Low-Light Imaging

Ever since the invention of the digital camera, new imaging applications have been explored. The increasing possibilities of fast digital cameras have resulted in applications that were unthinkable only twenty years ago. High-speed cameras nowadays are widely used for recording of dynamic events at high frame rates (e.g. 10000 fps). The results can then be inspected by playing individual frames at a lower speed.

High-speed imaging up to 100000 fps is easily feasible with current technology. But what if you need to create high-speed images when light conditions are far from optimal? Your high-speed camera will be no good under these circumstances, as a certain brightness of the object is required for the high frame rates that are used. The lack of light in combination with short exposure times will result in underexposed and noisy images. The obvious solution would be to increase the illumination level of the object. However, in some cases it is just not possible to add more light, for example because:

 

  • The object to be recorded generates light by itself. This may be the case for phenomena like the combustion process (flames and turbines), or in living cells that emit fluorescent light.
  • The radiation level corresponding to the required brightness would cause an unacceptable temperature rise of the object.

And what if the image signal has become too low because of the high frame rates? Camera noise will be an additional problem then. Fortunately, there is a high-tech solution for these problems: the image intensifier. It is used to intensify the image before it is projected onto the image sensor of the high-speed camera. The intensified image results in a sensor signal that is typically 10 000 times higher than without using an image intensifier—in the process elevating the signal above camera noise level.

Intensified CCD Cameras

An Intensified CDD (ICCD) camera is an electronic camera, equipped with an intensified CCD as image sensor. The sensor uses an image intensifier that is fiber-optically coupled to the CCD chip to increase the sensitivity down to single photon level.

An intensified CCD camera allows image acquisition at very low light levels over a wide light spectrum and at relatively high speeds. Single photons can be detected and discriminated from CCD noise. Ultra high-speed phenomena can be captured by using the image intensifier as a fast shutter (gating).

CCD Camera Sensitivity

At low light levels standard CCD/CMOS cameras are not sensitive enough to capture useful high-contrast images. There are ways to increase the sensitivity of such cameras. The first method is to allow the CCD to integrate for much longer times. In order to prevent high background noise, CCD cooling is applied when using long exposure times. A second method is to use an image intensifier to boost the input signal.

Cooled CCD

At longer integration times of a CCD, more light is captured to enhance images. However, not just more input signal is collected, but also more dark current from the CCD itself. The amount of dark current depends strongly on the temperature; for every 6 degrees C the CCD is cooled down, the noise (dark current) halves. When the CCD is cooled to -25 degrees C, integration times up to minutes can be applied. This enhances the sensitivity of the camera immensely.

To better improve the cameras SNR, the read-out noise is reduced by using a lower read-out speed. These techniques are used in high performance 14 and 16 bits digital cameras.

Intensified CCD with Fiber-Optic Coupling

An image intensifier helps to increase the sensitivity of a camera by amplifying the input light-signal before relaying it to the CCD/CMOS sensor of a camera. Roughly, there are two ways to relay the output image, from an image intensifier, to a CCD/CMOS sensor. The first is by means of a relay lens. A lens coupling is flexible, but the downside is that a lens coupling has a low transmission efficiency, caused by the limited aperture of a lens. A more efficient way is to use a fiber-optic window to transfer the image from the intensifier to the CCD. A fiber-optic window contains a large number of microscopic (6-10 micron) individual fibres and acts as an image guide. A tapered fiber-optic window will magnify or demagnify input images. Generally, demagnification is chosen to match the image intensifier to the CCD/CMOS sensor.

In summary, the advantages of a fiber-optic coupling are:

  • Low light losses
  • Intensifier/CCD combination is more compact
  • Camera design is sturdier
  • No optical adjustments are needed

Related Posts

Third Generation Image Intensifier

The next step in technology is the third generation (GenIII) image intensifier in which the multi-alkali photocathode is replaced by a Gallium-Arsenide (GaAs) or a Gallium-Arsenide-Phosphide (GaAsP) photo-cathode. The quantum efficiency (QE) of these types of photo-cathodes are much higher as compared to the multi-alkali photocathode of the second-generation image intensifiers.

Recently, new filmless Gen III intensifiers have been developed that are using the high QE to its full extend. The higher QE results in a better SNR or in shorter exposure times at equal SNR. In the graph, spectral sensitivity curves of multialkali photocathodes, such as S25, S20 and broadband, are shown in comparison with GaAs and GaAsP photocathodes.

Second Generation Image Intensifier

In the second-generation image intensifier a so-called Micro-Channel Plate or MCP is added, improving the gain of the image intensifier enormously. The MCP is placed between the cathode and the anode and acts as an electron multiplier.

The MCP is a 0.5 mm thick plate with millions of 6 micron wide holes. An accelerated electron coming from photocathode will be accelerated towards the MCP. When the electron hits the wall of one of the MCP channels it will spawn secondary electrons. Due to the voltage over the MCP, these electrons will also be accelerated, and hit the surface of the MCP in their turn. Again spawning new (tertiary) electrons. This process is repeated several times, resulting in an electron gain far higher (several thousand times higher) than in first generation intensifiers.

When an electron leaves the MCP, it is propelled to the phosphor screen where it will generate multiple photons. The overall gain of an image intensifier is up to ten thousand. With two or three MCPs amplifications up to 10 million times gain is possible. The gain of the image intensifier can even be controlled by changing the voltage over the MCP.
 

Gating the Image Intensifier

An important feature of the MCP, and therefore the second-generation image intensifier, is that it can be gated. Gating the image intensifier offers a whole new possibility of using the image intensifier as an ultra fast (electro-optical) shutter. Gating is achieved by controlling the photocathode voltage of the image intensifier, creating a shutter with effective exposure times down to a few nanoseconds.

By applying a negative voltage to the photocathode, typically -200 V with respect to the MCP input, photoelectrons are generated in the photocathode. They are emitted and accelerated to the MCP to be multiplicated. In this situation the image intensifier is "gated on". When applying a small positive voltage to the photocathode, typically 50 V with respect to the MCP, the photoelectrons can not be emitted and the intensifier is "gated off". With this gating option the input light range is extended significantly and it offers unique options for time resolved experiments.

In the Lambert Instruments Fluorescence Attachment, a second-generation image intensifier is used as a detector. The image intensifier, combined with a CCD camera, is attached to a widefield fluorescence microscope. The photocathode of the intensifier is located in the image (focal) plane of the microscope. In the frequency domain LIFA, the photocathode is switched from positive to negative at the same frequency as the light source is modulated.

Furthermore gating can be used to reduce or prevent the effect of motion blur when capturing fast moving objects. In our Intensified cameras gating is standard synchronised with the exposure period of the CCD or CMOS sensor.
 

Advantages

  • Fibre-optic/glass/quartz/MgF2 input windows
  • Many photocathode types from UV to NIR
  • High gain
  • Fast shuttering is possible (gating)
  • Good over-illumination protection
  • Maximum output brightness control
  • Wide gain control range
  • Many types of output phosphors
  • Distortion free


Disadvantages

  • Limited intra-scene dynamic ranges
  • Low maximum output brightness for fast phosphors
  • No de-magnifying models
  • MCP introduces extra noise

First Generation Image Intensifier

A first generation image intensifier does not use a micro-channel plate. The electrons are guided from the input to the output by means of electrostatic focussing. Two types can be distinguished: proximity focussed diodes and electrostatic inverters. In the latter a structure of electrodes form an electrostatic lens that focusses the electrons coming from cathode onto the anode. The advantage of electrostatic focussing is that it allows de-magnification of the image. This is especially interesting when these devices are coupled to small CCDs.

Advantages of first generation tubes

  • Available in de-magnifying formats

  • Therefore no fiber optic taper required

  • No MCP noise

  • High intra-scene dynamic range

  • Low cost (standard models)

Disadvantages

  • Electrostatic inverters show a few percent of image distortion

  • Relatively low gain

  • Gating not possible

  • No UV sensitivity

  • Limited external gain control

  • Poor over-illumination protection

The Image Intensifier

An image intensifier is a device that intensifies low light-level images to light levels that can be seen with the human eye or can be detected by a camera. An image intensifier consists of a vacuum tube with several conversion and multiplication screens.

An incident photon will hit a light sensitive photo-cathode screen. Photons are absorbed in the photocathode and give rise to emission of electrons into the vacuum. These electrons are accelerated by an electric field to increase their energy and focus them on the multi channel plate (MCP).

 

Inside the MCP the electron image is multiplied, after which the electrons are accelerated towards an anode screen. The anode screen contains a layer of phosphorescent material that is covered by a thin aluminium film.

The anode contains a phosphor such that when striking the anode the energy of the electrons is converted into photons again. Because of the multiplication and increased energy of the electrons the output brightness is higher as compared to the original input light intensity.


Related Posts

Intensified Cameras for Lifetime Imaging

Intensified cameras enable full-field frequency-domain and time-domain FLIM. The image intensifier becomes an ultra-fast electro-optical shutter by operating it at radio frequencies allowing time-resolved imaging. The high-resolution image intensifier is the key component of the TRiCAM (part of the LIFA) and the TRiCATT camera attachment. Its photon gain is typically in the range of 100 to 10000. Lambert Instruments provides different image intensifiers based on photocathodes with different spectral sensitivity to match a range of applications in the UV, visible and NIR.

For FLIM in the lifetime range of 0 ps to 1 ms we provide S20 (UV) and SuperS25 (visual) image intensifiers. For increased quantum efficiency of the photocathode in the visual part of the spectrum in this lifetime range, a GaAs intensifier is available. For near-infrared applications up to about 1100 nm an InGaAs photocathode is available.

The graph below shows the spectral sensitivity of these photocathodes.

Frequency-Domain FLIM: Basic Equations

In this article the first principles of frequency domain (FD) fluorescence lifetime imaging microscopy (FLIM) are further explained through use of equations. Although these principles not necessary for the execution of basic lifetime measurements, a thorough understanding provides the groundwork that enables deeper insight into your results and into the possibilities of FD lifetime imaging.

Fluorescence Lifetime

For an ensemble of fluorescent molecules the rate at which molecules decay from the excited state to the ground state (cf. Fig.1) is proportional to the number of excited state molecules \(N\):

\begin{equation}
\frac{d N}{d t} = - k_{F} N
\end{equation}

where \(k_{F}\) is the fluorescence decay rate (in units of \(s^{-1}\)).

Figure 1. Jablonski diagram and spectra showing the fundamental photophysical processes in organic molecules: absorption of a photon (      S  0  >  S  1  ,  S  2      ), internal conversion (      S  2  >  S  1      , non-radiative), fluorescence (      S  1  >  S  0      ), intersystem crossing (      S  1  >  T  1      ) and phosphorescence (      T  1  >  S  0      ).

Figure 1. Jablonski diagram and spectra showing the fundamental photophysical processes in organic molecules: absorption of a photon (S0>S1,S2), internal conversion (S2>S1, non-radiative), fluorescence (S1>S0), intersystem crossing (S1>T1) and phosphorescence (T1>S0).

In case the number of excited state molecules is not re-supplied and there are no quenching mechanisms other than fluorescence, the number of molecules that drop to the ground state at time \(t\) is given by the solution to equation [1]:
\begin{equation}
N(t) = N_{0}\ e^{-t/\tau}
\end{equation}

\noindent where \(\tau = k_{F}^{-1}\) is the fluorescence lifetime, or the average time the fluorescent molecules spend in the excited state. It is this characteristic timescale which is measured with Fluoresence Lifetime Imaging Microscopy (FLIM) technology.

Modulated excitation and fluorescence emission

The exponential decay is the impulse or fundamental response of an ensemble of fluorescent molecules and only follows after excitation by an infinitely short excitation pulse. In practice, the excitation light is not an infinitely short pulse, and the fluorescence emission \(F(t)\) is the convolution of the excitation light waveform \(E(t)\) with the impulse response:
\begin{equation}
F(t) = \int_{0}^{t} E(t') F_{\delta}(t-t') dt'
\end{equation}

For FLIM the excitation signal \(E(t)\) can be a train of pulses or any repetitive waveform:
\begin{equation}
E(t)=E_0+\sum_{\substack{n=-\infty\\n\ne0}}^{+\infty}\left|E_{\omega,n}\right|e^{i(\omega_nt + \phi_n^E)}
\end{equation}

Figure 2. Typical excitation signal used in a frequency-domain FLIM system, the fundamental frequency adopted in this example is 50 MHz.

Figure 2. Typical excitation signal used in a frequency-domain FLIM system, the fundamental frequency adopted in this example is 50 MHz.

which is a Fourier series with \(|E_{w,n}|\) the amplitude of the \(n\)th frequency component which has frequency \(w_{n}^{E}\) and phase \(\phi_{n}^{E}\). The fundamental frequency of the excitation light is chosen to best resolve particular fluorescence lifetime components. For example, nanosecond fluorescence decays can be probed using MHz frequencies.

For a single fluorescence lifetime species (i.e. equation 2) the fluorescence emission \(F(t)\) following excitation by this repetitive waveform is then:

\begin{equation}
F(t)=E_0\tau+\sum_{\substack{n=-\infty\\n\ne0}}^{+\infty}\left|E_{\omega,n}\right|\frac{\tau+\omega_n\tau^2}{1+(\omega_n\tau)^2}e^{i(\omega_nt + \phi_n^E - \phi_n)}
\end{equation}

Figure 3. Fluorescence emission (red), showing a phase lag and reduced amplitude with respect to the excitation signal (blue) due to the fluorescence lifetime.

Figure 3. Fluorescence emission (red), showing a phase lag and reduced amplitude with respect to the excitation signal (blue) due to the fluorescence lifetime.

Figure 4. Modulated detector gain (green), pulsed excitation (blue) and fluorescence emission (red) in a homodyne fluorescence lifetime imaging system. The phase of the detector gain is controllable.

Figure 4. Modulated detector gain (green), pulsed excitation (blue) and fluorescence emission (red) in a homodyne fluorescence lifetime imaging system. The phase of the detector gain is controllable.

This waveform can be measured in the time domain or the frequency domain. It oscillates at the same high-frequencies as the corresponding Fourier components of the excitation light \(E(t)\), but exhibits a time delay or phase lag \(\phi_{n}\) at each frequency.

\begin{equation}
\phi_{n} = \arctan(\omega_{n} \tau)
\end{equation}

In addition to this phase lag the fluorescence emission is demodulated, i.e. the modulation depth or relative amplitude of the emission is attenuated relative to the pure excitation waveform.

For a single lifetime fluorescence species the modulation depth is:

\begin{equation}
M_{\omega, n} = 1 / \sqrt{1 + (\omega_{n} \tau)^2}
\end{equation}

Homodyne detection

In frequency-domain (FD) FLIM systems the fluorescence lifetime can be obtained from measurements of the phase lag and demodulation of the emission as compared to the excitation light. For convenience, simplicity of the instrumentation and to avoid high frequency noise, the high-frequency (HF) fluorescence signal \(F(t)\) is not measured directly in the time domain but instead converted to a low-frequency (LF) signal. This is accomplished using a detector of which the gain is shuttered just as the excitation light source. This frequency mixing phenomenon, the conversion of HF signals to LF signals, is well-known and the basis of radio technology. In the homodyne detection method, the excitation light and detector are modulated at the same frequency (\(\omega_{n}^{G} = \omega_{n}^{E} = \omega_{n}\)). The detector gain is then:

\begin{equation}
G(t)=G_0+\sum_{\substack{n=-\infty\\n\ne0}}^{+\infty}\left|G_{\omega,n}\right|e^{i(\omega_nt + \phi_n^G)}
\end{equation}

with \(|G_{w,n}|\) the amplitude of the \(n\)th frequency component, \(\omega_{m}\).

At a certain phase difference between the detector gain curve and the modulated excitation, the measured signal in an FD system \(S\) is the real-time product of the fluorescence emission and detector gain:

\begin{equation}
S(t)=\left\{F(t) \cdot G(t)\right\}_\textrm{LF} \propto F_0\left(1 +M_{\omega,n}\cos(\phi_n)\right)
\end{equation}

In a homodyne system, \(S\) is measured at a series of phase steps covering \(360\) degrees, and at each phase setting the detector signal is integrated for a time period much longer than the period of the HF modulation, averaging the signal. The resulting homodyne signal or phase-modulation diagram (an integral of Eqn.~9 over time \(t\)) exactly preserves the phase lag and the demodulation of the high frequency fluorescence emission, and can be directly translated to a fluorescence lifetime (through e.g. Eqns.6 & 7 for  a single lifetime species).

Figure 5. FD homodyne signal of the fluorescence emission (red), showing a phase lag phi and demodulation M with respect to the excitation light (blue), exactly as in the time domain (figure 2).

Figure 5. FD homodyne signal of the fluorescence emission (red), showing a phase lag phi and demodulation M with respect to the excitation light (blue), exactly as in the time domain (figure 2).

For multiple fluorescence lifetime species the impulse response to an ultra-short excitation pulse is \(F_{\delta}(t) = \Sigma_{s=1}^{S} a_{s} e^{-t/\tau_{s}}\). It can be shown that the fluorescence emission following modulated excitation has a generalized form of Eqn.5, with corresponding generalized expressions for the phase and modulation (e.g. Spring & Clegg, 2009). The fluorescence lifetime components and their relative contributions can be identified using multi-frequency FLIM instrumentation.

The FD FLIM approach is implemented in the gain-modulated image intensified CCD camera of the Lambert Instruments LIFA for fast full-field lifetime imaging. The videorate imaging speeds attainable with full-field FD FLIM and its model free, rapid data analysis using the measured phase lag and demodulation make it applicable to real-time FRET imaging and promising for endoscopy and clinical applications. Recent advances in full-field FLIM include implementations of video-rate 3-D confocal FLIM with a multi-beam confocal spinning disk and hyperspectral FLIM.

Figure 6. Fluorescence intensity (left, grayscale) and fluorescence lifetime (right, pseudo-color) images obtained with a Lambert Instruments LIFA.

Figure 6. Fluorescence intensity (left, grayscale) and fluorescence lifetime (right, pseudo-color) images obtained with a Lambert Instruments LIFA.

 

References

Robert M . Clegg and Bryan Q . Spring, in FLIM Microscopy in Biology and Medicine', eds. Robert M. Clegg and Ammasi Periasamy, Chapman and Hall/CRC 2009, p.115-142