How it works. Absorption Spectroscopy Basics. Why molecular absorption is important. The math behind the physics of absorption spectroscopy.
Measuring molecular absorption spectra. The original paper can be found here. For more precise details about this technology, you should look here , but the general idea is provided below in the figure.
Example of the waveforms used to modulate the light sources in swept-frequency FDPM. As the source sweeps through all the frequencies, a spectrum is built up, in similar fashion to a Bode plot. For those with engineering backgrounds, we can think of the frequency response of the tissue as a filter; the absorption and scattering of the tissue "filters out" photons at frequencies corresponding to the absorption and scattering rates of the tissue.
For those without engineering backgrounds, you can think of it this way: some processes are very fast in light transport i. By sampling at different frequencies, we are changing our sensitivity to the rates of scattering and absorption.
These frequency spectra see below of phase and amplitude are generated by performing the cross-correlation analysis on the detected APD signal. Once we have these curves, we then fit them simultaneously to a diffusion model of light transport to recover the tissue absorption and scattering coefficients. By "simultaneous" we mean that the phase and amplitude are fitted to the model using a combined chi-squared merit function to compare model i. The fits are not iterative i.
But these measurements are so fare only at one optical wavelength. Thus we need to repeat this at additional colors of light. In the current DOSI instrument, this is accomplished by serially switching electronically to another laser diode. If we do this for say four discrete wavelengths, we would then generate an absorption and scattering spectrum at 4 discrete wavelengths, as shown in the figure below. FDPM absorption spectrum left and reduced scattering spectrum.
Each point is the corresponding coefficient for a single laser diode. Integration of Steady-state Broadband Spectra. The steady-state SS method cannot measure tissue optical properties from a single source-detector location. The SS method has the advantage that the light sources need not be modulated; hence it is easier to generate SS photons at many wavelengths. Without some knowledge of the system we can not be sure we recover the true optical properties.
Why is this? There are two main reasons: The absorption and scattering cannot be separated. By measuring these optical properties, quantifiable and qualitative information about the target tissue can be ascertained. The beauty of DOSI is that it is fundamentally based on basic spectroscopic theories. To be more specific, DOSI is based on optical spectroscopy in the approximate nm to nm near-infrared NIR region of the electromagnetic spectrum. It works by interrogating the target tissue with NIR light, detecting the remitted photons, and analyzing the remission via a set of mathematical photon transport models based on Beer-Lambert Law modified to predict multiple photon scattering and diffusion in living tissues.
From this, various useful information such as tissue oximetry can be used to diagnose conditions such as melanoma. As can be seen in Figure 1, a typical DOSI instrument consists of a tunable NIR laser light source and various photon detectors placed at differing distances away from the source.
The light propagation inside a highly scattering target is mathematically described by a forward problem solver based on the radiative transfer equation RTE , or its simplified version, the diffusion equation DE [ 5 , 6 ]. These re-emitted photons can be collected using photodetectors in either reflectance geometry or transmittance geometry. The reflectance geometry can be used for both thin and thick targets such as newborn and adult heads respectively, but the reachable depth of collected photons is limited to less than 4 cm.
The transmittance geometry is only applicable for thin targets less than 8 cm of thickness such as breasts and newborn heads [ 7 , 8 ]. Different measurement geometries can be achieved using multiple channels for both sources and detectors, but this is more complex and expensive than a scanning approach that moves around one or more source—detector pairs in order to can the target [ 9 , 10 ].
After the reemitted photons are detected, an inverse problem solver is implemented to analyze the raw data of the detected photons from a DOS setup.
Then, three-dimensional 3D or two-dimensional 2D images of the optical properties of turbid targets are reconstructed. From the reconstructed images, any inclusion or heterogeneity inside the target may be detected, localized, and its size estimated with good spatial resolution up to few mm [ 13 , 14 ].
Since the s, DOS has emerged as a powerful technique to explore the chemical properties and compositions of objects in several fields such as agriculture, food, pharmaceutical, and medical imaging [ 15 , 16 , 17 ]. In the last few decades, the applications of DOI for non-invasive tomographic and topographic imaging of tissues and organs have expanded and are known by several names such as diffuse optical spectroscopy DOS [ 18 ], optical topography OT [ 19 ], and diffuse optical tomography DOT [ 20 ].
In this article, the term DOS is used to describe either the part of TR prototypes that produce raw datasets DToF histograms or the prototypes that do not reconstruct images of the targets. On the other hand, the terms DOI and DOT are used to describe any prototype that utilizes an inverse problem solver to reconstruct images from raw data obtained from DOS.
DOT systems can produce 2D or 3D images slices in transmittance geometry detectors and sources are not on the same side for thin targets less than 8 cm thickness such as muscles, breasts, and heads of newborn babies [ 21 , 22 ]. The same can be done in reflectance geometry for thicker or high-absorption tissues, but the depth of the interrogated region in the tissue is shallower than in transmittance geometry [ 8 , 23 , 24 ].
On the other hand, topographic imaging can be applied in reflectance geometry to reconstruct only 2D images from illumination sources and detectors on the same side [ 7 , 8 ]. DOI systems can be used in optical mammography for tumor detection, localization, and evaluating the response from cancer treatment [ 25 , 26 , 27 ].
In addition, functional DOI or functional near infrared spectroscopy fNIRS is used to take images of the changes in optical properties due to the variation in tissue oxygen saturation StO 2 and blood flow in the brain during doing some functional activities [ 1 , 21 , 28 ]. CW usually employs dozens of sources and detectors which restrict the scalability of the DOI systems and increase the amount data that must be analyzed to reconstruct images [ 30 ]. Moreover, CW DOS can only monitor the variation of the optical properties, so without estimating the values of optical properties, its capabilities for structural imaging are limited [ 7 ].
On the other hand, TR-DOI systems were proven to be the most powerful approach among the three techniques with respect to depth sensitivity and recovery of the absolute value of the optical properties of the targets. TR-DOI has been continuously developed and improved since the late s [ 32 , 33 ].
Nonetheless, most of the reported TR prototypes were built in research centers and academic institutions. With the use of TR-DOI, many practical limitations—such as large size, high cost, and complexity—were reduced. These improvements are primarily due to the rapid advances in photodetector technology and timing electronics that led to reduced cost and size of photon counting and timing devices by more than three orders of magnitude during the last three decades [ 37 ].
These detectors represent cutting edge technology in the miniaturization and cost reduction of picosecond photon timing and counting applications such as TR-DOI. Therefore, by exploiting solid-state detectors, the usage of TR-DOI is expected to expand in several fields significantly, and more affordable and portable commercial devices are expected within the next few years [ 37 ].
To the best of our knowledge, this is the first article to review all reported TR-DOI prototypes using semiconductor detectors and characterize them based on their specific features. To aid in the logical flow of the recent evolution of TR-DOI prototypes, this paper is organized as follows. Next, in Section 3 , the parameters which are used to evaluate the performance of the entire TR-DOS systems as well as the figures of merit of the detectors are presented and discussed. Finally, in the Section 7 , the performance of FR vs.
The first subsystem is a pulsed laser source s that illuminates light into a turbid target. A third subsystem is required to reconstruct images by using forward and inverse modeling to analyze the histograms from the photon timing subsystems and estimate the optical properties for each point within the target. Eventually, the size, location, shape, and optical properties of a high-absorbent inclusion inside a turbid media, such as tissue, can be investigated based on the variation of the optical properties with respect to surrounding normal tissue [ 38 ].
From left to right: light illumination, turbid, photon timing, image reconstruction tool. Light sources in this range of wavelengths can be utilized to monitor hemodynamics and estimate the concentrations of oxyhemoglobin HbO 2 , deoxy-hemoglobin HHb , total-hemoglobin tHb , and tissue oxygen saturation StO 2.
In addition, to monitor water and collagen in a turbid medium, light sources with wavelengths longer than nm can be used [ 39 ]. However, the use of silicon semiconductor detectors in prototypes for water and collagen is hindered by the poor performance of the detectors at this longer range of wavelengths see Section 4.
Some lab prototypes have used bulky solid-state lasers—such as Mai Tai Spectra Physics—that have higher power and ultrashort pulse width FWHM in the tens of fs range , whereas other recent setups were developed by using a supercontinuum fiber laser sources FWHM in ps range [ 33 ].
Therefore, in the literature, it is reported that the pulsed diode lasers maintain low power, which leads to a small contribution from the light source to the IRF Total of the TR prototype [ 40 , 45 , 46 , 47 ]. However, pulsed diode lasers are still widely used because of their affordable cost and availability of many models at different wavelengths from several vendors [ 48 , 49 , 50 , 51 ]. Although the maximum permissible exposure MPE was not specified for some internal tissues such as the brain, the power of the illuminated light must be kept lower than the MPE for skin 1.
According to the International Electrotechnical Commission standards IEC , it should be considered that the MPE values vary depending on the exposure duration and the wavelengths used for illumination [ 52 ]. Moreover, pulsed lasers with high repetition rates can be approximated as CW laser by using the average power of the pulsed laser which is represented by the energy of the pulse multiplied by the number of pulses in 1 s [ 54 ].
The MPE must be considered, particularly for any prototype which is to be used in clinical measurements. The most common method to classify highly scattering targets is based on the similarity of the values of their optical properties.
Most tissues such as an adult head are heterogeneous, but some tissues such as breasts and neonatal heads can be modeled as homogenous to simplify the analysis of light propagation [ 45 , 55 , 56 ]. Turbid targets are categorized according of their type; that is, whether it is a real target such as tissue, or a phantom that mimics the optical properties of a specific organ.
Phantoms are more common in preliminary experiments because of their flexibility in shape, size, value of their optical properties and the fact that no permission from ethics boards is required for their use [ 57 ]. On the other hand, ethics board approval and patient consent are required for in vivo targets in clinical experiments according to strict protocols.
Photon counting and timing that has resolution below the tens of picoseconds range is needed for TR-DOI prototypes, and this can be achieved by two separate categories of equipment. The second category consists of standalone cameras such as streak cameras or time-gated intensified charge coupled device ICCD cameras [ 33 , 45 ].
Therefore, this review focuses on TR-DOI systems of the first category because of improvements achieved in recent years by exploiting the advances of semiconductor detectors to build affordable and compact TR-DOI prototypes. For TR-DOS measurements, it is vital to have detection responses faster than 1 ns as well as stable single electron responses for each detected event.
This is because having faster detection time helps in better distinguishing between different photon arrival times so that the DToF can have higher resolution.
This will help discriminate the differences in the delay between detected photons due to the variation in path-length for each detected photon in the turbid media [ 7 , 58 ]. The benefits of the TG mode are noticeable when the reflectance geometry configuration of measurement as shown in Figure 2. Early photons in the pulse are related to photons that have passed through superficial areas of the target whereas late photons are most likely to have reached deeper areas in the target.
These late photons provide useful information [ 68 ]. Photons reached areas within the target for each time gate window with different delays. An excellent way to distinguish between early and late photons is to use very fast time gating circuits that can record the arrival times of each photon.
However, these detectors are hindered by the huge numbers of early photons which significantly increase the noise and saturate the detectors [ 43 , 67 ]. Hence, SPAD detectors are a potential alternative when building fast TG detectors which are capable of ignoring early photons and detecting late photons, within selected delays of picosecond resolution.
This causes the contrast to improve and the number of detected photons within the determined gate-window to increase. The improvement of these factors will lead to significant advances in diffuse optical imaging, particularly in detecting deep inclusions in turbid targets such as tissues. This will in turn lead to improved quality and contrast of reconstructed images. Also, using a close to null SDD allows the maximum level of the lateral spatial resolution to be reached in DOI for highly diffusive targets such as tissues that are dominated by diffusion [ 43 , 63 , 72 ].
A DToF is a histogram of the different delay times between the time of triggering of the synchronized injected laser pulse, and the PTA of the detected photons belonging to the same pulse at the detector [ 75 ]. To generate a DToF, detectors detect photons that have migrated through turbid targets, whereas for IRF measurements, detectors count photons directly from the source without involving the target. Figure 3 illustrates the principle of TCSPC and TDC measurements, and how all detected photons are stored in the DToF histogram according to the differences in delay between each detected photon and the reference pulse pulse of laser.
The reference pulse can be connected to the TCSPC or TDC from the laser driver gain switching case or through the detected signal from a photodetector that measures the injected laser pulse.
From the DToF histograms, the optical properties of a target can be determined see Section 2. Top-right in red and green borders shows two methods to synchronize the laser pulse with the detected photons.
Top-right blue border illustrates how the counted photons are stored in the DToF histogram according to the differences in delay for each one of them. Images are reconstructed by estimating the optical properties of each point inside the target. Basically, the optical properties of any object can be recovered by utilizing inverse problem models or some formulas of the analytical solution of the DE that was demonstrated for many geometries [ 81 , 82 ].
The absorption coefficient can be estimated from the linear regression of the slope of a DToF as follows Figure 4 [ 83 ]. Recovering the optical properties of time-resolved diffuse reflectance measurement for a homogenous target.
However, for more complicated geometries and shapes of targets, a regulated inverse problem depends on an iterative forward modeling of multiple DToF of photons detected at different positions are required to estimate the optical properties [ 8 ].
The DOS prototypes should implement inverse modeling in OT or DOT prototypes to recover the optical properties and to detect any inclusions by analyzing simultaneous signals from source—detector pairs attached to a target [ 10 ]. For DOT prototypes, Figure 5 represents the iterative processes of the forward modeling until the deviation between measurements and the forward modeling solver for all detectors are reduced enough comparing the measured data with the forward modeling solver to judge convergence to reconstruct images for the distribution of the optical properties [ 8 ].
The forward modelling simulates the light propagation inside the high scattering target by solving the RTE, or the simplified DE, using stochastic or numerical models such as Monte Carlo MC and the finite element method FEM respectively [ 69 , 85 ]. To use these iterative processes with all the data points of the DToF are computationally expensive.
Therefore, some accelerated FEM approaches were recently demonstrated by analyzing a few critical points from the DToF curves for transmittance geometry measurements only for simulating data , instead of using all points on the histogram that are not needed, and significantly increases the processing time [ 86 , 87 ]. Flow diagram of a TR-DOT prototype, the left and right side represent the measurement setup DOS and the flowchart of the inverse modelling respectively.
Generally, high quality image reconstruction requires a prior knowledge of the anatomy of the tissue, but because of the highly scattering nature of the turbid targets, the solution of the inverse problem becomes ill-posed, nonlinear, and ill-conditioned [ 24 , 88 , 89 ].
Hence, if this anatomical information is considered in the inverse modeling, it is called a soft prior. It is called a hard prior if the anatomical information is also being considered in the forward modeling [ 8 , 88 ]. Overall, many forward and inverse models were developed by several groups to study the light propagation and calculate DToF histograms using different geometries and shapes of targets and estimate the optical properties.
However, it is beyond the scope of this paper to discuss these models, although some review papers about modelling and image reconstruction for DOT and OT are recommended for more information [ 8 , 19 , 88 , 90 ]. Therefore, in this paper all mentioned prototypes will be characterized based on their hardware component features and the performance of the whole setup as stated in the publications. These parameters will be described in the following subsections according to the order of the subsystems of TR-DOI prototype, as mentioned in Figure 1.
It can be calculated using the formula. A laser beam is emitted in a specific wavelength with a range e. The delivered power P Source is the average laser power that is actually delivered to the sample e. P Source is the measured output optical power using optical power meter from the fiber optics.
Lastly, the illuminated area A Source represents the area of the injected light on the surface of the sample. Increasing A Source leads to a decrease of the spatial resolution for the reconstructed images from the measurement [ 74 ]. Several features of photon timing in TR-DOI systems should be considered to evaluate the performance of prototype and particularly the detectors.
The main parameters are the photon detection efficiency PDE , noise, detection responsivity, dead-time T DEAD , timing jitter, fill-factor, and total active area [ 95 , 96 ]. The noise in SPADs represents false triggering that may or may not be correlated to time. Dark Count Rate DCR is the noise that is not correlated with the avalanche process for photon detection [ 99 ]. However, after-pulsing P AP and crosstalk refer to subsequent noise pulses that appear after a detected photon is generated.
P AP happens within the same pixel that detected a photon, while crosstalk happens in external pixels [ 95 , 99 ]. The main sources of DCR in SPADs are the free-carriers because of the thermal generation that occurs in the depletion region [ 98 , 99 ]. Most of the parameters stated above have an effect on the value of the FoM T as [ 97 ]. This nonlinearity is estimated using a continuous light source to illuminate the detector, and the TCSPC module to accumulate enough detected photons.
These photons are stored in channels based on the differences in the arrival time. Theoretically, the detected photons from a continuous light source should have equal counts for all time channels [ 74 , 80 ].
However, practically, the width of channels of the timing electronics suffers from instability [ 80 ]. Figure 6 shows how the measured DToF represents the convolution of the IRF Total and the real DToF histogram of the reemitted photons from a turbid target , plus the noise in the prototype [ 74 , ].
Broaden of measured DToF. The IRF Total of the setup is estimated from experiments measuring the transmitted light from the laser source to the detector when a thin, highly scattering material with small temporal dispersion, such as a white sheet of paper or a Teflon layer, is inserted between the source and detector [ 65 , , ]. Using a thin, highly scattering material between laser source and detector ensures that the detected photons are diffused and have multiple directions when they impinge the detector similar to the re-emitted photons from a turbid target [ ].
DR could be calculated using the formula. Having high orders of magnitude of DR is vital for TR-DOI systems, particularly when small SDDs are used in reflectance geometries as they require the system to have at least five orders of magnitude 10 5 for it to be an appropriate system in applications that require deep detection capabilities such as functional brain imaging [ 43 ].
These parameters evaluate how good the estimation of the size is, the optical properties and the depth and lateral localization. Parameters that are typically evaluated are based on the depth penetration, sensitivity, localization, and spatial resolution depth and lateral of inclusions. In this review, these features will be used to compare the published performance of prototypes, because there is no standard inverse problem model for image reconstruction using TR-DOI systems [ ].
Single-photon counting detectors are used to generate an electrical signal for each photon that is absorbed. The process of counting incident photons is called the Geiger mode [ , , ]. PMTs were the most common used detectors in photon counting and timing systems for low light such as TR-DOI [ 10 , 16 , , , ]. The use SPAD detectors have increased recently because they possess several advantages over PMTs: low DCR , high quantum efficiency QE , timing jitter of less than picoseconds, small size, low power dissipation, low supply voltage, high reliability, and ultrafast gating [ ].
SPADs in standard silicon technology typically have lower sensitivity to photons of wavelengths longer than the visible range — nm due to their long absorption depths [ ]. The avalanche current flows into the junction until the quenching circuit lowers the bias voltage below the V BD. Because of the low absorption coefficient of silicon in the NIR range, the larger thickness of the absorption region plays a significant role in increasing the detection efficiency of the SPAD.
However, timing jitter in the detectors increases with thicker absorption regions. SPAD detectors can be categorized into two types based on their implementation technology: standard silicon Complementary Metal-Oxide-Semiconductor CMOS technology and custom silicon technology [ 98 , 99 ].
A SiPM is an array of hundreds or thousands of SPADs connected in parallel, and perform as a single large area detector few mm 2 with two terminals one cathode, another anode [ , ]. The total area of the SiPM can be estimated by multiplying the number of pixels by the fill-factor FF for each pixel [ 95 ]. If an active quenching circuit is connected to each pixel, this is a digital SiPM [ 95 , ]. In analog SiPMs, the number of photons can be estimated from the output current which represents a summation of all photons that are absorbed [ 95 ].
However, each single SPAD in digital SiPM is connected to a circuit to generate a signal for each counted photon, and a quenching circuit to turn off the SPAD when it exceeds the maximum time of activity [ , ]. Presently, most commercially available SiPMs cannot reach high enough resolutions to perform single photon counting with high temporal resolution in ps range without a custom module integrated with the SiPM to extract the timing information for each photon, as demonstrated in [ ].
Overall, SiPMs combine the benefits of both photocathode-based e. However, SiPMs suffer from low dynamic range of around two orders of magnitude 10 2 and a long diffusion tail because of the sequence of carriers generated inside the detector for each single photon response [ 65 ].
Detectors can be classified into two categories based on their operation mode: free-running time-invariant or time-gated time-variant [ ]. Photons can only be detected if they arrive while the gate is ON [ ].
Figure 8 illustrates the main differences between photon counting in FR and TG detectors photons—blue colored symbols , reasons for losing photons photons—green colored symbols in FR and photons—green and red colored symbols in TG, and the importance of T DEAD to reduce the effects of false triggering, particularly P AP. Photon counting process for identical incident photons using a FR detector, b TG detector.
Reasons of missing photons are indicated on the right for each one. Table 1 shows a summary of the features of the detectors. An abbreviated name will be used for each detector to indicate which one is utilized in each reported prototype.
Therefore, TG-TR-DOI prototypes require a common pulse generation unit to trigger the laser pulses and time gate windows in the same time precisely.
In addition, a delay unit is required in TG-TR-DOI prototypes to adjust the delay of the gate and enable the detection to focus only on early or late reemitted photons. Nevertheless, FR-TR-DOI prototypes are simpler than TG because the laser pulse source is triggered from the driver internally according to the adjustable repetition rate in the range of one to tens of MHz.
Moreover, the detector is always on and able to count incident photons unless a photon arrives during the T DEAD. However, it must be noted that there is a compromise between increasing the number of the integrated on-chip TDCs with reducing the fill-factor and PDE of the SPAD imagers which degrades the detectors capabilities in very low light intensity applications such as DOI.
This setup works in the reflection mode to avoid the restriction of the thickness of targets, and it aims to merge the advantages of the conventional CCD systems with information of PTA. Moreover, if more than one photon arrives in one cycle, only one of them will be recorded during the same clock cycle.
The IRF SPAD is enhanced significantly for light detection at nm from ps down to ps by increasing the excess bias voltage from 1. Recently, a TR-DOI for the imaging of small animals was demonstrated in [ 58 ] and two versions of the multichannel prototype were built.
In this prototype, 3D images are reconstructed by exploiting time of arrival of early photons EPTA which focus on the photons that have the shortest path-length from the source to the detectors.
Images were reconstructed slice-by-slice similar to the concept of Computed Tomography CT. However, all of them can reach the brain tissue of an adult when they used for functional brain imaging. This TR-DOS prototype using SiPM demonstrated high accuracy and linearity in quantifying the optical properties as well as good depth sensitivity for detecting inclusion with acceptable contrast.
Recently, in [ 65 ], the feasibility of using SiPMs for a TR-DOT scanning reflectance geometry prototype was evaluated based on several parameters such as depth sensitivity, absolute quantification of the optical properties, size estimation, and both lateral and depth localization. A supercontinuum fiber laser was used to illuminate light at nm 26 ps IRF into a liquid phantom [ ]. Five standard black cylinders of different sizes with high absorption inclusions made of black polyvinyl chloride PVC were embedded inside the liquid phantom at different depths 10, 15, 20, 25, 30 mm [ ].
The results of all the parameters were good down to 20 mm depth, but they all degraded for deeper inclusions. It is worth noting that the absolute quantification values of the optical properties of the inclusions are not accurate, mainly when they are embedded deeper than 20 mm, but this is a common limitation of DOI technology when prior knowledge is not used [ 13 ]. This prototype used 2 mW average power pulses at nm and nm wavelengths with up to 40 MHz repetition rates.
The absorption and reduced scattering coefficients were estimated from the raw data of prototype over a broad range of wavelengths — at six locations of bone prominence in the human body. Extracting this information about tissue components can lead to diagnosing some bone related pathologies.
Therefore, such PMT detectors can be used when extracting information about constituents of tissue such as collagen and water in this wavelength range [ , ]. This system uses three diode laser sources , , and nm wavelengths to illuminate the light, MPPC SiPMs connected to TDC units are used for photon counting and timing item details not specified [ ]. Using three wavelengths of laser sources aims to distinguish between HbO 2 , HHb, and tHb and quantify their percentage.
Although this prototype suffers from large IRF Total 1. In Table 2 , a summary of the main components and the major features of the above-mentioned FR prototypes is provided. Features and details for each component of the FR-TR prototypes are stated. The first part of the table light source focuses on parameters of light illumination in these prototypes. The sixth part summarizes the method and performance of image reconstruction, and these details are only available for FR-TR-DOI prototypes.
There, they used two pulsed diode lasers and nm Picoquant, Berlin, Germany to illuminate two different targets solid homogeneous phantom and in vivo head of an adult [ 51 ]. Two different widths for the SPAD gates were used ps and ps and the rise time was around ps with delay steps of ps. The IRF Total was measured for the two laser sources 90 ps for nm source and ps for nm source.
This setup proved the benefits of using time gating to improve the spatial resolution and the depth sensitivity. The contrast of detecting deep buried inclusions 18 mm within a turbid medium for small SDD is better with late time-gate windows. However, the contrast is better with early gates for the larger SDD 20 mm. Finally, the setup of TG mode was applied in-vivo for functional brain monitoring with very good contrast and sensitivity to detect activities in the brain cortex [ ].
They proved the advantages of TG to provide a large DR which leads to higher spatial resolution and better sensitivity for deep inclusions. In this setup, the gate width for SPADs is flexible from less than 1 ns up to 10 ns by very small steps of 10 ps using an external passive delayer, and the gates can be turned ON within less than ps. To compare the signal from TG vs. FR, a proper background subtraction of the noise is needed for the curves to increase the SNR at longer acquisition times.
It was concluded that, in reflectance mode of TR-DOS measurements, using TG can achieve up to eight orders of magnitude 10 8 of DR within a measurement time much faster than FR which requires a much longer and impractical measurement time to reach the same DR [ 43 ].
There, they performed measurements using a setup containing two laser pulses nm and nm to distinguish the effect of a finger tapping task on HbO 2 and HHb dynamics of an adult brain. A fiber for detection was attached at 6 mm from the laser fiber and connected to the MPD-SPAD detector through a band pass filter used to transmit the re-emitted photons. They applied 5 ns gate widths with ps second delay to focus on late photon detection.
Their measurements confirmed the advances of the TG over FR in distinguishing brain tasks with better contrast [ 63 ]. Thereafter, in [ ] a Monte Carlo model was developed to simulate the effects of variation of the gates widths on the light propagation and the collected photons for each gate.
Therefore, by applying the time-gating with a stable open gate that is synchronized precisely and long enough e. It was realized that PDM which produce smaller IRF Total ps and has shorter decay time 85 ps can distinguish inclusions up to 3 cm in depth.
On the other hand, measurements by PDM could gather information only from the early photons that primarily come from superficial regions of the target. The main cause of this limitation is the long decay time ps and larger IRF Total ps [ ]. In [ ], a non-contact TR-DOS prototype was reported which was used to investigate the contrast variations of inclusions buried in different depths using fast gating windows at null SDD.
A PDM SPAD was used, with a fixed gate width 6 ns and nine measurements with constant delay steps ps from 0 to ps were taken. The contrast was calculated experimentally with different delays and depths of inclusions, and these results agreed with the predictions using a custom Monte Carlo simulator [ ]. Eventually, this non-contact TR-DOI prototype was extended and validated successfully for in vivo applications such as functional brain imaging for adults [ , ]. Two illumination wavelengths and nm were used in the system to distinguish the changes of the HbO 2 and HHb in healthy adult brain by running the measurements during motor and cognitive tasks that activate the brain [ ].
0コメント