Partitioning histopathological images: an integrated framework for supervised color-texture segmentation and cell splitting.

For quantitative analysis of histopathological images, such as the lymphoma grading systems, quantification of features is usually carried out on single cells before categorizing them by classification algorithms. To this end, we propose an integrated framework consisting of a novel supervised cell-image segmentation algorithm and a new touching-cell splitting method. For the segmentation part, we segment the cell regions from the other areas by classifying the image pixels into either cell or extra-cellular category. Instead of using pixel color intensities, the color-texture extracted at the local neighborhood of each pixel is utilized as the input to our classification algorithm. The color-texture at each pixel is extracted by local Fourier transform (LFT) from a new color space, the most discriminant color space (MDC). The MDC color space is optimized to be a linear combination of the original RGB color space so that the extracted LFT texture features in the MDC color space can achieve most discrimination in terms of classification (segmentation) performance. To speed up the texture feature extraction process, we develop an efficient LFT extraction algorithm based on image shifting and image integral. For the splitting part, given a connected component of the segmentation map, we initially differentiate whether it is a touching-cell clump or a single nontouching cell. The differentiation is mainly based on the distance between the most likely radial-symmetry center and the geometrical center of the connected component. The boundaries of touching-cell clumps are smoothed out by Fourier shape descriptor before carrying out an iterative, concave-point and radial-symmetry based splitting algorithm. To test the validity, effectiveness and efficiency of the framework, it is applied to follicular lymphoma pathological images, which exhibit complex background and extracellular texture with nonuniform illumination condition. For comparison purposes, the results of the proposed segmentation algorithm are evaluated against the outputs of superpixel, graph-cut, mean-shift, and two state-of-the-art pathological image segmentation methods using ground-truth that was established by manual segmentation of cells in the original images. Our segmentation algorithm achieves better results than the other compared methods. The results of splitting are evaluated in terms of under-splitting, over-splitting, and encroachment errors. By summing up the three types of errors, we achieve a total error rate of 5.25% per image.

Computer-aided classification of centroblast cells in follicular lymphoma.

OBJECTIVE: To distinguish centroblast cells from non-centroblast cells using a novel automated method in follicular lymphoma cases and measure its performance on cases obtained by a consensus of six pathologists. STUDY DESIGN: Geometric and color texture features were used in the training and testing of the supervised quadratic discriminant analysis classifier. The technique was trained and tested on a data set composed of 218 centroblast images and 218 non-centroblast images. Computer performance was tested by measuring sensitivity and specificity among cells classified as centroblast and non-centroblast by consensus of six board-certified hematopathologists. RESULTS: Automated classification distinguished centroblast from non-centroblast cells with a classification accuracy of 82.56% and sensitivity and specificity of 86.67% and 86.96%, respectively, when the approach was tested. CONCLUSION: The novelty of our approach is the identification of the centroblast cells with prior information and the introduction of the principal component analysis in the spectral domain to extract texture color features.

Extraction of color features in the spectral domain to recognize centroblasts in histopathology.

In this paper, we are proposing a novel automated method to recognize centroblast (CB) cells from non-centroblast (non-CB) cells for computer-assisted evaluation of follicular lymphoma tissue samples. The method is based on training and testing of a quadratic discriminant analysis (QDA) classifier. The novel aspects of this method are the identification of the CB object with prior information, and the introduction of the principal component analysis (PCA) in the spectral domain to extract color texture features. Both geometric and texture features are used to achieve the classification. Experimental results on real follicular lymphoma images demonstrate that the combined feature space improved the performance of the system significantly. The implemented method can identify centroblast cells (CB) from non-centroblast cells (non-CB) with a classification accuracy of 82.56%.

Multimodality in the superior colliculus: an information theoretic analysis.

The deep superior colliculus (DSC) integrates multisensory input and triggers an orienting movement toward the source of stimulation (target). It would seem reasonable to suppose that input of an additional modality should always increase the amount of information received by a DSC neuron concerning a target. However, of all DSC neurons studied, only about one half in the cat and one-quarter in the monkey were multimodal. The rest received only unimodal input. Multimodal DSC neurons show the properties of multisensory enhancement, in which the neural response to an input of one modality is augmented by input of another modality, and of inverse effectiveness, in which weaker unimodal responses produce a higher percentage enhancement. Previously, we demonstrated that these properties are consistent with the hypothesis that DSC neurons use Bayes' rule to compute the posterior probability that a target is present given their stochastic sensory inputs. Here we use an information theoretic analysis of our Bayesian model to show that input of an additional modality may indeed increase target information, but only if input received from the initial modality does not completely reduce uncertainty concerning the presence of a target. Unimodal DSC neurons may be those whose unimodal input fully reduces target uncertainty and therefore have no need for input of another modality.