Ludguier D. Montejo

Columbia University, New York City, NY, United States

Are you Ludguier D. Montejo?

Claim your profile

Publications (14)8.94 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: This is the first part of a two-part paper on the application of computer-aided diagnosis to diffuse optical tomography (DOT). An approach for extracting heuristic features from DOT images and a method for using these features to diagnose rheumatoid arthritis (RA) are presented. Feature extraction is the focus of Part 1, while the utility of five classification algorithms is evaluated in Part 2. The framework is validated on a set of 219 DOT images of proximal interphalangeal (PIP) joints. Overall, 594 features are extracted from the absorption and scattering images of each joint. Three major findings are deduced. First, DOT images of subjects with RA are statistically different (p<0.05) from images of subjects without RA for over 90% of the features investigated. Second, DOT images of subjects with RA that do not have detectable effusion, erosion, or synovitis (as determined by MRI and ultrasound) are statistically indistinguishable from DOT images of subjects with RA that do exhibit effusion, erosion, or synovitis. Thus, this subset of subjects may be diagnosed with RA from DOT images while they would go undetected by reviews of MRI or ultrasound images. Third, scattering coefficient images yield better one-dimensional classifiers. A total of three features yield a Youden index greater than 0.8. These findings suggest that DOT may be capable of distinguishing between PIP joints that are healthy and those affected by RA with or without effusion, erosion, or synovitis.
    Journal of Biomedical Optics 07/2013; 18(7):76001. · 2.88 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: This is the second part of a two-part paper on the application of computer-aided diagnosis to diffuse optical tomography (DOT) for diagnosing rheumatoid arthritis (RA). A comprehensive analysis of techniques for the classification of DOT images of proximal interphalangeal joints of subjects with and without RA is presented. A method for extracting heuristic features from DOT images was presented in Part 1. The ability of five classification algorithms to accurately label each DOT image as belonging to a subject with or without RA is analyzed here. The algorithms of interest are the k-nearest-neighbors, linear and quadratic discriminant analysis, self-organizing maps, and support vector machines (SVM). With a polynomial SVM classifier, we achieve 100.0% sensitivity and 97.8% specificity. Lower bounds for these results (at 95.0% confidence level) are 96.4% and 93.8%, respectively. Image features most predictive of RA are from the spatial variation of optical properties and the absolute range in feature values. The optimal classifiers are low-dimensional combinations (<7 features). These results underscore the high potential for DOT to become a clinically useful diagnostic tool and warrant larger prospective clinical trials to conclusively demonstrate the ultimate clinical utility of this approach.
    Journal of Biomedical Optics 07/2013; 18(7):76002. · 2.88 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: We apply the Fourier Transform to absorption and scattering coefficient images of proximal interphalangeal (PIP) joints and evaluate the performance of these coefficients as classifiers using receiver operator characteristic (ROC) curve analysis. We find 25 features that yield a Youden index over 0.7, 3 features that yield a Youden index over 0.8, and 1 feature that yields a Youden index over 0.9 (90.0% sensitivity and 100% specificity). In general, scattering coefficient images yield better one-dimensional classifiers compared to absorption coefficient images. Using features derived from scattering coefficient images we obtain an average Youden index of 0.58 +/- 0.16, and an average Youden index of 0.45 +/- 0.15 when using features from absorption coefficient images.
    Proc SPIE 03/2013;
  • [Show abstract] [Hide abstract]
    ABSTRACT: We introduce here a temporally constrained image reconstruction algorithm for fast dynamic imaging of the spatial distribution of tissue parameters such as oxy-hemoglobin, HbO2, or deoxy-hemoglobin, Hb, and their derived parameters, e.g., HbT or StO2. An unknown spatial-temporal distribution of the tissue parameter is represented by a combination of basis functions where bases are predefined and their coefficients are unknown. The performance of the new algorithm is evaluated using experimental studies with dynamic imaging of vascular disease in foot. The results show that the temporally constrained algorithm leads to 26- fold acceleration in the image reconstruction as compared to more traditional methods that have to reconstruct all time frames data sequentially.
    Proc SPIE 03/2013;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We are presenting data from the largest clinical trial on optical tomographic imaging of finger joints to date. Overall we evaluated 99 fingers of patients affected by rheumatoid arthritis (RA) and 120 fingers from healthy volunteers. Using frequency-domain imaging techniques we show that sensitivities and specificities of 0.85 and higher can be achieved in detecting RA. This is accomplished by deriving multiple optical parameters from the optical tomographic images and combining them for the statistical analysis. Parameters derived from the scattering coefficient perform slightly better than absorption derived parameters. Furthermore we found that data obtained at 600 MHz leads to better classification results than data obtained at 0 or 300 MHz.
    IEEE transactions on medical imaging. 10/2011; 30(10):1725-36.
  • Ludguier D. Montejo, Hyun-Keol K. Kim, Andreas H. Hielscher
    [Show abstract] [Hide abstract]
    ABSTRACT: In this work we introduce the finite volume (FV) approximation to the simplified spherical harmonics (SPN) equations for modeling light propagation in tissue. The SPN equations, with partly reflective boundary conditions, are discretized on unstructured grids. The resulting system of linear equations is solved with a Krylov subspace iterative method called the generalized minimal residual (GMRES) algorithm. The accuracy of the FV-SPN algorithm is validated through numerical simulations of light propagation in a numerical phantom with embedded inhomogeneities. We use a FV implementation of the equation of radiative transfer (ERT) as the benchmark algorithm. Solutions obtained using the FV-SPN (N > 1) algorithm are compared to solutions obtained with the ERT and the diffusion equation (SP1). Compared to the SP1, the SP3 solutions obtained using the FV-SPN algorithm can better approximate ERT solutions near boundary sources and in the vicinity of void-like regions. Solutions using the SP3 algorithm are obtained 9.95 times faster than solutions with the ERT-based algorithm.
    Proc SPIE 02/2011;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We present a study on the effectiveness of computer-aided diagnosis (CAD) of rheumatoid arthritis (RA) from frequency-domain diffuse optical tomographic (FDOT) images. FDOT is used to obtain the distribution of tissue optical properties. Subsequently, the non-parametric Kruskal-Wallis ANOVA test is employed to verify statistically significant differences between the optical parameters of patients affected by RA and healthy volunteers. Furthermore, quadratic discriminate analysis (QDA) of the absorption (mua) and scattering (mua or mu's) distributions is used to classify subjects as affected or not affected by RA. We evaluate the classification efficiency by determining the sensitivity (Se), specificity (Sp), and the Youden index (Y). We find that combining features extracted from mua and mua or mu's images allows for more accurate classification than when mua or mua or mu's features are considered individually on their own. Combining mua and mua or mu's features yields values of up to Y = 0.75 (Se = 0.84 and Sp = 0.91). The best results when mua or mu's features are considered individually are Y = 0.65 (Se = 0.85 and Sp = 0.80) and Y = 0.70 (Se = 0.80 and Sp = 0.90), respectively.
    Proc SPIE 02/2011;
  • Source
    L.D. Montejo, A.D. Klose, A.H. Hielscher
    [Show abstract] [Hide abstract]
    ABSTRACT: The frequency domain (FD) radiative transfer equation (RTE) for fluorescence excitation and emission is implemented on block-structured grids. The numerical grid is adaptively refined near boundaries to accurately represent tissue of arbitrary shape using only Cartesian-type grids of varying refinement. The algorithm for the FD-RTE on block-structured grids is tested on a homogeneous numerical phantom with tissue like optical properties.
    Bioengineering Conference, Proceedings of the 2010 IEEE 36th Annual Northeast; 04/2010
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Linear discriminant analysis (LDA) and support vector machines (SVM) are used to classify reconstructed absorption coefficient distributions of the proximal interphalangeal joints as affected or not affected by rheumatoid arthritis. The performance of each classification method is quantified using the leave-n-out method. LDA is shown to yield high sensitivities, while SVM yields high specificities.
    Bioengineering Conference, Proceedings of the 2010 IEEE 36th Annual Northeast; 04/2010
  • [Show abstract] [Hide abstract]
    ABSTRACT: Using optical tomographic data from fingers affected by RA we compare the performance of 3 different classification methods. Linear discriminant and quadratic discriminant analysis methods yield high sensitivities while support-vector machine-based methods yield high specificities.
    Biomedical Optics; 04/2010
  • [Show abstract] [Hide abstract]
    ABSTRACT: Presenting data from the largest clinical trial on optical tomographic imaging of finger joints to date, we show that sensitivities and specificities better than 0.89 can be achieved, using frequency-domain techniques and advanced classification methods.
    04/2010;
  • [Show abstract] [Hide abstract]
    ABSTRACT: We solve the frequency domain equation of radiative transfer on block-structured grids (BSG) that are adaptively refined only near boundaries. We compare solutions on BSG to solutions on single finely discretized grids.
    04/2010;
  • Source
    Ludguier D Montejo, Alexander D Klose, Andreas H Hielscher
    [Show abstract] [Hide abstract]
    ABSTRACT: We present the first algorithm for solving the equation of radiative transfer (ERT) in the frequency domain (FD) on three-dimensional block-structured Cartesian grids (BSG). This algorithm allows for accurate modeling of light propagation in media of arbitrary shape with air-tissue refractive index mismatch at the boundary at increased speed compared to currently available structured grid algorithms. To accurately model arbitrarily shaped geometries the algorithm generates BSGs that are finely discretized only near physical boundaries and therefore less dense than fine grids. We discretize the FD-ERT using a combination of the upwind-step method and the discrete ordinates (S(N)) approximation. The source iteration technique is used to obtain the solution. We implement a first order interpolation scheme when traversing between coarse and fine grid regions. Effects of geometry and optical parameters on algorithm performance are evaluated using numerical phantoms (circular, cylindrical, and arbitrary shape) and varying the absorption and scattering coefficients, modulation frequency, and refractive index. The solution on a 3-level BSG is obtained up to 4.2 times faster than the solution on a single fine grid, with minimal increase in numerical error (less than 5%).
    Biomedical Optics Express 01/2010; 1(3):861-878. · 3.18 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: We developed a method for solving the fluorescence equation of radiative transfer in the frequency domain on blockstructured grids. In this way fluorescence light propagation in arbitrarily shaped tissue can be modeled with high accuracy without compromising on the convergence speed of these codes. The block-structure grid generator is developed as a multi-purpose tool that can be used with many numerical schemes. We present results from numerical studies that show that it is possible to resolve curved boundaries with grids that maintain much of the intrinsic structure of Cartesian grids. The natural ordering of this grid allows for simplified algorithms. In simulation studies we found that we can reduce the error in boundary fluence by a factor of five by using a two-level block structured grid. The increase in computational cost is only two-fold. We compare benchmark solutions to results with various levels of refinement, boundary conditions, and different geometries.
    Proc SPIE 02/2009;