Biophotonics Imaging Laboratory

• Home
• Background
  • Technology
    • Optical Coherence Tomography
    • Coherent Anti-Stokes Raman Scattering
    • Multi-Photon Microscopy
    • Targeted Molecular Imaging
    • Elastography
    • Optical Properties
    • Light Sources
  • Optics in Biology and Medicine
  • Academics
    • Professor Boppart's Courses
    • Other UIUC Courses
    • Instructional Resources
• Research
  • Cancer Detection and Diagnosis
  • OCT Imaging Technology
    • Light Sources
    • Beam Delivery
    • Portable Systems
  • Molecular Imaging
    • Spectroscopic
    • Contrast Agents
    • NIVI
  • Optical Computational Imaging
  • Multimodal Imaging
  • Elastography
• Publications and Presentations
  • Publications
    • Cover Figures
    • Books, Chapters
    • Reviews
    • Invited Publications
    • Contributed Publications
  • Presentations
    • Invited
    • Contributed with Proceedings
    • Contributed with Abstracts
  • Patents
• People
  • Professor Stephen Boppart
    • Biography and Contact
    • Professional
    • Education and Training
    • Awards
  • Current Members
  • Past Members
  • Collaborators
• Resources
  • Facilities
  • Links
  • Funding Sources
• Gallery
• Group Events

Optical Computational Imaging

Interferometric Synthetic Aperture Microscopy (ISAM)

State-of-the-art methods in high-resolution three-dimensional optical microscopy require that the focus be scanned through the entire region of interest. However, an analysis of the physics of the light-sample interaction reveals that the Fourier space coverage is independent of depth. We have shown that by solving the inverse scattering problem for coherence microscopy, computed reconstruction yields volumes with a resolution in all planes that is equivalent to the resolution achieved only at the focal plane for conventional high-resolution microscopy. The entire illuminated volume has spatially-invariant resolution, thus eliminating the compromise between resolution and depth-of-field. This novel computational image-formation technique is capable of real-time imaging, and has the potential to broadly impact real-time 3-D microscopy and analysis in the fields of cell and tumor biology, as well as in clinical diagnosis where in vivo imaging is preferable to biopsy. We have analyzed and demonstrated this technique for rectangular Cartesian-coordinate imaging, for rotational catheter-based imaging, and for full-field imaging.

Renderings of 3-D OCT (left) and 3-D ISAM reconstruction (right) of a silicone tissue phantom containing 1 m TiO2 scatterers, demonstrating spatially invariant resolution.

En face images from human breast tissue. Images are shown for different depths. Histological sections (a,b) show comparable features with respect to the unprocessed interferometric data (c,d) and the ISAM reconstructions (e,f). The ISAM reconstructions resolve features in the tissue that are not decipherable from the unprocessed data. The green dashed arrow indicates the fast-scanning direction for the volumetric data acquisition.

En face ISAM and OCT images from rat adipose tissue ex vivo showing that ISAM reconstruction of a plane far from focus provides comparable resolution to moving the focus to that plane. (a) En face OCT of a plane approximately 8 Rayleigh ranges above focus. (b) ISAM reconstruction of the same en face plane. (c) En face OCT with the focal plane moved to the plane of interest in (a). The field of view in each panel is 500 m by 500 m.

Original object model of point scatterers (left) and simulated OCT image (right) showing the central in-focus region with blurring above and below the focal region.

Unfiltered reconstruction of data shown above (left) and Tikhonov regularized solution of data shown above (right). Note that spatially-invariant resolution is apparent, both inside and outside of the confocal parameter of the focused Gaussian beam.

Simulated (left) and reconstructed (right) radial OCT catheter images of randomly distributed point scatterers.

Simulated volume of point scatterers imaged with OCT to demonstrate spatially-invariant resolution. Projection of (a) unprocessed and (b) processed data on 2-D plane, and full 3-D rendering of (c) unprocessed and (d) processed volume of point scatterers. All axes are labeled in units of wavelength. Arrows in (c) indicate wavefront interference which is resolved in (d).

Dispersion Compensation Algorithms

Practical clinical optical coherence tomography systems require automatic tools for identifying and correcting flaws in OCT images. One type of flaw is the loss of image detail owing to the dispersion of the medium, which in most cases is unknown. We have developed an autofocus algorithm for estimating the delay line and material dispersion from OCT reflectance data. This autofocus algorithm can be used in conjunction with a high-speed, digital-signal-processor-based OCT acquisition system for rapid image correction.

On the left is a cross section schematic of a PDMS microfluidic structure. On the right are autofocus digitally corrected reflections off of interfaces of the microfluidic structure. Plots (a), (c), (e), and (g) correspond to the uncorrected reflectance functions, whereas (b), (d), (f), and (h) are the corrected point spread functions.

Manually-scanned OCT data acquisition

Accurate image formation in OCT requires the acquisition of A-scans (depth-resolved scattering profile) in synchronization with the movement of the focused beam in the sampling arm. Current OCT scanning mechanisms are based on precision lateral scanning of the incident light beam using mechanical scanning mechanisms. Although these methods are extremely accurate; their limited field of view and fixed scan geometry can restrict their usability in clinical and intraoperative settings. Alternatively, freehand scanning may be used for image formation using motion estimation techniques to combat the artifacts arising due to variations in lateral scan velocity and probe orientation. Sensor-based position estimation based on acoustic, optical or magnetic sensing techniques can be used for image assembly but these techniques have limited spatial resolution, make the sample arm more complicated and inflexible in addition to imposing a number of constraints on image acquisition

We have developed a cross-correlation-based algorithm that can assemble images from lateral sensorless manual-scanning of the sample. Quantifying the correlation by the Pearson correlation-coefficient enables the estimation of the relative displacement between the sample and the probe, and by selecting an appropriate threshold for discarding the correlated A-scans we can compensate for variations in the lateral scan velocity.

Image assembly for human breast tissue (on the left) and plasticine sample (on the right) over a sample length of 1 cm. (a, b) Galvanometer-scanned image. (c, d) Manually-scanned image. (e, f) Assembled image by cross-correlation based algorithm.

This sensorless manual-scanning technique may offer a flexible and computationally simple alternate scanning mechanism that can be used in conjunction with hand-held probes, OCT catheters and imaging needles to provide high-resolution images of tissue structures over an extended field-of-view with user-defined scan geometry.

Speckle Reduction in OCT Images

In OCT, ultrasound, synthetic-aperture radar, and other coherent ranging methods, speckle can cause spurious details that detract from the utility of the image. Speckle is a problem inherent to imaging densely scattering objects with limited bandwidth. We have developed methods to minimize speckle and improve image quality while maintaining features that are consistent with the known data.

OCT images of structures in a Xenopus laevis (African frog) tadpole before and after application of the I-divergence regularization method to remove speckle.

Projected Index Computed Tomography (PICT)

Projected index computed tomography (PICT) is an imaging technique that provides a computed reconstruction of the index of refraction of a sample. PICT makes use of data from standard optical coherence tomography images taken from several view angles to determine a mapping of the refractive indices of the sample. A rectilinear propagation model is assumed, so the data are understood to be related to the line integral of the refractive index in the beam paths. These data thus provide a set of angular projections of the sample. The spatial distribution of the index of the object may then be reconstructed by use of standard filtered backprojection techniques. The resultant PICT images are free of the spatial distortion that is inherent in standard optical cross-sectional images and correspond well to the manufactured dimensions of specific samples. PICT has the potential to produce high-resolution, highly sensitive, and spatially accurate images of a variety of biological and material samples.

Glass capillary tube sample (left) is imaged by PICT method (right), which shows a spatial map refractive index. Linear artifacts in the image stem from the sharp discontinuities in projected data.

Adaptive Spectal Adopization

In optical coherence tomography (OCT), as in other imaging modalities, one desires to achieve the highest possible resolution given instrument limitations. The bandwidth of the OCT source chiefly determines the useful resolution. However, if the spectrum is not smooth, then the point response in the interferogram will have large sidelobes that cause a degradation of effective resolution and introduce artifacts. As a result, a smooth, Gaussian-like spectrum is often employed to minimize sidelobes. However, there are many sources that do not produce smooth spectra but still produce a wide bandwidth, such as Ti-sapphire oscillators, nonlinear supercontinuum generation optical fibers including microstructured and tapered fibers, and ultrahigh-numerical-aperture fiber sources. We have created a method that produces an estimate of object scattering density (or reflectivity profile) that minimizes sidelobes while not overly increasing the contribution of noise. This method can make broad spectrum sources much more useful for high-resolution OCT.

Spectrum of source (solid line) and sidelobe-corrected instrument response function (dashed line). The OCT system used a UHNA3 fiber as a source, and spectra were obtained from the reflections off of a microscope slide.

Point response of original spectrum (top) and sidelobe-corrected instrument response function (bottom) obtained with a microscope slide as a test sample.

Image Processing

It is possible to achieve Doppler imaging with OCT when imaging blood vessels or any other movement. When blood flows towards or away from the transducer, the frequency of the backscattered light varies depending on the velocity of the fluid the light is incident upon. This shift can then be used to calculate the velocity and direction of the blood. This technology is especially useful in the study of microfluidics. Below is an example of Doppler imaging in microfluidics. The left image is a cross-section of a tubular microfluidic channel after Doppler OCT processing. On the right is a graph showing the flow velocity measured vs. the ideal flow profile.

Questions or Comments about the website? Please contact Guillermo Monroy at: gmonroy2@illinois.edu
Copyright © 2005-2012, Stephen A. Boppart, Biophotonics Imaging Laboratory