Chaoyang Zhang

Fluorescence-enhanced optical imaging in large tissue volumes using a gain-modulated ICCD camera

A novel image-intensified charge-coupled device (ICCD) imaging system has been developed to perform 3D fluorescence tomographic imaging in the frequency-domain using near-infrared contrast agents. The imager is unique since it (i) employs a large tissue-mimicking phantom, which is shaped and sized to resemble a female breast and part of the extended chest-wall region, and (ii) enables rapid data acquisition in the frequency-domain by using a gain-modulated ICCD camera. Diffusion model predictions are compared to experimental measurements using two different referencing schemes under two different experimental conditions of perfect and imperfect uptake of fluorescent agent into a target. From these experimental measurements, three-dimensional images of fluorescent absorption were reconstructed using a computationally efficient variant of the approximate extended Kalman filter algorithm. The current work represents the first time that 3D fluorescence-enhanced optical tomographic reconstructions have been achieved from experimental measurements of the time-dependent light propagation on a clinically relevant breast-shaped tissue phantom using a gain-modulated ICCD camera.

Diagnostic imaging of breast cancer using fluorescence-enhanced optical tomography: phantom studies

Molecular targeting with exogenous near-infrared excitable fluorescent agents using time-dependent imaging techniques may enable diagnostic imaging of breast cancer and prognostic imaging of sentinel lymph nodes within the breast. However, prior to the administration of unproven contrast agents, phantom studies on clinically relevant volumes are essential to assess the benefits of fluorescence-enhanced optical imaging in humans. Diagnostic 3-D fluorescence-enhanced optical tomography is demonstrated using 0.5 to 1 cm(3) single and multiple targets differentiated from their surroundings by indocyanine green (micromolar) in a breast-shaped phantom (10-cm diameter). Fluorescence measurements of referenced ac intensity and phase shift were acquired in response to point illumination measurement geometry using a homodyned intensified charge-coupled device system modulated at 100 MHz. Bayesian reconstructions show artifact-free 3-D images (3857 unknowns) from 3-D boundary surface measurements (126 to 439). In a reflectance geometry appropriate for prognostic imaging of lymph node involvement, fluorescence measurements were likewise acquired from the surface of a semi-infinite phantom (8x8x8 cm(3)) in response to area illumination (12 cm(2)) by excitation light. Tomographic 3-D reconstructions (24,123 unknowns) were recovered from 2-D boundary surface measurements (3194) using the modified truncated Newtons method. These studies represent the first 3-D tomographic images from physiologically relevant geometries for breast imaging.

Comparison of probabilistic Boolean network and dynamic Bayesian network approaches for inferring gene regulatory networks

BackgroundThe regulation of gene expression is achieved through gene regulatory networks (GRNs) in which collections of genes interact with one another and other substances in a cell. In order to understand the underlying function of organisms, it is necessary to study the behavior of genes in a gene regulatory network context. Several computational approaches are available for modeling gene regulatory networks with different datasets. In order to optimize modeling of GRN, these approaches must be compared and evaluated in terms of accuracy and efficiency.ResultsIn this paper, two important computational approaches for modeling gene regulatory networks, probabilistic Boolean network methods and dynamic Bayesian network methods, are compared using a biological time-series dataset from the Drosophila Interaction Database to construct a Drosophila gene network. A subset of time points and gene samples from the whole dataset is used to evaluate the performance of these two approaches.ConclusionThe comparison indicates that both approaches had good performance in modeling the gene regulatory networks. The accuracy in terms of recall and precision can be improved if a smaller subset of genes is selected for inferring GRNs. The accuracy of both approaches is dependent upon the number of selected genes and time points of gene samples. In all tested cases, DBN identified more gene interactions and gave better recall than PBN.

Fluorescence-enhanced optical imaging of large phantoms using single and simultaneous dual point illumination geometries

Fluorescence-enhanced optical tomography is typically performed using single point illumination and multiple point collection measurement geometry. Single point illumination is often insufficient to illuminate greater volumes of large phantoms and results in an inadequate fluorescent signal to noise ratio (SNR) for the majority of measurements. In this work, the use of simultaneous multiple point illumination geometry is proposed for acquiring a large number of fluorescent measurements with a sufficiently high SNR. As a feasibility study, dual point excitation sources, which are in-phase, were used in order to acquire surface measurements and perform three-dimensional reconstructions on phantoms of large volume and/or significant penetration depth. Measurements were acquired in the frequency-domain using a modulated intensified CCD imaging system under different experimental conditions of target depth (1.4-2.8 cm deep) with a perfect uptake optical contrast. Three-dimensional reconstructions of the fluorescence absorption from the dual point illumination geometry compare well with the reconstructions from the single point illumination geometry. Targets located up to 2 cm deep were located successfully, establishing the feasibility of reconstructions from simultaneous multiple point excitation sources. With improved excitation light rejection, multiple point illumination geometry may prove useful in reconstructing more challenging domains containing deeply embedded targets. Image quality assessment tools are required to determine the optimal measurement geometry for the largest set off imaging tasks.

A novel gene network inference algorithm using predictive minimum description length approach

BackgroundReverse engineering of gene regulatory networks using information theory models has received much attention due to its simplicity, low computational cost, and capability of inferring large networks. One of the major problems with information theory models is to determine the threshold which defines the regulatory relationships between genes. The minimum description length (MDL) principle has been implemented to overcome this problem. The description length of the MDL principle is the sum of model length and data encoding length. A user-specified fine tuning parameter is used as control mechanism between model and data encoding, but it is difficult to find the optimal parameter. In this work, we proposed a new inference algorithm which incorporated mutual information (MI), conditional mutual information (CMI) and predictive minimum description length (PMDL) principle to infer gene regulatory networks from DNA microarray data. In this algorithm, the information theoretic quantities MI and CMI determine the regulatory relationships between genes and the PMDL principle method attempts to determine the best MI threshold without the need of a user-specified fine tuning parameter.ResultsThe performance of the proposed algorithm was evaluated using both synthetic time series data sets and a biological time series data set for the yeast Saccharomyces cerevisiae. The benchmark quantities precision and recall were used as performance measures. The results show that the proposed algorithm produced less false edges and significantly improved the precision, as compared to the existing algorithm. For further analysis the performance of the algorithms was observed over different sizes of data.ConclusionsWe have proposed a new algorithm that implements the PMDL principle for inferring gene regulatory networks from time series DNA microarray data that eliminates the need of a fine tuning parameter. The evaluation results obtained from both synthetic and actual biological data sets show that the PMDL principle is effective in determining the MI threshold and the developed algorithm improves precision of gene regulatory network inference. Based on the sensitivity analysis of all tested cases, an optimal CMI threshold value has been identified. Finally it was observed that the performance of the algorithms saturates at a certain threshold of data size.

A comparison of exact and approximate adjoint sensitivities in fluorescence tomography

Many approaches to fluorescence tomography utilize some form of regularized nonlinear least-squares algorithm for data inversion, thus requiring repeated computation of the Jacobian sensitivity matrix relating changes in observable quantities, such as emission fluence, to changes in underlying optical parameters, such as fluorescence absorption. An exact adjoint formulation of these sensitivities comprises three terms, reflecting the individual contributions of 1) sensitivities of diffusion and decay coefficients at the emission wavelength, 2) sensitivities of diffusion and decay coefficients at the excitation wavelength, and 3) sensitivity of the emission source term. Simplifying linearity assumptions are computationally attractive in that they cause the first and second terms to drop out of the formulation. The relative importance of the three terms is thus explored in order to determine the extent to which these approximations introduce error. Computational experiments show that, while the third term of the sensitivity matrix has the largest magnitude, the second term becomes increasingly significant as target fluorophore concentration or volume increases. Image reconstructions from experimental data confirm that neglecting the second term results in overestimation of sensitivities and consequently overestimation of the value and volume of the fluorescent target, whereas contributions of the first term are so low that they are probably not worth the additional computational costs.

Supervised learning method for the prediction of subcellular localization of proteins using amino acid and amino acid pair composition

BackgroundOccurrence of protein in the cell is an important step in understanding its function. It is highly desirable to predict a proteins subcellular locations automatically from its sequence. Most studied methods for prediction of subcellular localization of proteins are signal peptides, the location by sequence homology, and the correlation between the total amino acid compositions of proteins. Taking amino-acid composition and amino acid pair composition into consideration helps improving the prediction accuracy.ResultsWe constructed a dataset of protein sequences from SWISS-PROT database and segmented them into 12 classes based on their subcellular locations. SVM modules were trained to predict the subcellular location based on amino acid composition and amino acid pair composition. Results were calculated after 10-fold cross validation. Radial Basis Function (RBF) outperformed polynomial and linear kernel functions. Total prediction accuracy reached to 71.8% for amino acid composition and 77.0% for amino acid pair composition. In order to observe the impact of number of subcellular locations we constructed two more datasets of nine and five subcellular locations. Total accuracy was further improved to 79.9% and 85.66%.ConclusionsA new SVM based approach is presented based on amino acid and amino acid pair composition. Result shows that data simulation and taking more protein features into consideration improves the accuracy to a great extent. It was also noticed that the data set needs to be crafted to take account of the distribution of data in all the classes.

Time lagged information theoretic approaches to the reverse engineering of gene regulatory networks.

BackgroundA number of models and algorithms have been proposed in the past for gene regulatory network (GRN) inference; however, none of them address the effects of the size of time-series microarray expression data in terms of the number of time-points. In this paper, we study this problem by analyzing the behaviour of three algorithms based on information theory and dynamic Bayesian network (DBN) models. These algorithms were implemented on different sizes of data generated by synthetic networks. Experiments show that the inference accuracy of these algorithms reaches a saturation point after a specific data size brought about by a saturation in the pair-wise mutual information (MI) metric; hence there is a theoretical limit on the inference accuracy of information theory based schemes that depends on the number of time points of micro-array data used to infer GRNs. This illustrates the fact that MI might not be the best metric to use for GRN inference algorithms. To circumvent the limitations of the MI metric, we introduce a new method of computing time lags between any pair of genes and present the pair-wise time lagged Mutual Information (TLMI) and time lagged Conditional Mutual Information (TLCMI) metrics. Next we use these new metrics to propose novel GRN inference schemes which provides higher inference accuracy based on the precision and recall parameters.ResultsIt was observed that beyond a certain number of time-points (i.e., a specific size) of micro-array data, the performance of the algorithms measured in terms of the recall-to-precision ratio saturated due to the saturation in the calculated pair-wise MI metric with increasing data size. The proposed algorithms were compared to existing approaches on four different biological networks. The resulting networks were evaluated based on the benchmark precision and recall metrics and the results favour our approach.ConclusionsTo alleviate the effects of data size on information theory based GRN inference algorithms, novel time lag based information theoretic approaches to infer gene regulatory networks have been proposed. The results show that the time lags of regulatory effects between any pair of genes play an important role in GRN inference schemes.

Identification and Optimization of Classifier Genes from Multi-Class Earthworm Microarray Dataset

Monitoring, assessment and prediction of environmental risks that chemicals pose demand rapid and accurate diagnostic assays. A variety of toxicological effects have been associated with explosive compounds TNT and RDX. One important goal of microarray experiments is to discover novel biomarkers for toxicity evaluation. We have developed an earthworm microarray containing 15,208 unique oligo probes and have used it to profile gene expression in 248 earthworms exposed to TNT, RDX or neither. We assembled a new machine learning pipeline consisting of several well-established feature filtering/selection and classification techniques to analyze the 248-array dataset in order to construct classifier models that can separate earthworm samples into three groups: control, TNT-treated, and RDX-treated. First, a total of 869 genes differentially expressed in response to TNT or RDX exposure were identified using a univariate statistical algorithm of class comparison. Then, decision tree-based algorithms were applied to select a subset of 354 classifier genes, which were ranked by their overall weight of significance. A multiclass support vector machine (MC-SVM) method and an unsupervised K-mean clustering method were applied to independently refine the classifier, producing a smaller subset of 39 and 30 classifier genes, separately, with 11 common genes being potential biomarkers. The combined 58 genes were considered the refined subset and used to build MC-SVM and clustering models with classification accuracy of 83.5% and 56.9%, respectively. This study demonstrates that the machine learning approach can be used to identify and optimize a small subset of classifier/biomarker genes from high dimensional datasets and generate classification models of acceptable precision for multiple classes.

A Fourier Transformation based Method to Mine Peptide Space for Antimicrobial Activity

BackgroundNaturally occurring antimicrobial peptides are currently being explored as potential candidate peptide drugs. Since antimicrobial peptides are part of the innate immune system of every living organism, it is possible to discover new candidate peptides using the available genomic and proteomic data. High throughput computational techniques could also be used to virtually scan the entire peptide space for discovering out new candidate antimicrobial peptides.ResultWe have identified a unique indexing method based on biologically distinct characteristic features of known antimicrobial peptides. Analysis of the entries in the antimicrobial peptide databases, based on our indexing method, using Fourier transformation technique revealed a distinct peak in their power spectrum. We have developed a method to mine the genomic and proteomic data, for the presence of peptides with potential antimicrobial activity, by looking for this distinct peak. We also used the Euclidean metric to rank the potential antimicrobial peptides activity. We have parallelized our method so that virtually any given protein space could be data mined, in search of antimicrobial peptides.ConclusionThe results show that the Fourier transform based method with the property based coding strategy could be used to scan the peptide space for discovering new potential antimicrobial peptides.