This document describes a method for active 3D drift control in super-resolution fluorescent microscopy. Super-resolution microscopy requires capturing many images of sparse fluorescent subgroups and reconstructing a high-resolution image, but 3D drift can cause previously sparse subgroups to overlap, reducing localization capabilities. The method uses a proportional-integral-differential controller to provide closed-loop feedback stage control based on image analysis, moving the sample stage to compensate for estimated drift in real-time and allow continuous super-resolution image acquisition. Experimental results show the system can correct for drift on the order of nanometers, enabling visualization of structures less than 200nm apart.
1. Active 3D drift control for a super-
resolution fluorescent microscope
Patrick Llull
Advisor: Rafael Piestun
University of Colorado at Boulder
Summer 2011
2. Microscopy: pushing for scientific discoveries
• Primary targets: cellular structures, small
organisms, chemical reactions
3. Imaging resolution problem
• Response to point source: point spread
function (PSF) governed by optical system
geometry
http://www.scanco.ch/support
Resolved PSFs Unresolved PSFs
4. Super-resolution method: STORM
STochastic Optical Reconstruction
Microscopy
• Fluorescent molecular subgroups
• Series of point images, subgroups of PSFs analyzed
• Stochastic reconstruction: stack of images, localization
–How can we achieve 3D localization?
http://www.scie
ncedirect.com
5. Double-Helix PSF
• PSF that rotates with defocus
– Encodes defocus information with angle of orientation
Double-Helix PSF (fluorescent
bead) with defocus
Camera
Engineered
Optical
Element
Sample
mounted on
control stage
Standard PSF with defocus
6. Problem: 3D drift and vibrations
• Reduced localization capabilities
– Ambient conditions
• Potential overlapping of previously sparse PSFs
Acquired image before drift After drift
7. Drift correction block diagram
• Use of Proportional Integral Differential (PID) control method
• Camera image acquisition, PSF tracker VI, stage movement,
Image
acquisition
(camera)
Image
analysis
(estimate
drift)
Compensate
for drift
(move
sample
stage)
http://www.thorlabs.com
8. Correcting drift: active PID stage control
• Proportional Integral Differential (PID)
– Closed loop feedback control to move microscope stage
– Present, past, and future error estimation
http://zone.ni.com/devzone/cda/tut/p/id/3782
9. Determining drift: image location estimation
• LabVIEW machine vision matches template twice to
an image’s region of interest
– Templates: experimental and ideal Gaussian intensity
distribution
Fluorescent molecule
Experimental lobe template
Ideal lobe template
Angle to
defocus
calibration
11. Calibration: PSF Tracker VI systematic error
• Drift test through twenty 100 defocus steps: 7.5nm
transverse; 30nm axial
Transverse (x,y) movement during readout
15. Acknowledgements
• Piestun research group (Rafael Piestun, Ginni
Grover, Ariel Libertun, Sean Quirin, Anurag
Agrawal, Jerry Brown, Antonio Caravaca, Don
Conkey, Anthony Barsic, Albert Brown.
• SMART Program, University of Colorado