Mobile Microscopy Group #33 Rui Guo, Yongli Chen, Xiaoyu Qin
Overview Incorporated LED circuit into QPI(Quantitative Phase Imaging) Design An migrated implementation of phase extraction program on iOS devices
Objectives Want to process and capture image on mobile devices Process different kinds of images and analyze the performance Incorporate PCB in the microscopy design
Why phase image? Provide quantitative information about biological specimens(i.e. dry mass) Phase of a field is much more sensitive to the specimen structure than its amplitude
Why DPM(Diffraction Phase Microscopy)? Noise in phase due to mechanical vibrations and air fluctuations Compact configuration that cancels out most mechanisms responsible for noise acquisition speed is limited only by the speed of the camera employed
System Overview Hardware: 1. DPM system 2. Microscope 3. LED circuit 4. iPhone 5 Software: 1. Built in image capture session 2. Process multiple images at one time (up to 100 images) 3. Scroll view for user to view the preprocess images or phase image results
Hardware Overview LED circuit Provide illumination to the microscope DPM system Active filter used in the microscopy Microscope Zeiss Z1 observer with adjustable magnification up to 40x iPhone 5 iOS device
System Overview Block Diagram LED Light Source Microscope Port DPM System CCD & ProjectoriPhone camera Data Collection & Preprocess
Setup
DPM System Block Diagram Microscope Output Image Iris Diffraction Grating 4f plane with Pinhole filter Final Image Output to CCD
DPM Schematics
DPM System (our implementation) Diffraction grating First lens 40mm focal length Pinhole filter Second lens 100mm focal length
Diffraction Grating Starting from the normal incidence condition, rotate the rotation stage until the diffraction order = −1 propagates straight back to the incident beam. The diffracted beam goes back to the laser cavity (or close to it). Read this angle on the rotation stage and subtract the offset angle.
Verification on focal length of the lens #1 Method 532 nm laser diffuser Collimated lens iris Testing lens Image plane
Verification on focal length of the lens #2 Method
Pinhole Filter 1 st order wave is filtered down using a small pinhole 0 th order wave is fully passed Two fields interfere at the final image plane to create the interferogram We tested pinhole by adjusting its position in x, y, z direction. From the CCD camera, we can see how much light passed through the filter. Cross section of pinhole filter
Verification of DPM system Adjust the DPM system until three lobes show up in the Fourier domain Middle main lobe – DC – zeroth order Upper and lower lobes – 1 st order
Circuit Overview LED light source utilizing Printed Circuit Board The circuit provides illumination to the microscope, so that the detector will be able to generate pictures with enough brightness. First prototype was built with bread board. Final product was built with printed circuit board. There are three major parts: Arduino Uno microcontroller, switches, and LEDs.
Circuit schematic switches Arduino Uno LED module
LED module consists of: 1.9 x White LEDs 2.3 x Bipolar Junction Transistors 3.9 x 100Ω Resistors at collectors 4.3 x 1k Ω Resistors at bases 5.1 x Power supply from Arduino 6.1 x Ground from Arduino
Switch module
Arduino Uno microcontroller Arduino Uno microcontroller serves as power supply and ground for the whole circuit. The microcontroller also controls intensity of LEDs through switches, as described in the previous slide.
First prototype
Printed Circuit Board
Verification of light source circuit We made very careful adjustment to the position of white LED. The LED needs to sit on the focus of collimated lens. Through observing the output from CCD camera, we found an optimal position for the LED.
Software Overview Utilized iOS OpenCV Framework: Computation efficiency Quick functions to use(i.e. dft, normalize, magnitude) Created bridge between C++ and objective C Image Format Conversion: UIImage (oc) to Mat(c++). Phase extraction written in c++ Event and outlet design & link written in objective c Basic steps to extract a phase image
Solved challenges and bug fixes Xcode built in image picker can only select one image. Solved by writing own image picker which can select and save multiple images into an array Fourier spectrum is not centered because OpenCV is lack of shifting algorithm Solved by designing a function which cut matrix into pieces and form a centered spectrum after reorganization. Supporting different format of images Created an image conversion function that can support most common 16 bit image format (i.e. jpeg, tiff, png, bmp)
Major obstacles in software Information lost due to image format (i.e. JPEG vs TIFF) fftshift algorithm doesn’t work as fast as in MATLAB Runtime limited by the computation speed of the phone Maximum images to process because of limitation in memory
Simulation result on iPhone Left: red blood cell Right: phase image
More Simulation Results Left: hand imposed on a grating Right: Phase image of hand
Conclusion We could encapsulate the whole system so that the DPM will have more mobility All the test results are within the range of the requirement As a result, we can apply DPM system to other microscopes Overall design works properly Successful migration to mobile device
Further Development Optimize the application, do the video processing instead of image sequence Try with powerful white LED so that we can ideally build a portable microscope Take various images of different objects and discover the potential application of the software