Xuqing Wu

To track or to detect? an ensemble framework for optimal selection

This paper presents a novel approach for multi-target tracking using an ensemble framework that optimally chooses target tracking results from that of independent trackers and a detector at each time step. The ensemble model is designed to select the best candidate scored by a function integrating detection confidence, appearance affinity, and smoothness constraints imposed using geometry and motion information. Parameters of our association score function are discriminatively trained with a max-margin framework. Optimal selection is achieved through a hierarchical data association step that progressively associates candidates to targets. By introducing a second target classifier and using the ranking score from the pre-trained classifier as the detection confidence measure, we add additional robustness against unreliable detections. The proposed algorithm robustly tracks a large number of moving objects in complex scenes with occlusions. We evaluate our approach on a variety of public datasets and show promising improvements over state-of-the-art methods.

Embedding Topic Discovery in Conditional Random Fields Model for Segmenting Nuclei Using Multispectral Data

Segmentation of cells/nuclei is a challenging problem in 2-D histological and cytological images. Although a large number of algorithms have been proposed, newer efforts continue to be devoted to investigate robust models that could have high level of adaptability with regard to considerable amount of image variability. In this paper, we propose a multiclassification conditional random fields (CRFs) model using a combination of low-level cues (bottom-up) and high-level contextual information (top-down) for separating nuclei from the background. In our approach, the contextual information is extracted by an unsupervised topic discovery process, which efficiently helps to suppress segmentation errors caused by intensity inhomogeneity and variable chromatin texture. In addition, we propose a multilayer CRF, an extension of the traditional single-layer CRF, to handle high-dimensional dataset obtained through spectral microscopy, which provides combined benefits of spectroscopy and imaging microscopy, resulting in the ability to acquire spectral images of microscopic specimen. The approach is evaluated with color images, as well as spectral images. The overall accuracy of the proposed segmentation algorithm reaches 95% when applying multilayer CRF model to the spectral microscopy dataset. Experiments also show that our method outperforms seeded watershed, a widely used algorithm for cell segmentation.

A bottom-up and top-down model for cell segmentation using multispectral data

Cell segmentation is a challenging problem in histology and cytology that can benefit from additional information obtained in using multispectral imaging. Unique transmission spectra of biological tissues are potentially useful for better classification and segmentation of sub-cellular structures. In this paper, we propose a conditional random field (CRF) model that interprets high-dimensional spectral data during inference and pixel labeling. High quality segmentations are computed by combining low-level cues and high-level contextual information extracted by unsupervised topic discovery. Comparative analysis of the proposed model against the commonly used 2-D CRF model in color space is also performed. Results of this evaluation show the benefits of our proposed model.

Comparative analysis of cell segmentation using absorption and color images in fine needle aspiration cytology

Segmentation of cytological smears plays a critical role in the automated analysis of histological abnormalities by fine needle aspiration cytology. However, smears obtained from fine needle aspiration biopsy are often contaminated with blood. Segmentation of such an image is not a trivial task and the false positive rate could be high if the blood cells cannot be correctly separated from the rest of the sample. Moreover, the fine textured nature of the cell chromatin gives it a non-uniform intensity appearance in both color and gray images. In this paper, we propose an enhanced watershed approach to remove background noise by using short wavelength spectral image and the computed absorption image to improve segmentation accuracy. We also demonstrate a color image segmentation method by applying watershed to the minima imposed aggregation image. Results of segmentation on 20 images of cytological smears are presented and the accuracy compared for the two methods.

Level set with embedded conditional random fields and shape priors for segmentation of overlapping objects

Traditional methods for segmenting touching or overlapping objects may lead to the loss of accurate shape information which is a key descriptor in many image analysis applications. While experimental results have shown the effectiveness of using statistical shape priors to overcome such difficulties in a level set based variational framework, problems in estimation of parameters that balance evolution forces from image information and shape priors remain unsolved. In this paper, we extend the work of embedded Conditional Random Fields (CRF) by incorporating shape priors so that accurate estimation of those parameters can be obtained by the supervised training of the discrete CRF. In addition, non-parametric kernel density estimation with adaptive window size is applied as a statistical measure that locally approximates the variation of intensities to address intensity inhomogeneities. The model is tested for the problem of segmenting overlapping nuclei in cytological images.

Multispectral microscopy and cell segmentation for analysis of thyroid fine needle aspiration cytology smears

This paper discusses the needs for automated tools to aid in the diagnosis of thyroid nodules based on analysis of fine needle aspiration cytology smears. While conventional practices rely on the analysis of grey scale or RGB color images, we present a multispectral microscopy system that uses thirty-one spectral bands for analysis. Discussed are methods and results for system calibration and cell delineation.

Regularized Multi-view Multi-metric Learning for Action Recognition

Although multi-view datasets have become more accessible in the real-world applications, most state-of-the-art action recognition methods applied to those datasets rely on simple view agreement when combining local information from various views together. This leads to deteriorated performance in situations with view insufficiency and view disagreements. In this paper, we propose a novel framework for boosting action recognition performance by quantifying the connection between the viewpoint and an action. The proposed approach searches for the best combination of multiple views based on a co-learning strategy that simultaneously learns a local distance metric related to each action class and the relationships between each viewpoint and the action category. Consequently, the spatio-temporal representation of each action class in different viewpoints plays a key role in shaping the local distance metric space. We test our method on the IXMAS dataset and shows competitive performance compared to other state-of-the-art methods.

Cell segmentation in multispectral images using level sets with priors for accurate shape recovery

In this paper, we demonstrate the effectiveness of using statistical shape priors to recover shape descriptors from occluded objects in a level set based variational framework. Parameters that balance curve evolution forces are estimated systematically through embedded discrete Conditional Random Field (CRF). In addition, our approach exploits the benefit of using spectral data to construct a local appearance model for images with intensity inhomogeneity. The proposed segmentation approach is evaluated on cytological smears imaged using spectral microscopy and compared against traditional cell segmentation algorithms.

Random field model for cell segmentation in transmission mode multispectral microscopy images

Multispectral microscopy for applications in histology and cytology has shown that the unique transmission spectra of biological tissue provides additional information that is potentially useful for better classification and segmentation of sub-cellular structures. In this paper, we propose a conditional random field (CRF) model that incorporates spectral data during inference for the problem of segmenting cells in images of cytological smears. Relationship between neighboring bands is weighted by the gradient of spectral profile of the sample. Experimental results show that the proposed approach effectively suppresses the non-uniform appearance of the cell chromatin by integrating spatial and spectral constraints within the segmentation process and robustly labels fine textured cells. Comparative analysis of the proposed model against the commonly used 2-D model in color space is also performed. Results of this evaluation show the benefits of the proposed model.

Automated Prototype Generation for Multi-color Karyotyping

This chapter presents an algorithm for automatically generating a prototype from multicolor karyotypes obtained via multi-spectral imaging of human chromosomes. The single representative prototype of the color karyotype that is generated represents the analytical integration of a group of karyotypes obtained via Multicolor Fluorescence In Situ Hybridization (MFISH) method. Multicolor karyotyping is a 24-color MFISH method that allows simultaneous screening of the genome. It allows for the detection of a wide variety of anomalies in human chromosomes, including subtle and complex rearrangements. Although, multicolor karyotyping allows visual detection of gross anomalies, misclassified pixels make manual examination difficult. Additionally, in the absence of prior knowledge of the anomaly, interpretation of the karyotypes can be ambiguous. In this study we have developed an automated method for the generation of a single representative prototype of the color karyotype, which assists the screening of chromosomal aberrations by computational removal of non-physiological anomalies. We hypothesize that generation of a single representative prototype of the color karyotype from multiple karyotypes (k) for a given specimen can highlight all the aberrations, while minimizing misclassified pixels arising from inconsistencies in sample preparation, hybridization and imaging procedures. A three-tier approach is implemented to achieve the generation of the representative color karyotype from a set of multiple (>2) karyotypes. The first step involves the automated extraction of individual chromosomes from each karyotype in the set, followed by chromosome straightening and size normalization. In the second step, the extracted and normalized chromosomes belonging to each of the 24 color classes are automatically assigned to a particular group (1, 2, 3, etc.) based on the ploidy level (monoploid, diploid, triploid, etc.), respectively. For automated group assignment, Bayesian classification is utilized to determine the probability that a particular chromosome belongs to a specific group based on the similarity between the chromosomes within the group. Similarity is evaluated using two distance metrics: (1) two-dimensional (2D) histogram based descriptors, and (2) Eigen space representation based on Principal Component Analysis (PCA). Finally in the third step, we compute the prototype of the color karyotype by generating the representative chromosome for each group in the 24 color classes using pixel-based fusion. This approach allows us to generate the representative prototype color karyotype that reflects all anomalies for a given specimen, while rejecting non-physiological inconsistencies. Furthermore, automation not only reduces the workload, but also allows alleviation of subjectivity by providing a quantitative formulation based on statistical analysis.