-
[show abstract]
[hide abstract]
ABSTRACT: The image reconstruction enhancement schemes of total variation minimization, dual meshing and iterative spatial filtering have been applied to laboratory data collected from continuous light illumination of tissue-like phantoms. Experiments include both single- and multi-target cases where variations in object size (4 mm to 20 mm), position (centred to near boundary) and contrast with the background (2:1 to 8:1) have been explored. The results show that dramatic improvements in image quality have been obtained in terms of geometric and spatial resolution measures relative to those previously reported for continuous light, but quantitative information on the actual optical properties of embedded heterogeneities is still lacking. Specifically, the geometric characteristics of object size, position and shape are generally accurate to 10-20% and the spatial resolution metrics of background-to-object size and neighbouring-edge separation are approximately 10:1. Direct comparisons are also made with images obtained with intensity-modulated light under identical experimental conditions. Images from intensity-modulated light are found to be superior to continuous light in several important ways, most notably in terms of the ability to quantitatively discriminate the optical property values of embedded targets from the surrounding background. Continuous-light images are also found to have centrally located artefacts in many instances which do not appear in the corresponding intensity-modulated cases.
Physics in Medicine and Biology 03/1998; 43(3):675-93. · 2.83 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: In this paper, an initial evaluation of our finite element based frequency-domain image reconstruction algorithm is performed for experiments where multiple millimeter-sized heterogeneities are embedded within a tissue-equivalent (optically) background medium having multicentimeter dimensions. The cases considered consist of several interesting geometry and optical property contrast combinations including (i) two different-sized targets with the same contrast at three different separation distances; (ii) two different-sized targets with different contrasts at two different separation distances; and (iii) three targets with the same and different sizes and contrasts, respectively. The reconstruction algorithm that has been used is an enhanced version of our originally developed regularized least squares approach that now includes total variation minimization, dual meshing, and spatial low-pass filtering. Quantitative measures of image quality including the size, location, and shape of the embedded heterogeneities along with errors in their recovered optical property values are presented. The results show that multiple targets can be clearly detected for all combinations of locations, sizes, and contrast levels considered, but the quantitative nature of this detection is influenced by these parameters.
Medical Physics 02/1998; 25(2):183-93. · 2.83 Impact Factor
-
Applied Optics 06/1997; 36(13):2995-8. · 1.41 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: An experimental study of the detectability of an object embedded in optically tissue-equivalent media by frequency-domain image reconstruction is presented. The experiments were performed in an 86-mm-diameter cylindrical phantom containing an optically homogeneous cylindrical target whose absorption and scattering properties presented a 2:1 contrast with the background medium. The parameter space explored during experimentation involved object size (15-, 8-, and 4-mm targets) and location (centered, 20-mm off-centered, and 35-mm off-centered) variations. Image reconstruction was achieved with a previously reported regularized least-squares approach that incorporates finite-element solutions of the diffusion equation and Newton's method solutions of the nonlinear minimization problem. Also included during image formation were image enhancement schemes-(1) total variation minimization, (2) dual meshing, and (3) spatial low-pass filtering-which have recently been added. Quantitative measures of image quality including the size, location, and shape of the heterogeneity along with errors in its recovered optical property values are used to quantify the image reconstructions. The results show that a near 22:1 ratio of tissue thickness relative to detectable object size has been achieved with this approach in the laboratory conditions and parameter space that have been investigated.
Applied Optics 02/1997; 36(1):52-63. · 1.41 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: In this paper, we explore optical image formation using a diffusion approximation of light propagation in tissue which is modelled with a finite-element method for optically heterogeneous media. We demonstrate successful image reconstruction based on absolute experimental DC data obtained with a continuous wave 633 nm He-Ne laser system and a 751 nm diode laser system in laboratory phantoms having two optically distinct regions. The experimental systems used exploit a tomographic type of data collection scheme that provides information from which a spatially variable optical property map is deduced. Reconstruction of scattering coefficient only and simultaneous reconstruction of both scattering and absorption profiles in tissue-like phantoms are obtained from measured and simulated data. Images with different contrast levels between the heterogeneity and the background are also reported and the results show that although it is possible to obtain qualitative visual information on the location and size of a heterogeneity, it may not be possible to quantitatively resolve contrast levels or optical properties using reconstructions from DC data only. Sensitivity of image reconstruction to noise in the measurement data is investigated through simulations. The application of boundary constraints has also been addressed.
Physics in Medicine and Biology 09/1996; 41(8):1483-98. · 2.83 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Diffuse optical tomography is an imaging technique whereby spatial maps of absorption and scattering coefficients are derived from the characteristics of multiply scattered light transmitted through the object. The system described here used four intensity-modulated light sources and measurements of the intensity and phase (relative to each source) at 16 or 20 detectors on the surface of a 10 cm diameter cylinder. An iterative Newton-Raphson algorithm was used to estimate the absorption and scattering coefficients at each pixel in a 17 x 17 array minimizing the difference between measured and calculated values of the intensity and phase at the measurement sites. Forward calculations of the intensity and phase were based on a multigrid finite-difference solution of the frequency domain diffusion equation. Numerical simulations were used to examine the resolution, contrast, and accuracy of the reconstructions as well as the effects of measurement noise, systematic uncertainties in source-detector location, and accuracy of the initial estimates for the optical properties. Experimental tests also confirmed that the system could identify and locate both scattering and absorbing inhomogeneities in a tissue-simulating phantom.
Physics in Medicine and Biology 11/1995; 40(10):1709-29. · 2.83 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Simultaneously reconstructed images of both absorption and scattering coeff icients in optically heterogeneous media based on frequency-domain measurements are presented. The images are obtained from absolute measured data by use of a diffusion approximation to light propagation in tissue that is modeled with a finite-element method. Images that have different contrast levels are demonstrated.
Optics Letters 10/1995; 20(20):2128-30. · 3.40 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: A finite element reconstruction algorithm for optical data based on a diffusion equation approximation is presented. A frequency domain approach is adopted and a unified formulation for three combinations of boundary observables and conditions is described. A multidetector, multisource measurement and excitation strategy is simulated, which includes a distributed model of the light source that illustrates the flexibility of the methodology to modeling adaptations. Simultaneous reconstruction of both absorption and scattering coefficients for a tissue-like medium is achieved for all three boundary data types. The algorithm is found to be computationally practical, and can be implemented without major difficulties in a workstation computing environment. Results using simulated data suggest that qualitative images can be produced that readily highlight the location of absorption and scattering heterogeneities within a circular background region of close to 4 cm in diameter over a range of contrast levels. Absorption images appear to more closely identify the true size of the heterogeneity; however, both the absorption and scattering reconstructions have difficulty with sharp transitions at increasing depth. Quantitatively, the reconstructions are not accurate, suggesting that absolute optical imaging involving simultaneous recovery of both absorption and scattering profiles in multicentimeter tissues geometries may prove to be extremely difficult.
Medical Physics 07/1995; 22(6):691-701. · 2.83 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Electrical impedance imaging is a technique which is under investigation as a noninvasive method of tracking subsurface temperature distributions and/or associated cellular response during hyperthermia. In previous work, a finite element image reconstruction algorithm for converting surface potential distributions recorded at discrete electrode positions into spatial maps of conductivity values was developed. This paper reports on a series of significant improvements in the basic image reconstruction approach. Specifically, the ability to recover both the resistive and capacitive components of tissue electrical impedance have been incorporated. In addition, the image enhancement schemes of (1) total variation minimization, (2) dual meshing, and (3) spatial low-pass filtering, have been added. Through a series of simulation studies involving both phantom-like and clinically-relevant geometries having discrete regions and continuously-varying electrical property profiles, a significantly improved ability to recover spatial images of electrical properties in the impedance imaging context is demonstrated. The results show that the new algorithm is much more tolerant of measurement noise with levels up to 1% causing relatively modest degradations in image quality (compared to 0.1% which was needed previously in order to produce high quality images). The recovered electrical properties, themselves, both resistive and capacitive, are also found to be quantitative in value with errors in the 10-20% range occurring in the majority of cases, although deviations can reach 40% or more when noise levels as high as 10% are used. Temperature estimation simulations show that maximum temperature errors are significantly reduced (to approximately 2 degrees C relative to more than 10 degrees C in previous thermal simulations) with the new algorithm; however, temperature accuracies of better than 0.5 degree C on average are still found to be difficult to achieve with electrical impedance imaging even when the enhanced image reconstruction approach is used.
International Journal of Hyperthermia 13(5):459-80. · 1.92 Impact Factor
