Optical imaging through biological tissue has the significant problems of scattering which degrades the image resolution
and quality. Research has shown that Angular Domain Imaging (ADI) improves image quality by filtering out the
scattered light in the biological tissue images based on the angular direction of photons. The advantage of this technique
is that it is independent of the wavelength, coherent, pulse, or duration compared to OCT or time domain. This allows us
to couple ADI with conventional fluorescence imaging technique. Previous work was creating test media by varying
Intralipid/water concentration to produce different scattering levels. This showed difficulties in producing a consistent
scattering medium in liquid states. Hence, ideally we want a reusable solid medium which has a stable scattering
characteristic. Our target is to investigate fluorescence ADI on skin with cancerous collagen tissue where healthy
collagen fluoresces while the cancerous collagen tissue does not. To mimic the characteristic of skin, a solid scattering
medium over a patterned fluorescence material with non-emitting structures is created. We used a solid agar medium, or
a transparent polymer, infused with Intralipid at different concentrations, as the scattering medium. The solid media with
similar scattering characteristic of skin (μs = 20cm-1, g = 0.85) is placed on top of a fluorescence plastic (415nm
excitation, ≈ 530nm emission) which is patterned by strips of non-emitting structures (200-400μm). Using small
apertures with acceptance angles of 0.171° a distance away from the solid scattering medium, these non-emitting
structures are detectable at shallow scattering tissue depth (1-2mm).
Conventional fluorescence imaging often does not have a mechanism to remove the scattering effect in biological tissue.
We use Angular Domain Imaging (ADI) to improve the detection of smaller structures in fluorescence layer over that
can be provided by existing systems. ADI is a high resolution, ballistic imaging method that utilizes the angular
spectrum of photons to filter multiple-scattered photons and accepts only photons with small angular deviation from their
original trajectory. Advantages of the ADI technique are that it is insensitive to wavelength and the sources are not
required to be high quality, coherent, or pulse, as with OCT or time domain. Our target is to perform fluorescence ADI at
shallow tissue such as skin (≈ 1mm) with a buried collagen layer. To experimentally model shallow tissue with phantoms,
a thin layer of scattering medium with similar scattering characteristic (μs = 200cm-1, g = 0.85) is placed on top
fluorescence plastic (415nm excitation, ≈ 555-585nm emission) which is patterned by strips of non-emitting structures
(200-400μm). Positioning multiple collimated arrays with acceptance angles of 5.71° on top of the scattering medium,
test structures (200μm wide) can be detected at shallow scattering medium thickness (1mm). Monte Carlo simulation
confirms that fluorescence ADI can image structures at shallow tissue depth by using collimator array with modest
filtration angles. Results show micromachined collimator arrays provide both high spatial resolution and angular
filtration on scattered photons.
Researchers have been using simple optics to image optically induced fluorescence in tissues. We now apply the Angular
Domain Imaging technique using a Spatiofrequency filter which accepts only photons within a small deviation angle
from its original trajectory to image a fluorescing medium beneath a scattering layer. A Rhodamine 6 G dye fluorescing
layer or fluorescence slides, under an Intralipid scattering medium was used. By applying ADI with acceptance angle of
0.17°, the structures are distinguishable at low scattering depth depending of the emission wavelength of the
fluorescence source. It was established previously that as the acceptance angle increases, the amount of scattered
light/noise in the images increases, however, the resolution also deteriorates. Simulations using a Monte-Carlo program
are done for both angular filters, Spatiofrequency filter and Linear Collimating Array. Due to the additional positional
filtration on top of the angular filtration with Linear Collimating Array, collimators with aspect ratio as low as 10:1 can
improve the quality of the fluorescence images significantly in both contrast ratio and resolution.
Angular Domain Imaging (ADI) is a high resolution, ballistic imaging method that utilizes the angular spectrum of
photons to filter multiply-scattered photons which have a wide distribution of angles from ballistic and quasi-ballistic
photons which exit a scattering medium with a small distribution of angles around their original trajectory. An
advantage of the ADI method is that it is suitable with a wide variety of light sources, as it is not sensitive to coherence
or wavelength and does not require a pulsed source or a highly collimated beam. We extend the ADI method to
transmissive imaging of scattering media using incoherent, collimated sources with a spatial filter comprised of a
converging lens (focal distance of 50 to 100 mm) and pinhole aperture (diameter of 100 to 500 μm) giving acceptances
angles of 0.06 to 0.6° to produce wide-beam, full-field images of planar, high contrast, phantom test objects through 5
cm thick scattering media at optical depths of up to 14.6 (scattered to ballistic photon ratio ≈ 2×106). Experimental
images, obtained using a 12 mm diameter beam produced by a
quartz-halogen incandescent source (beam divergence
angle 0.52°, beam power < 10 mW), demonstrate the advantages of this combination of broadband, incoherent source
and spatial filter: lack of interference artifacts seen with laser sources, ease of changing image magnification, simple
correlation between system geometry and resolution, and ease of spectral filtration to obtain multispectral images.
Monte Carlo simulation with angular tracking is used to validate the experimental results and determine system tradeoffs.
Angular Domain Imaging is an optical tomography technique that filters out scattered light by accepting only photons
with small deviation angles from their original trajectories. Previously, angular filters of linear collimating array (0.29°
acceptance) or spatiofrequency filter of a +50mm lens with a 214um aperture (0.25° acceptance) were used. In the linear
collimating array system, using a wedge prism to deviate the light source by 2-3x the acceptance angle creates a second
image of only the scattered components which can then be subtracted from the filtered image to enhance detectability.
We now apply this technique to the spatiofrequency filter system at an angle 2x the acceptance. Utilizing several
wavelengths of laser sources with different beam symmetries, test phantoms are placed in a 5cm thick sample of diluted
intralipid solution, with a maximum SR of 1.64×106:1 (μs' = 1.8cm-1). By digitally subtracting the background scattered
light, test phantoms previously unobservable are now distinguishable. Using background subtraction, the SR limitation
of the SFF system improves 3x under full illumination and ~40x under line of light illumination. The improvement under
partial illumination is similar to the result using the collimator array, but with resolution limited by the optics used in the
system.
Imaging structures within a turbid medium using Angular Domain Imaging (ADI) employs angular filter array aligned
to a laser source to separate ballistic and quasi-ballistic photons from the highly scattered light by means of angular
filtration. The angular filter consists of a high aspect ratio linear array of silicon micromachined tunnels, 51 micron
wide by 10mm long with a 0.29 degree acceptance angle. At heavy scattering ratios of >1E7 image detectability
declines due to the non-uniform scattered background light fraction still within the acceptance angle. This scattered
signal can be separated out by introducing a wedge prism to deviate the laser source where it enters the medium by an
angle slightly larger than the acceptance angle. This creates a second image consisting of pure scattering photons with
the filtration characteristics of the angular filter, and a pixel by pixel correspondence to the fully scattered illumination
emitted from the medium. Experiments used an 808 nm laser diode, collimated to an 8×1 mm line of light, entering a
5cm thick medium with a scattering ratio of > 1E6, with a wedge prism creating a 0.44 degree deviation. Digitally
subtracting the deviated scattered signal from the original image significantly reduced the scattered background and
enhanced image contrast. We can have about images at least 40 times more of our previous scattering limits. Depending
on test phantom object location, the contrast level can be increased from 4% of the total dynamic range to over 50%
which results in higher definition and visibility of our micro-scale test structures in the turbid medium.
Angular Domain Imaging (ADI) is a high resolution, ballistic imaging method that utilizes the angular spectrum of
photons to filter multiply-scattered photons which have a wide distribution of angles from ballistic and quasi-ballistic
photons which exit a scattering medium with a small distribution of angles around their original trajectory. Such spatial
gating has been previously accomplished using a scanning array of collimating holes micromachined into a silicon wafer
section. We now extend that work to include using a wide-beam, full-field, converging lens and pinhole aperture system
to capture images in a single exposure. We have developed an analysis of resolution and sensitivity trade-offs of such a
system using Fourier optics theory to show that the system resolution is primarily governed by collimation ability at
larger aperture sizes and by spatiofrequency (Fourier space-gated) filtering at smaller aperture sizes. It is found that
maximum sensitivity is achieved when spatiofrequency resolution and collimation resolution are equal. Planar, high
contrast, phantom test objects are observed in 5 cm thick media with effective scattered to ballistic photon ratios
>1.25×107:1 using a wide-beam, full-field lens and aperture system. Comparisons are made between ballistic imaging
with the lens and aperture system and with the scanning silicon micromachined collimating array. Monte-Carlo
simulations with angular tracking validate the experimental results.
Angular Domain Imaging (ADI) within highly scattering media employs micromachined angular filter tunnels to detect
nonscattered photons which pass through the tunnels unattenuated while scattered photons collide with the tunnel walls.
Each tunnel is micromachined approximately 51 &mgr;m wide by 10 mm long in silicon, giving a maximum acceptance
angle of 0.29 degrees. The ADI technique is inherently independent of wavelength, and thus multispectral laser sources
can be incorporated. Previous ADI experiments employed a 488-514 nm Argon ion laser source. This paper describes
the construction of a new imaging system utilizing a high-power (up to 0.5 W) laser diode at the 670 nm wavelength,
along with an aspheric and cylindrical lens system for shaping the beam into a collimated line of light. ADI results of
biological samples (i.e. chicken breast tissue) are also presented. Image resolution is 204 &mgr;m or better in compressed
chicken breast tissue approximately 3.8 mm in thickness. Digital image processing techniques are employed to improve
image contrast, definition, and detectability of test structures. Because silicon is 40% reflective, scattered light at up to
three times the acceptance angle is not sufficiently absorbed by the angular filter tunnels and contributes significant
background noise, thus decreasing image contrast and detectability. Roughening of the tunnel surface using a NH4OH
etchant solution scatters light hitting the walls, thus allowing it to be absorbed. Images after roughening show dramatic
reductions in background scattered light levels between tunnels, suggesting that further experiments will make progress
towards improved contrast and detectability of structures.
Coherence or Time Domain Optical tomography within highly scattering media observes the shortest path photons over the dominant randomly scattered background light. Angular Domain Imaging employs micromachined collimators detecting photons within small angles of aligned laser light sources. These angular filters are micromachined silicon collimator channels 51 microns wide by 10 mm long on 102 micron spacing, giving an acceptance angle of 0.29 degrees at a CMOS detector array. Phantom test objects were observed in scattering media 5 cm thick at effective scattered to ballistic ratios from 1:1 to greater than 1E8:1. Line and space test objects detection limits are set by detector pixel size (5.5 microns) not collimator hole spacing. To maximize the ballistic/quasi-ballistic photons observed, a line of light aligned with the collimator holes increases detectability by reducing the amount of scattered background light. A Cylindrical Spherical Cylindrical beam expander/shrinker creates a 16 mm by 0.35 mm line of light. Best results occur when the scattering medium, collimator and detector are within 3X the Rayleigh Range of the beam's narrow vertical axis, allowing imaging of 51 micron lines/spaces at 3E8:1 scattering ratios. Restricting the light to a 1 mm line extends this to 8E9:1. Carbon coating the SMCA to reduce reflectivity shows that at high scattering levels absorbing walls will reduce background light, improving contrast. ADI has also been shown to work when the illumination is unaligned with the detector. This allows for side illumination with detection of structures at depths of 3mm with a scattering ratio of 1E6:1.
Optical tomography within highly scattering media has usually employed coherence domain and time domain imaging, which observe the shortest path photons over the dominant randomly scattered background light. An alternative, Angular Domain Imaging, employs micromachined collimators which detect photons within a small angle of the aligned laser light source. These angular filters consist of micromachined silicon collimator channels 51 micron wide by 10 mm long on 102 micron spacing giving an acceptance angle of 0.29 degrees at a CCD detector. Phantom test objects were observed in turbid mediums ranging from 1 to 5 cm thick at effective scattered to ballistic ratios from 1:1 to greater than 100,000,000:1. Simple line and space test objects detection limits are set by detector pixel size not collimator hole spacing. Restricting the light emission to only the collimating array hole area provides increased detectability by reducing the amount of scattered light background. This is best done using cylindrical spherical cylindrical lens beam expanders/shrinkers to create a wide line of light of small thickness aligned to the collimator array. As object locations within the medium are moved closer to the detector/collimator, image detectability appears to depend on the scattering ratio after the test object rather than the total medium scattering. Hence, objects located closer to the detector than the middle of the medium are observed at a much higher scattering levels than those nearer the light source.
Quasi-ballistic or "snake" photons carry useable information on the internal structure of scattering mediums such as tissues. By defining quasi-ballistic photons to be those photons that have been scattered but have not exceeded a specified radial distance threshold from their initial trajectory (equivalent to the resolving limit of the quasi-ballistic photons) and by using the Henyey-Greenstein phase function, Monte Carlo modeling has shown that the number of quasi-ballistic photons increases with depth in an isotropic scattering medium until a maximum is reached and then the quantity decreases. The quantity of quasi-ballistic photons at a specified depth can be shown to be governed by two competing processes: the decay of ballistic photons into quasi-ballistic photons and the decay of quasi-ballistic photons into scattered photons. These well-defined behaviors allow one to write a rate equation governing the growth and decay in the quantity of quasi-ballistic photons with depth. It is found that as the anisotropy factor increases with forward scattering and as the resolution limit is widened, the quantity of quasi-ballistic photons begins to exceed the quantity of ballistic photons at a specified depth and the rate of decay of quasi-ballistic photon quantity decreases. The development of a rate equation for the formation of quasi-ballistic photons allows one to analyze how efficient various detection methods are in extracting these quasi-ballistic photons, and it can be seen that there is a compromise between desired resolution and the effective scattering ratio at a detector.
Most optical tomography work within highly scattering media has employed coherence domain and time domain methodologies, both detecting the shortest path photons over the dominant randomly scattered background. Angular domain imaging instead uses micromachined collimators to observe only those photons within a small angle of the aligned laser light source, which simulations show are the shortest path photons, while rejecting heavily scattered light. These angular filters consist of micromachined silicon collimator channels 51 micron wide by 10 mm or 20 mm long on 102 micron spacing giving acceptance angles of 0.29 to 0.15 degrees on a CCD detector. Phantom test objects were observed in mediums ranging from 1 to 5 cm thick at scattered to ballistic ratios of 500,000:1 to 10,000,000:1 depending on the illumination pattern. Object detection was retained at the same scattering levels for either 1 cm or 5cm thick mediums, demonstrating little dependence on medium thickness. Detection was also independent of the object size: phantoms ranging from thin structures of 100 micron wide lines and spaces to 4 mm spheres were detected at approximately the same scattering ratios. Minimum size resolution depends on CCD pixel size, not the collimator characteristics. Furthermore, detection was a function of the scattering ratio produced after the phantom's position, not of the whole medium’s scattering ratio. This means objects nearer the detector are much more observable. Longer collimators significantly increase the scattered light rejection. Monte-Carlo simulations with angular tracking demonstrate the object size independence and are undertaken to verify the other behaviors.
Optically Tomography within highly scattering material has focused on Coherence Domain and Time Domain methods: both detecting the shorting path photons over the dominant randomly scattered background light. Angular Domain Imaging instead uses collimators, small acceptance angle filters, to observe only those photons closely aligned to a laser light source. A linear collimating array was fabricated using silicon surface micromachining consisting of 51 μm wide by 10 mm long etched channels with 102 μm spacing very high aspect ration (200:1) 20 mm wide array. With careful array alignment to a laser source, restricted to a linear beam, the unscattered laser light passes directly through the channels to a CCD detector, and the channel walls absorb the scattered light at angles >0.29 degrees. With a computer controlled Z axis objects within a 1 cm thick scattering material were scanned quickly. High contrast 150 μm lines/spaces at the medium front were observed at scattered to ballistic photon ratios >5×105:1 with a 10 mm beam. Narrowing the beam to 130 μm width produces detectable images >3×108:1. Objects closer to the detector were more visible, and mid point objects were detectable >109:1. Smaller channels and longer arrays should enhance detection by factors of >100.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.