Nicolino J. Pizzi

Unsupervised hierarchical image segmentation with level set and additive operator splitting

This paper presents an unsupervised hierarchical segmentation method for multi-phase images based on a single level set (2-phase) method and the semi-implicit additive operator splitting (AOS) scheme which is stable, fast, and easy to implement. The method successively segments image subregions found at each step of the hierarchy using a decision criterion based on the variance of intensity across the current subregion. The segmentation continues until a specified number of levels has been reached. The segmentation information for sub-images at each stage is stored in a tree data structure, and is used for reconstructing the segmented images. The method avoids the complicated governing equations of the multi-phase segmentation approach, and appears to converge in fewer iterations. The method can easily be parallelized because the AOS scheme decomposes the equations into a sequence of one dimensional systems.

Fuzzy wavelet packet based feature extraction method and its application to biomedical signal classification

In this paper, we develop an efficient fuzzy wavelet packet (WP) based feature extraction method for the classification of high-dimensional biomedical data such as magnetic resonance spectra. The key design phases involve: 1) a WP transformation mapping the original signals to many WP feature spaces and finding optimal WP decomposition for signal classification; 2) feature extraction based on the optimal WP decomposition; and 3) signal classification realized by a linear classifier. In contrast to the standard method of feature extraction used in WPs, guided by the criteria of signal compression or signal energy, our method is used to extract discriminatory features from the WP coefficients of the optimal decomposition. The extraction algorithm constructs fuzzy sets of features (via fuzzy clustering) to assess their discriminatory effectiveness. This paper includes a number of numerical experiments using magnetic resonance spectra. Classification results are compared with those obtained from common feature extraction methods in the WP domain.

Identifying core sets of discriminatory features using particle swarm optimization

Forming an efficient feature space for classification problems is a grand challenge in pattern recognition. New optimization techniques emerging in areas such as Computational Intelligence have been investigated in the context of feature selection. Here, we propose an original two-phase feature selection method that uses particle swarm optimization (PSO), a biologically inspired optimization technique, which forms an initial core set of discriminatory features from the original feature space. This core set is then successively expanded by searching for additional discriminatory features. The performance of the proposed PSO feature selection method is evaluated using a nearest neighbor classifier. The design of the optimally reduced feature space is investigated in a parametric setting by varying the size of the core feature set and the training set. Numerical experiments, using data from the Machine Learning Repository, show that a substantial reduction of the feature space is accomplished. A thorough comparative analysis of results reported in the literature also reveals improvement in classification accuracy.

Neural network classification of infrared spectra of control and Alzheimer's diseased tissue.

Artificial neural network classification methods were applied to infrared spectra of histopathologically confirmed Alzheimers diseased and control brain tissue. Principal component analysis was used as a preprocessing technique for some of these artificial neural networks while others were trained using the original spectra. The leave-one-out method was used for cross-validation and linear discriminant analysis was used as a performance benchmark. In the cases where principal components were used, the artificial neural networks consistently outperformed their linear discriminant counterparts; 100% versus 98% correct classifications, respectively, for the two class problem, and 90% versus 81% for a more complex five class problem. Using the original spectra, only one of the three selected artificial neural network architectures (a variation of the back-propagation algorithm using fuzzy encoding) produced results comparable to the best corresponding principal component cases: 98% and 85% correct classifications for the two and five class problems, respectively.

Multicamera vision-based approach to flexible feature measurement for inspection and reverse engineering

The vision-based coordinate measurement system, developed at the National Research Council Canada, is a multicamera passive system that combines the principles of stereo vision, photogrammetry, knowledge-based techniques, and an object-oriented design methodology to provide precise coordinate and dimension measurements of parts for applications such as dimensional inspection, positioning and tracking of objects, and reverse engineering. For a vision system to be considered for such applications, its performance and design parameters must be well understood. A description of the system, the techniques employed for calibration, a performance evaluation procedure, an accuracy analysis, and test results are presented.

EvIdent TM : a functional magnetic resonance image analysis system

EvIdent (EVent IDENTification) is a user-friendly, algorithm-rich, exploratory data analysis software for quickly detecting, investigating, and visualizing novel events in a set of images as they evolve in time and/or frequency. For instance, in a series of functional magnetic resonance neuroimages, novelty may manifest itself as neural activations in a time course. The core of the system is an enhanced variant of the fuzzy c-means clustering algorithm. Fuzzy clustering obviates the need for models of the underlying requisite biological function, models that are often statistically suspect.

Mapping high-dimensional data onto a relative distance plane: an exact method for visualizing and characterizing high-dimensional patterns

We introduce a distance (similarity)-based mapping for the visualization of high-dimensional patterns and their relative relationships. The mapping preserves exactly the original distances between points with respect to any two reference patterns in a special two-dimensional coordinate system, the relative distance plane (RDP). As only a single calculation of a distance matrix is required, this method is computationally efficient, an essential requirement for any exploratory data analysis. The data visualization afforded by this representation permits a rapid assessment of class pattern distributions. In particular, we can determine with a simple statistical test whether both training and validation sets of a 2-class, high-dimensional dataset derive from the same class distributions. We can explore any dataset in detail by identifying the subset of reference pairs whose members belong to different classes, cycling through this subset, and for each pair, mapping the remaining patterns. These multiple viewpoints facilitate the identification and confirmation of outliers. We demonstrate the effectiveness of this method on several complex biomedical datasets. Because of its efficiency, effectiveness, and versatility, one may use the RDP representation as an initial, data mining exploration that precedes classification by some classifier. Once final enhancements to the RDP mapping software are completed, we plan to make it freely available to researchers.

Software quality prediction using median-adjusted class labels

Software metrics aid project managers in predicting the quality of software systems. A method is proposed using a neural network classifier with metric inputs and subjective quality assessments as class labels. The labels are adjusted using fuzzy measures of the distances from each class center computed using robust multivariate medians.

Scopira - a system for the analysis of biomedical data

With the proliferation of high-dimensional biomedical data, an acute need exists for a comprehensive, user-friendly software suite that allows investigators, in the health care disciplines, to classify their data through the detection of discriminating features. Scopira is a software initiative that attempts to achieve these goals in addition to providing intuitive visual computation, logic construction and parallel execution. We describe the architecture of Scopira, and various design and implementation issues that surfaced during development.

Fuzzy pre-processing of gold standards as applied to biomedical spectra classification

Fuzzy gold standard adjustment is a novel fuzzy set theoretic pre-processing strategy that compensates for the possible imprecision of a well-established gold standard (reference test) by adjusting, if necessary, the class labels in the design set while maintaining the gold standards discriminatory power. The adjusted gold standard incorporates robust within-class centroid information. This strategy was applied to biomedical data acquired from a MR spectrometer for the purpose of classifying human brain neoplasms. It is shown that consistent improvement (10-13%) to the discriminatory power of the underlying classifier is obtained when using this pre-processing strategy.