View
215
Download
1
Category
Preview:
Citation preview
Biosensors for Pathogen
DetectionAggiE Challenge Fall 2015
Dr. Kameoka
Jaskirat Batra
Agenda• Introduction
–Project Overview
–Group Responsibilities
•Group Progress
– Fabrication Team
– Optics Team
– Software Team
Project Overview• Objectives
– Create a microcantilever able to produce a measurable cantilever
deflection
– Initiate a clear image using a smart-phone
– Develop detection software from pictures from a smart-phone
• Requirements
– Affordable
– User-friendly
– Highly Sensitive
Group Assignments• Fabrication
– Create a polymer-based platform that can support the identification of a
biological sample in order to produce a deflection
• Optics
– Design the most sensitive optical configuration possible for a smart-
phone platform
• Software
– Develop a program to measure the deflection of the cantilever from the
smart-phone image
Fabrication Team
Meredith Davies [BMEN]
Rebecca Valencia [ECEN]
Last Semester’s Progress• Mold updates and parameterized files
• Fabricating cantilevers without silane coating
• Dyeing cantilevers to improve contrast
• Platform design
Overview• Fabrication process
• Sudan IV
• Finding the Young’s modulus
• Other Experiments of PDMS Properties
Fabrication Process
Soft Mold Fabrication Process1. Mix PDMS
2. Pour PDMS into hard mold
3. Cure at least 2 hrs @ 65C
4. Remove soft mold from hard mold
5. Cure soft mold only for at least 48 hrs @
125C
• They need to be remade every few
months for easier cantilever extraction
• Cantilevers were static and responsive to
our gloves and petridish
• Dipping in water usually helped but added
lint from paper towels
Example of soft mold with failed cantilevers
Cantilever Fabrication ProcessTwo-Tone Cantilever
1. Mix dye with PDMS
2. Pour PDMS into block
3. Cure 30 min @ 65° (or Cure 3~4 min @ 110°C)
4. Pour additional PDMS (may be different color) into cantilever channels
5. Use razor to smooth and rid excess PDMS
6. Cure at least 2 hours @ 65°C
7. Remove cantilever
Example of two colored cantilever with very
flexible cantilevers
Hard Molds
0.2 mm thickness
0.4 mm thickness
4 mm long8 mm long4, 6, 8 mm long
Clear Cantilever Problems
8mm 6mm 4mm
Modified Cantilevers: 4 mm long0.2 mm thick cantilevers
0.4 mm thick cantilevers
0.2 mm thick cantilevers
0.4 mm thick cantilevers
Concentration: 0.3 mL dye / 10 g PDMS
Sudan IV
Sudan IV Interest• Properties of PDMS
• Bleed less (on hands, into water)
• Better contrast image for processing
Sudan IV Solution
Instructions from Flinn Scientific
1. 37.5 mL of ethyl alcohol
2. Heated
3. Added 0.25 g of Sudan IV powder
4. Shake with vortex mixture
5. Add 12.5 mL of DI water0.3 mL Sudan/10 g PDMS
Sample created
Sudan IV• Cured at 65°C for 2 hours
– Samples reacted to heat
• Cured at room temperature for 48
hours
Finding the Young’s Modulus
Tensile SamplesThe gage length was pre measured and
marked on the sample and then inserted into
the headers
ASTM D412: Extension Rate is 500 mm/min
Used 12.7 mm/min for majority of the tests
Using increased extension rate (50 mm/min)
did not change the young’s modulus
In the process of being stretched
Clear PDMSSudan Dye
Food Dye
Food Dye Concentrations
No Dye 0.77 MPa
0.1 mL/10 g 0.39 MPa
0.3 mL/10 g 0.38 MPa
0.5 mL/10 g Not Found
• Young’s modulus did not appear to
change at different dye concentrations
• Higher concentrations are harder to
remove
Other Experiments of PDMS
Properties
Food Dye Water Soluble PropertyGoal
Test how much dye in the PDMS is
leaking
Experiment
Placed samples in water for 24 hrs and
then in the oven (65C) for 24 hrs repeatedly
Alternated environment for 6 days
Recorded the mass
Findings
Observed that the dyed PDMS leaked dye
into the water every time it was introduced
into fresh water
The change in mass was not found to be
statistically significant in relation to the
scale’s accuracy
Heat Strain
Heat Strain• Test if changing the temperature is changing the diffusion
rate of dye out of PDMS
• Control: Plain PDMS in green dye
• Experiment: Heating samples of clear PDMS then adding
into dye alternating for 6 days
Heat Strain Results• 6 days alternated heat/dye
• Tears on the surface
• Some green dye on outside
Food Dye vs SudanFood Dye
Cured at higher temperatures, faster cure
time
Water soluble, dye leaks out of PDMS
Contains corn syrup, at higher concentrations
the PDMS is sticky
Sudan IV
Larger Young’s modulus, less flexible
Class 3 carcinogen
Mutagenic for mammalian somatic cells
Next Semester’s Steps• Sudan IV tensile test
• Sudan IV in water
• Indicate the best dye option
• Fabricate successful cantilevers for imaging
• Modify detection platform as needed
Questions?
Optics Team
Nirmal Patel [ECEN]
Joe Slagel [ISEN]
Calvin Theriot [ECEN]
Images Without Optics
• Limited to 5x with Digital Zoom
• Very Lossy - High Noise
• Impractical Focal Length
• Low Reproducibility
Unutilized Pixels
Goals
• Review optics principles
• Find multiple potential lens which would
provide a high-resolution picture.
• Select the best lens among all potentials.
Preliminary
Before any research could be done, it was
important for us to review the following optics
principles. Reflection and Refraction• Images
• Mirrors
• Snell’s Law
• Lenses
Finding the right lens3 lenses were chosen:
a) Neewer 60X Zoom LED Clip-On Microscope Magnifier
Micro Lens with Universal Clamp.
b) TECHNO Universal Professional HD Camera Lens Kit
(12.5X Super Macro was used).
c) Also a lens provided by Dr. Kameoka “Gakken Suteifuru
Microscope” was considered.
Option A: Neewer 60X Zoom LED Clip-On Microscope Magnifier Micro
Lens with Universal Clamp.
● Advantages:
○ Very large zoom.
○ LED light attachment.
● Disadvantages:
○ Large amount of fish-eye.
○ Inability to take picture without
touching object.
○ Poor holder design. Testing the Neewer lens
using currency
Kit contained 3 lens: In our research project the 12.5X
Super Macro Lens was used
12.5X Super Macro Lens details:
• 12.5X zoom
• focal length = 23.54 mm
• d = 10 mm
• f/2.354
Source: Amazon.com
Option B: TECHNO Universal Professional HD Camera Lens Kit (12.5X
Super Macro was used)
Results: Galaxy S6 Stock Lens
• Taken using Galaxy S6
stock lens and 5X digital
zoom
• f/1.9
• Significant amount of
noise
• Distance of image: ~3.11”
but could vary.
Demonstration
of Clarity
Results: Galaxy S6 stock lens + 12.5X lens
• Taken using a Samsung
Galaxy s6 (no digital zoom)
and added 12.5X Super
Macro lens
• f/2.354
• Reduced noise
• distance of image = 0.75”
Demonstration
of Clarity
• Image taken with lens
• Less noise
• Long depth of field
● Picture without lens
● Significantly more
noise
● Shallow depth of field
Gakken Microscope
• Three optical powers
• 2x, 4x, 16x
• Easy integration with
existing systems
• Measurements taken
with 8 MP sensor
(image is 66% of a 16
MP image in y-axis)
Note: The Light Source is not included in this kit.
Phone Platform
Adjustment Wheel
Lens Housing
Stage
Light Source
Technical SpecificationsLens 2x 4x 16x
Focal Length 2.778 cm 1.27 cm 0.1588 cm
f/number* 2.692 1.455 0.333
Numerical
Aperture0.82 0.485 0.0378
Depth of Field 0.515 cm 0.23 cm 0.25 cm
Minimum
Theoretical
Detection**
58.8 µm 25 µm 5.2 µm
* The f/number without a lens is 2.4
**The minimum theoretical detection without a lens is 210 µm
MethodologyIn order to find minimum theoretical detection:
1. Measure Pixel Height
2. Measure Pixel Buffer
3. Measure Real Height
4. Calculate Buffer Height
2x Lens
No Magnification 2x External Magnification
4x Lens
4x External MagnificationNo Magnification
16x Lens
No Magnification 16x External Magnification
Miscellaneous
2x Magnification + Digital Zoom
Dark Cantilever with No Deflection
2x Magnification + Digital Zoom
0.2mm Cantilever with Deflection
Decision MatrixLenses Clarity No
DistortionNo
Background Noise
Ease of Use
Detail No Fisheye
Focus Practicality Deflection Picture Area
Total
Gakken Lens
System 2x
8 6 6 9 6 7 8 7 8 9 74
Gakken Lens
System 4x
8 4 7 9 8 6 6 5 7 7 67
Gakken Lens
System 16x
8 1 9 9 9 8 4 1 7 4 60
Macro Lens
7 8 7 4 4 8 7 3 5 10 63
60x Lens 5 1 1 2 4 2 3 1 3 1 23
No Lens 2 3 2 9 2 10 6 3 4 10 51
Conclusion• 2X and 4X Gakken lens ranked highest
according to decision matrix.
• <10X lens with focal length ~1cm gives highest
precision.
• 16x Gakken Lens and 60X Neewer lens
disqualified due to practicality
Future Goals• Reproduce high resolution pictures of cantilevers.
• Research and recommend low-distortion lens.
–Preferably 16x
• Improve phone holder design.
Software Team
Lucas Cheung [ECEN]
Charlie Zhang [ECEN]
Stefan Manoharan [ECEN]
Last Semester’s Work
• Edge Detection Software–Greyscale
–Binary
–Edge Detection
– Invert Edge Detection
–Tabulate in Matrix
– Invert Matrix
–Shift
Pros
• Provides very good basis for further code
• Show a rough edge to measure
displacement
Cons• Not applicable for different types and angles
• Image provided did not show any displacement
• Small amount of noise
New Considerations
• New Images–Colored Cantilevers
–Lens
–Lighting
–New Angle
Goals
• Make code applicable to multiple situations
• Remove as much noise as possible– Image
–Matrix
Displacement Test• If able to detect rough displacement, then able
to calculate displacement - pixel scale
• Crop area of interest to reduce excess
information
• Different edge detection method
Crop Results• 1mm Cantilever
– ~5 pixel displacement
Crop Results (cont.)• 0.6mm cantilever
– ~20 pixel displacement
Threshold OptimizationProgramming Team
Overview of Fixes by Mid-Term
• Binary Detection Mechanism is more
calibrated (Used background as threshold
source)
• Edge Detection Source was changed from
Binary to Greyscale
• Better Edge Detection Algorithm was
proposed (Roberts)
Where We Left Off
Problem
Definition of ‘Salt and Pepper’, anyone?
• The salt and pepper in the image looks ugly.
• Edge detection algorithms were ineffective with this
much noise.
• Without a clear edge, image can’t be used
Possible Reasons• Threshold,
though better,
was not good
enough.
• Too much noise
in peripheral of
cantilever
Possible Reasons
• Slight changes in light intensity and color
in a greyscale image would be interpreted
as an edge according to the algorithm.
Solution
• We reintegrated binary to the process (better contrast
between two sides of an edge)
• Manual crop function was added to get rid of extraneous
peripherals.
• Threshold Selection: Three possible areas to look out for
thresholds
The New Cantilever Image
Threshold Selection Areas
• Inside the colored Cantilever tip
Threshold Selection Areas
• Containing the Cantilever-Background Edge
Threshold Selection Areas-
Background of Cantilever!
• Background was the most interesting
selection to play around with
• We found out that the amount and position of
salt and pepper depended on the dimensions
and position of threshold rectangle.
Threshold Selection- Background
• Selecting a ‘fat’ rectangle
Threshold Selection- Background
• Selecting a ‘vertical’ rectangle
Threshold Selection- Background• Selecting a Horizontal Image
• Found that Thinner rectangle is better (thinner salt and
pepper curve)
Correlations Between Rectangle
and Salt & PepperThere is possibly a radial light gradient
from the bottom centre, causing all
isoreflective areas to interact with the edge
detection in the same way.
Demonstration
• Matlab Code Module: Bin_Conv.m
• Must have in Path: Threshold_Test.jpg
Results
• Eliminated (for the most part) salt and
pepper from the necessary parts of the
figure
• Attained a clear, single line detection of
edge
OpenCV + Android
We eventually want our
software on mobile devices.
• Camera + Computer
• Portable
• They’re everywhere
Motivation
• OpenCV–Most popular open source image
processing library
• Android–More cameras to choose from
–Cheaper to iterate
Why OpenCV? Why Android?
Image processing basics
• I/O
• Matrix conversions and operations
Edge detection implementations
• Canny
• Sobel
OpenCV Features
Outdated documentation on OpenCV
makes using it difficult
Learning Android
Porting to Android
Android Demonstration
Questions?
Acknowledgments Thank you to the following for aiding our research:
• AggiE Challenge Program
• Magda Lagoudas
• Dr. Jun Kameoka
• Dr. Lei Fang
• Jaskirat Singh Batra
Recommended