital to have no lab/internet
facility whatsoever? Is that the main scenario of
usefulness?
The Malaria Game


Marli: Wouldn't it be
SO much more difficult
with the thicker blood
smears?
Erin: How do you tell if
malaria is present in a
thick smear such as
this? I certainly can't!
The Malaria Game Cont’d


Erin: This is a lot better.
What do you think, class,
could we play the Malaria
game with this image?
Erin: Seems like their
“automated bacillus
counting of the fluorescent
TB images” could be useful
for that time-intensive cell
photo cropping we were
talking about for the
Malaria game.
Fluorescence Imaging of Tuberculosis
and Automated Image Analysis


Sunitha: This means there are two versions of the
hardware needed with the mobile phone: One for
fluorescence microscopy technique and another for
brightfield imaging technique. The former is
expensive than the latter.
Just how much more expensive? Let’s find out…
Materials and Methods
Material for Brightfield
Cost
20X wide field microscope eyepiece
$115
60X microscope objective
$70
Total $185
Additional Material for Fluorescence
Cost
Luxeon III 455nm LED
$5
3x3” microprocessor heat sink
$40
Silver conductive epoxy
$30 (per tube)
5˚ spot lens
$10
25.4mm focal length biconvex lens
$50
Excitation filter
$300
Emission interference filter
$80
Total $515
Traditional Methods Cost comparison
Method
Cost
Traditional fluorescence microscope
$1700
iPhone fluorescence microscope
$700
Savings $1000
Traditional 30x microscope
$40-130
iPhone 30x microscope
$185
Savings -$55 to -$145
Other Materials and Methods

Free water drop method
 http://blogs.scientificamerican.com/compound-
eye/2012/03/12/transform-your-iphone-into-amicroscope-just-add-water/

$5 mini-microscope method
 http://crabfuartworks.blogspot.com/2010/09/crabfu-
5-iphone-microscope-mod.html

$20 tiny lens method
 http://www.wired.com/wiredscience/2011/03/diy-
cellphone-microscope/?pid=1112&viewall=true
$20 Tiny Lens Method Mini-Lab
1.
2.
3.
4.
5.
Find some thin, dark, rubbery material and poke a
small hole in it (less than 1 millimeter in diameter). This
can be done using a pin or needle.
Carefully mount the lens to this “iris,” covering as little
of the lens as possible.
Center the iris with the ball lens tucked in the middle
over the camera of the cellphone (above).
Mount the iris to the phone with electrical tape. Our
rubbery material is sticky, so we don’t need to.
As with a standard microscope, use plenty of light to
illuminate your sample. Liquid samples should be
placed between a glass slide and coverslip.
But how useful is it?

Sunitha: I wonder where this technology could be
applicable. There are millions of under-equipped
hospitals throughout the world, but the question is:
who will they send the images to? Will a medical
professional sitting in a bigger hospital do the
diagnosis of the images? Who will pay them? May
be, there is an affiliation of all the under-equipped
hospitals with a more sophisticated medical facility.
I guess I am thinking out loud :)
Discussion - Viability

“Microscope-enabled mobile phones have the potential to
significantly contribute to the technology available for
global healthcare, particularly in the developing world and
rural areas where mobile phone infrastructure is already
ubiquitous but trained medical personnel, clinical laboratory
facilities, and clinical expertise are scarce.”


Sunitha: I guess we still need trained medical personnel to use the
equipment - especially to handle the blood smear, etc. I can
visualize the use of this equipment in rural hospitals where there
are one or two doctors covering a fairly big low income
population.
Erin: True. Technology like this might be very helpful for
community health initiatives, though (info on community health
initiatives in India)
Discussion – Viability Cont’d

“Combining the mobile phone microscopy system
with automated sample preparation systems…”
 Sunitha:
This seems to be part of an answer to my
previous question, but it is mentioned here for the first
time. Not sure what it really means and how they
automate the process so personnel not trained to
prepare the sample can manage it. Probably this is
part of the "future field study" they mentioned at the
end of this section.
 Something like this? That’s got to be expensive, though,
right?
Discussion – Post-processing Issues

Post-processing of images
 Sunitha:
This article doesn't give sufficient examples of
specifically what kinds of post-processing a mobile
phone can do! Wonder how one can do automated
particle counting on a mobile phone! It must be a super
advanced cell phone! Are they talking about special
apps running on smartphones?
 They said they used ImageJ on a desktop for simplicity,
could they also be using it because current photo
editing apps aren’t up to the task? How valid is this
concept, then?
Discussion – Acceptance

“Since we are developing a technology that makes the current and longstanding internationally accepted standards for disease screening in
developing countries more portable – rather than creating an entirely new
diagnostic assay – we anticipate that a relatively fast time to adoption by
clinicians and health workers may be possible.”

Erin: Would there be a difference between this method and the malaria game in
terms of quickness of acceptance, then? What do you guys think?


Sunitha: I guess there is a difference between the two, especially in the purpose each
of the two serves. This gadget is intended for getting the smear readings out to a
professional quicker, where as the game is for diagnosis. I guess a combination of the
two could be something quite new and possibly quick and cheap.
Rob: Once you get the product approved, initiate mass-production of the
microscope attachment, deal with the programming issues, and set up a network
of people who can do the diagnoses (you still need an expert (or at least a
sufficiently programmed computer) to analyze the images) Still promising.
Discussion – Proprietary Camera
Software; Privacy

Proprietary camera software leads to lack of
information on several fronts (e.g., sensor
integration time unknown)
 Erin:
Seems like they should look into this.
 Rob: This would seem like an area where either a more
open-source platform or manufacturer collaboration
would be useful to actually control these settings.
 Rob: Might some of these be proprietary?

Location-tagging patient data
 Rob:
Privacy concerns?
System Design & Characterization

“Both the brightfield and fluorescence instruments are
designed to work with a typical camera-enabled mobile
phone.”


Rob: Phones have a variety of shapes/sizes; are they going to try
a one-size fits all approach, or produce different models for
different phones?
“~50,000 hour lifetimes of LEDs make them particularly
suitable for use in portable systems and systems designed
for use in developing areas where replacement parts may
be unavailable or unaffordable.”

Erin: Always good to factor in maintenance costs. I have a fancy
newer HP printer that I never use because the ink is 4x as
expensive as my old printer's ink.
Brightfield Microscopy


Erin: Brightfield is
traditional microscopy
(a bright light shines
through the cell)
Sunitha: To understand
brightfield, the
following link helped
Fluorescence Microscopy



Erin: Fluorescence
method stains the cells
so they can be seen
glowing
Erin: Have been around
since at least the 1950's
Sunitha: I found a nice
article to understand the
details on fluorescent
imaging here
Sensor Integration Time

Sunitha: Not sure what this means. Is this the time
needed to focus before taking the picture?

Same thing as exposure time/shutter speed in photography
– the time during which a sensor acquires ambient light
Epi-Illumination Geometry

Rob: Microscope
geometries  trans-illuminated=light
source (here the LED)
behind the sample(as
in picture)
 epi-illuminated=light
source coming from
same direction as
viewer.
Full Width at Half Maximum

Rob: Full Width at Half
Maximum (FWHM) - a
frequently used heuristic.
Basically, if you have some
plotted distribution with peak
height H, you measure the
distance between where it drops
to H/2 on either side of the
peak. This provides a simple,
easy way to characterize the
distribution, without worrying
about noise, etc. You guys may
already know this; I'm not sure
how commonly used it is outside
the natural sciences, but given the
talk about statistical measures a
couple weeks back, I thought it
might be worth addressing.
Nominal Rayleigh Resolution Limit

Rob: Nominal Rayleigh
resolution limit - Basically,
assuming normal Rayleigh
diffraction, a light source will
give you a bright inner spot,
surrounded by a dark region,
then a bright ring, then
another dark region, etc. The
Rayleigh resolution limit is the
point at which the bright inner
spot from one object would
fall in the first dark ring from
the other object.