from ccd to emccd scientific imaging for today’s microscopy
TRANSCRIPT
From CCD to EMCCDScientific imaging for today’s microscopy
From CCD to EMCCD
Is this a right topic?
Who should care what?
The CCD manufactories: Sony, Kodak, Texas Instruments, e2V ….
2. Cameras: Read noise, cooling, interface, speed, sensitivities …..
MAG !!
3. Imaging Systems:
1. CCD, CMOS, EMCCD, Interline, Color, Frame Transfer……
System Integrators, LIN Trading !!
From CCD to EMCCD
Is this a right topic?
What actually do you/researchers care about?
Beautiful image!!!
Publication and
Quantization !!!
From CCD to EMCCD
The right topic
What make a good image?
• better resolution
• higher Signal-Noise Ratio
• Good Contrast (Dynamic Range)
• Resolution (Digital Resolution, Spatial Resolution)
• Signal-Noise Ratio
• Contrast / Dynamic Range
From CCD to EMCCD
Revised topic: important characters of digital image
electronics
CCD / CMOS
image sensor
Camera (CCD?)
• Resolution (Digital Resolution, Spatial Resolution)
• Signal-Noise Ratio
• Contrast (Dynamic Range)
From CCD to EMCCD
Revised topic: important characters of digital image
From CCD to EMCCD
Digital Resolution
Approach:
1. Bigger Chip;
High cost for high grade chips
2. Smaller Pixels;
Lower sensitivity (Signal/Noise Ratio)
3. Micro Scanning
Good Balance of the above, slow speed.
From CCD to EMCCD
Digital Resolution
Useful resolution for microscopy
Camera Resolving Power > Optical Resolving Power
1. Specimen details resolved by the objectives, need to be acquired by the camera ;2. Avoid “empty resolution“, empty resolution only create unnecessary large files.
Tip to be remembered:Camera resolution should match optical resolution;Low magnifications normally require higher camera resolutions.
From CCD to EMCCD
Digital Resolution : The higher, the better ?
25 mm
2/3 “ Chip
Adaptation 1,0 x 18, 3% of view area 20
Adaptation 0,5 x , 58% of view area 20
sensor size 8,8 mm x 6,6 mm
8.8mm6
.6m
m
1.22 x N.A.Objective + N.A.Condenser
d0 =
Field of View = Field Number/mag.
Requested resolution = 2* FOV / d0
Requested Resolution
The higher, the better?
1435 x 108184714 x 53842 1,4100
921 x 69354459 x 34627 0,9100
2270 x 17091341139 x 85867 1,463
3320 x 24991951666 x 1254981,340
1906 x 1435112952 x 717560,7540
3255 x 24511911632 x 1229960,8025
3830 x 28862251921 x 14461130,7520
2548 x 19181501275 x 960 750,520
5097 x 38373002550 x 19201500,510
2548 x 19181501275 x 960750,2510
3063 x 23051801530 x 1152900,155
4905 x 36932882448 x 18431440,122,5
3210 x 24511921632 x 1229960,041,25
Necessary camera
resolution
Lines/mm(TV- 0,5 x)
Necessary camera
resolution
Lines/mm(TV-1.0 x)
N.A.Magnificati
on
Nyquist Theorem: Sampling frequency should be double the frequency of the signal.
From CCD to EMCCD
Digital Resolution: the higher, the better?
1. GREEN resolution/QE ->50%;2. RED resolution/QE -> 25%; 3. BLUE resolution/QE -> 25%; 4. The color interpolation decreases camera resolution;
4.2 Pixels
From CCD to EMCCD
Digital Resolution: Color or Mono?
1 Pixels
Exposure Time: Color 3.34 ms VS. Mono 0.9 ms
From CCD to EMCCD
Digital Resolution: Color or Mono?
• Resolution (Digital Resolution, Spatial Resolution)
• Signal-Noise Ratio
• Contrast (Dynamic Range)
From CCD to EMCCD
Revised topic: important characters of digital image
Important Camera Specs affect Signal-Noise Ratio
1. Quantum Efficiency: higher signal
2. Noise: Photon noise, readout noise, dark current
3. Signal-Noise Ratio, Camera sensitivity
From CCD to EMCCD
Signal-Noise Ratio: Low light considerations
Quantum Efficiency : The Spectral Response / Photon to Electron converting efficiency
From CCD to EMCCD
Signal-Noise Ratio: Quantum Efficiency
Front vs Backside Illuminated CCD
From CCD to EMCCD
Signal-Noise Ratio: Quantum Efficiency
– Photon-induced shot noise– Readout noise– Dark current noise
• Total System Noise = all noise sources added in quadrature
Main Noise Sources in CCDs
From CCD to EMCCD
Signal-Noise Ratio: Noise
Photon Noise (Shot Noise)
- Law of physics- Square root relationship between signal and noise
noise = square root of number of electrons- Poisson distribution- When photon noise exceeds system noise, data is photon (shot) noise limited
- Law of physics- Square root relationship between signal and noise
Photon noise = √Signal electrons
- Poisson distribution- When photon noise exceeds system noise, image data is photon (shot) noise limited
From CCD to EMCCD
Signal-Noise Ratio: Noise
Serial Register
Preamplifier
Output NodeActive Array
ADC
From CCD to EMCCD
Signal-Noise Ratio: Noise
CCD Readout
Read Noise (preamplifier noise)
- Higher readout speed leads to higher Read Noise; example: Readout speed = 1 MHZ, Readout Noise = 3 e; -> 0.5 frame/second Readout speed = 20 MHZ, Readout Noise = 8 e; -> 10 frames/second
- Minimized by careful electronic design;
- Under low-light/low-signal conditions where read noise exceeds photon noise, data is read noise limited
- Read noise not as relevant in high-signal applications
From CCD to EMCCD
Signal-Noise Ratio: Noise
Dark current: • Electrons created by thermal emission;• Increases with time and temperature;• Cooling CCD reduces Dark Current;
Dark current is cut in half as the CCD temperature drops approximately every 6.7° C• Reduced by utilizing multi-pinned-phase (MPP) technology
Rule: 6~7 degree doubling
From CCD to EMCCD
Signal-Noise Ratio: Noise
Tip: Readout Noise is the major equipmental noise contributor for a cooled camera!
Total equipment noise=√readout noise2+dark noise2
We use a typical readout noise = 8e, Dark noise =√total dark current =√dark current x exposure time
camera A cooled 25°C lower than ambient, dark current = 0.15e/p/scamera B cooled to -25°C, dark current = 0.015e/p/s
With exposure 30s, Total noise of camera A = 8.27e Total noise of camera B = 8.02e
With exposure 1mins, Total noise of camera A = 8.54e Total noise of camera A = 8.06e
From CCD to EMCCD
Signal-Noise Ratio: the cooler, the better?
Ultimately, a High-Performance CCD camera is limited only by Readout Noise and Photon Noise.
– Photon Noise - A law of physics!
– Readout Noise - Reduced by careful electronics design
– Dark Current Noise - Reduced by cooling and MPP
Noise Reduction in CCD
From CCD to EMCCD
Signal-Noise Ratio
From CCD to EMCCD
Signal-Noise Ratio: The final Equation
Signal-to-Noise Ratio of an Image = Total Photon collected / Noise1. Total Photon Collected where P=total incident photons,
DQE = QE at specific wavelength
2. Shot Noise
From CCD to EMCCD
Signal-Noise Ratio: The final Equation
Sensitivity of a camera: the lowest signal can be differentiated from background noise by the camera
Read Noise limited region
Photon Noise limited region
• Resolution (Digital Resolution, Spatial Resolution)
• Signal-Noise Ratio
• Dynamic Range (Contrast)
From CCD to EMCCD
Revised topic: important characters of digital image
Well capacity (Well depth):Number of electrons can be hold by a pixel before saturation
Well Capacity will be higher when pixel size is bigger:
• Same resolution, larger chip size;
• Same chip size, lower resolution;
• Binning
Note: If the charge capacity is exceeded,the excess charge will overflow into
adjacent pixels and produce artifacts known as blooming and smear.
NoiseOverflowing and Blooming
Charging
From CCD to EMCCD
Revised topic: important characters of digital image
Dynamic Range
Dynamic Range = Well capacity / Read noiseDynamic Range (dB) = 20 x Log10 (Well capacity /Read noise)
Tip: if your sample contains both very dark and very bright signals, a higher dynamic range camera is needed to imaging them in one shot!
From CCD to EMCCD
Revised topic: Dynamic Range
Dynamic Range of CCD should be matched to A/D Converter. 12, 14, 16 bit
Binning
- Higher Dynamic Range- Higher Signal-to-Noise Ratio- Faster Readout- Dynamically Change Pixel Size/Aspect Ratio
From CCD to EMCCD
Revised topic: Dynamic Range
Slower readout -> Lower noise
From CCD to EMCCD Low light – slower readout or longer exposure
Longer exposure -> Stronger
Signal
same exposure same readout speed
short exposure less photon collected
high readout high noise
From CCD to EMCCD
Low light & High Speed -> Short exposure + High Readout ??!!
Signal-to-Noise Ratio (SNR) = Total Photon collected / Noise
• EMCCD: Electron Multiplying Charge Coupled Device
• Operates by applying high voltage during readout before the preamp stage of the CCD. Occurs through a probabilistic phenomenon where the gain is determined by:
Gain = (1 + g)N
where g is the probability of creating a second electron (typically in the vicinity of 0.01 – 0.016) and N is the number of elements (usually 500+)
From CCD to EMCCD Low light & High Speed -> The EMCCD Technology !
Signal-to-Noise Ratio (SNR) = Total Photon collected / Noise
Frame Transfer CCD
Serial Register
Preamplifier
Output NodeActive Array
Frame Transfer CCD
Serial Register
Preamplifier
Output NodeActive Array
Frame Transfer CCD
Serial Register
Preamplifier
Output NodeActive Array
Frame Transfer CCD
Serial Register
Preamplifier
Output NodeActive Array
Frame Transfer CCD
Serial Register
Preamplifier
Output NodeActive Array
ADC
ReadoutSignal = 1
Note: if read noise is 1 then S/N = 1/1
Frame Transfer EMCCD
Active Array
Output Node
Frame Transfer EMCCD
Serial Register
Active Array
Preamplifier
Output Node
EM Register
Frame Transfer EMCCD
Serial Register
Active Array
Preamplifier
Output Node
EM Register
Frame Transfer EMCCD
Serial Register
Active Array
Preamplifier
Output Node
EM Register
Frame Transfer EMCCD
Serial Register
Active Array
Preamplifier
Output Node
EM Register
Frame Transfer EMCCD
Serial Register
Active Array
Preamplifier
Output Node
EM Register
Frame Transfer EMCCD
Serial Register
Active Array
Preamplifier
Output Node
EM Register
Frame Transfer EMCCD
Serial Register Preamplifier
Output Node
Active Array
EM Register
Frame Transfer EMCCD
Serial Register
Active Array
Preamplifier
Output Node
EM Register
Frame Transfer EMCCD
Serial Register Preamplifier
Output Node
Active Array
ADC
ReadoutSignal = 5
Note: if read noise is 1 then S/N = 5/1! Vast improvement
EM Register
On-Chip Multiplication Gain CCD SNR:
SNR=[S*QE]÷√[S*QE*F2 + D*F2 +(σR/G)2]
Note: F is the excess noise factor.
From CCD to EMCCD SNR: The new equation
From CCD to EMCCD Types of Noise in EM Cameras
• Dark Current– Dependent on exposure time– Increases when gain is increased -> cooling important
• Read Noise– Changes with readout speed
• Spurious Noise (aka clock induced charge)– Not dependent on exposure time– Lower cooling increases chance of spurious charge– Occurs during high pulse clocking of CCD and generates a
secondary electron, even though no primary is present– Usually combined with the overall dark charge
• Excess Noise Factor– Based on deviation or uncertainty in on-chip multiplication gain
EM camera Applications
• Total internal reflection fluorescence (TIRF) microscopy
• Spinning-disk confocal microscopy
• Dynamic ratio imaging (e.g., pH and low-concentration flux)
• Fluorescence recovery after photo bleaching (FRAP)
• Live-cell fluorescent protein imaging
Very high sensitivity up to single molecular detection!
From CCD to EMCCD When to use EMCCD?
Signal-to-Noise ratio curve
From CCD to EMCCD
Dual Amplifier EMCCD: Traditional Amplifier for Wide-dynamic range operation
From CCD to EMCCDScientific imaging for today’s microscopy
Thank you for your attention!