Thomas Cajgfinger, Eric Chabanat, Agnes Dominjon, Quang T. Doan, Cyrille Guerin, Julien Houles, Remi Barbier.

Size: px

Start display at page:

Download "Thomas Cajgfinger, Eric Chabanat, Agnes Dominjon, Quang T. Doan, Cyrille Guerin, Julien Houles, Remi Barbier."

Reginald Jordan
5 years ago
Views:

1 Thomas Cajgfinger, Eric Chabanat, Agnes Dominjon, Quang T. Doan, Cyrille Guerin, Julien Houles, Remi Barbier. IPNL, Université de Lyon, Université Lyon 1, CNRS/IN2P3, 4 rue E. Fermi Villeurbanne cedex, France

$objective diffraction limit resolution Wish List of imaging sensors: 10 nm resolution on position of targets (=µm on sensor) Fast frame rate ~ khz Photon counting ~ 1-10 Multi target tracking ~ 1000$

2 Fluorescence microscopy : 2D imaging system Population of nanometer scale single emitters: Static or dynamic fluorescent beads (protein, Quantum Dots ) Phototoxicity->low signal(~photons/ms) Below objective diffraction limit resolution Wish List of imaging sensors: 10 nm resolution on position of targets (=µm on sensor) Fast frame rate ~ khz Photon counting ~ 1-10 Multi target tracking ~ 1000 Does that type of camera exist?

3 Single photon sensitive detector: Hybrid detector : electro bombarded CMOS CMOS + photocathode + vacuum tube Gain = accelerated e- by electric field in vacuum tube Point spread function(psf) : Tube: radial velocity of emitted e- CMOS : thermal diffusion ebcmos detector ebcmos Camera ebcmos Principle ebcmos PSF ebcmos DAQ

4 Continuous acquisition Frame rates: 125, 250 & 500 fps FPGA DDR custom board Ethernet 1 Gb/s Next 10 Gb/s

Secondary e- diffusion & charge sharing Impact pattern Photo-electron

intra pixel localization Natural CMOS noise filtering Counting

<80nm Epitaxial layer <10µm Secondary electrons Diode Read out integrated

5 Secondary e- diffusion & charge sharing Impact pattern Photo-electron reconstruction by clustering Computation of centre of gravity (COG) -> intra pixel localization Natural CMOS noise filtering Counting possibilities with gain linearity Accelerated Photo-electron Dead Layer <80nm Epitaxial layer <10µm Secondary electrons Diode Read out integrated circuit Diffusion of secondary e- in the CMOS Photoelectron Event COG reconstruction Resolution ~ 2 µm

6 Zoom in a raw frame Zoom in a reconstructed frame Frame 1 2µm Frame 2 Frame 5 Frame 10 Image stack & Kalman filtering Reconstructed Target Photoelectron State vector Measurement vector Zoom (~4x4pixels) What resolution can be achieved with this method?

Time (ms) ~mm/s 1000 1000 Comparable resolution EMCCD/ebCMOS QE 20% with

7 Localization accuracy [µm] <300 photons 300ph α1/ N Number of emitted photons Relative error On sensor position [µm] Resolution ~ 10µm Time (ms) Time (ms) ~mm/s Comparable resolution EMCCD/ebCMOS QE 20% with faster frame rate Adaptive tool (signal, background, motion ) Promising results

56 photon/2ms mean received signal Example : 50 frames period 400 measured positions Compared to statistical

8 Neutral density filters Spot <1µm Triggered LED (640nm) Optical fiber (8 µm core) Microscope objective 1/50 Integrating sphere Sketch : focusing a spot (Ø< 1µm) with various background noise conditions 1.56 photon/2ms mean received signal Example : 50 frames period 400 measured positions Compared to statistical true position ( frames) ~78 photons resolution:1.2µm ebcmos camera system Y Axis [µm] ebcmos resolution in the noise free case

9 Resolution on sensor [µm] 2µm α1/ N Signal [ph] 80 Localization accuracy: <3µm after 30 photons <2µm in noise free case Finding targets: Fake rate <10% & efficiency>90% after 30ms 90% 10% Time (ms) 30 Efficiency Fake rate

10 <1ph/2ms = <1ph/frame SUSS MicroOptics One raw frame 684 spots 1 Signal /2ms/target Mcrolenses Spot Mean Signal Targets after ~ 20 frames Average signal (3500 frames)

11 Setup at Nanoptec Center in Lyon: Spin-coated QDs Wide field microscopy setup Magnification: 100X Quantum Dots ID: Emission wavelength: 605 nm (Invitrogen) Excitation wavelength: 473 nm QD size: nm Detected targets after ~ 20 frames Average image ( frames) Parallel processing and data extraction

Temporal signal tracking: 30 frame average signal Blinking Quantum Dots: ON and OFF states Blinking ON OFF ON Mean signal [ph] OFF Frame number Temporal Tracking of a Blinking QD Parallel

12 Temporal signal tracking: 30 frame average signal Blinking Quantum Dots: ON and OFF states Blinking ON OFF ON Mean signal [ph] OFF Frame number Temporal Tracking of a Blinking QD Parallel tracking of 300 nano-photo-emitters : Position (localization accuracy ~ µm on sensor) Signal (photon counting) High data throughput No time limit 3ph 7ph Signal [ph] Signal Distribution of a Blinking QD

13 Camera system improvements: Larger and faster CMOS Target tracking with parallel processing : FPGA & GPU computing Real time implementation Applications: Physical properties of nano-objects: mean square displacement, viscosity Photo Activated Localization Microscopy (PALM)

14 IPNL: L Vagneron, D Chaize, P Calabria, W Tromeur, P Depasse IPHC Strasbourg: Jerome Baudot, Andrei Dorokhov, W Dulinski, M Winter PHOTONIS SA : C T Kaiser University of Lyon, Centre NanOptec: David Amans, Christophe Dujardin, Gilles Ledoux

17 3T

Lusipher 400x400 Continuous readout 640 Mbits/s 500 fps Hardware configuration Dell Precision T7400 2 QuadCore E5430 2.

thread for socket reception 20 % of a core 1 thread for frame reconstruction and SNR computation 50 % of a core 2 threads for clustering and ion feedback filtering 100 % of a core 1 thread

18 Lusipher 400x400 Continuous readout 640 Mbits/s 500 fps Hardware configuration Dell Precision T QuadCore E GHz 8 GB of RAM 4 disks 500 GB RAID0 for data storage 1 disk 500 GB for the system 1 dual port Gb Eth Board CPU load measurement for 400x400 pixels at 500 Hz frame rate = 640 Mbits/s 1 thread for socket reception 20 % of a core 1 thread for frame reconstruction and SNR computation 50 % of a core 2 threads for clustering and ion feedback filtering 100 % of a core 1 thread for the display slow motion 50 % of a core 1 thread for the display low frequency (Histogram) 3 % of a core 1 thread for the continuous data UDP packets storage 20 % total = 2,5 cores/8 cores 30% of the total CPU without real time deconvolution or tracking of the emitters

Single-photon sensitive fast ebcmos camera system for multiple-target tracking of single fluorophores: application to nano-biophotonics.

Single-photon sensitive fast ebcmos camera system for multiple-target tracking of single fluorophores: application to nano-biophotonics. Thomas Cajgfinger a Eric Chabanat a Agnes Dominjon a Quang T. Doan