Rapid Prototyping for Functional Electrical Stimulation Control

The clinical set-up tool integrates fast parameter selection and a user-friendly interface to help electrical muscle stimulators more efficiently treat patients with neurological injuries. A key challenge in increasing functional electrical stimulation systems' clinical acceptance is facilitating or automating parameter selection, optimization, and programming to make the underlying engineering transparent to the user. To this end, we present the clinical set-up tool (CST), a finite-state-machine-based controller that integrates accurate, automatic parameter optimization in an intuitive user interface. Unlike other approaches, we employ a numerical algorithm that uses real-life data and well-defined criteria to rapidly optimize parameter values.     

Continuation methods for approximate large scale object sequencing

Machine Learning - We propose a set of highly scalable algorithms for the combinatorial data analysis problem of seriating similarity matrices. Seriation consists of finding a permutation of data...     

Automated design of robust discriminant analysis classifier for foot pressure lesions using kinematic data

In the recent years, the use of motion tracking systems for acquisition of functional biomechanical gait data, has received increasing interest due to the richness and accuracy of the measured kinematic information. However, costs frequently restrict the number of subjects employed, and this makes the dimensionality of the collected data far higher than the available samples. This paper applies discriminant analysis algorithms to the classification of patients with different types of foot lesions, in order to establish an association between foot motion and lesion formation. With primary attention to small sample size situations, we compare different types of Bayesian classifiers and evaluate their performance with various dimensionality reduction techniques for feature extraction, as well as search methods for selection of raw kinematic variables. Finally, we propose a novel integrated method which fine-tunes the classifier parameters and selects the most relevant kinematic variables simultaneously. Performance comparisons are using robust resampling techniques such as Bootstrap 632+ and k-fold cross-validation. Results from experimentations with lesion subjects suffering from pathological plantar hyperkeratosis, show that the proposed method can lead to approximately 96% correct classification rates with less than 10% of the original features.     

A collective-based adaptive symbiotic model for surface reconstruction in area-based stereo

This paper proposes a novel optimization algorithm for image-space matching and three-dimensional space analysis, using an adapted scheme of evolutionary computation that employs the concept of symbiosis in a collective of homogeneous populations. It is applied to the automatic generation of disparity surfaces used for depth estimation in stereo vision. The global task of approximating the complete disparity surface is decomposed to a large number of smaller local problems, each solvable by a smaller processing unit. Coevolution is sustained in such a way as to counteract the arbitrary decomposition of the original super-problem, so that the local evolutions of all the subproblems become interlocked. This, in the long run, provides a consistent global solution, and it does so via an asynchronous and massively parallel architecture. The entire surface is partitioned to a set of adjoining patches represented by distinct species or populations, with phenotypes corresponding to different polynomial functionals. The credit assignment functions take into account both self and symbiotic terms in an adaptive and dynamic manner, in order to produce disparity patches that are fit within their own domain and at the same time fit in association with their symbionts. This persistent propagation of local interactions to a global scale throughout evolution generates a unified disparity surface composed of the many smaller patch surfaces.     

Spatiotemporal Visualizations for the Measurement of Oropharyngeal Transit Time From Videofluoroscopy

Videofluoroscopy remains one of the mainstay methods for clinical swallowing assessment, yet its interpretation is both complex and subjective. This, in part, reflects the difficulties associated with estimation of bolus transit time through the oral and pharyngeal regions by visual inspection, and problems with consistent repeatability. This paper introduces a software-only framework that automatically determines the time taken for the bolus to cross 1-D anatomical landmarks representing the oral and pharyngeal region boundaries ( Fig. 1). The user-steered delineation algorithm live-wire and straight-line annotators are used to demarcate the landmark on a frame prior to the swallow action. The rate of change of intensity of the pixels in each landmark is used as the detection feature for bolus presence that can be visualized on a spatiotemporal plot. Artifacts introduced by head and neck movement are removed by updating the landmark coordinates using affine parameters optimized by a genetic-algorithm-based registration method. Heuristics are applied to the spatiotemporal plot to identify the frames during which the bolus passes the landmark. Correlation coefficients between three observers visually inspecting twenty-four 5-mL single swallow clips did not exceed 0.42. Yet the same measurements taken using this framework on the same clips had correlation coefficients exceeding 0.87.      

Hybrid symbiotic genetic optimisation for robust edge-based stereo correspondence.

This work proposes a novel algorithm for performing robust feature-based stereo-matching, without the ordering constraint. The calculation of the disparity map is decomposed to a set of disjoint intra-row subproblems, each one having two objectives: the search for a high confidence intra-row matching and the enforcement of figural continuity at the inter-row level. A separate genetic algorithm (GA) is allocated at each epipolar to search the feasible solution space. All GAs evolve parallely in a symbiotic fashion and continuously exchange currently available solution information to enable optimisation of figural continuity. To accelerate the search, we adapt a deterministic solver to seed the GAs and design problem-specific genetic operators for greater efficiency.     

Density-driven generalized regression neural networks (DD-GRNN) for function approximation.

This paper proposes a new nonparametric regression method, based on the combination of generalized regression neural networks (GRNNs), density-dependent multiple kernel bandwidths, and regularization. The presented model is generic and substitutes the very large number of bandwidths with a much smaller number of trainable weights that control the regression model. It depends on sets of extracted data density features which reflect the density properties and distribution irregularities of the training data sets. We provide an efficient initialization scheme and a second-order algorithm to train the model, as well as an overfitting control mechanism based on Bayesian regularization. Numerical results show that the proposed network manages to reduce significantly the computational demands of having individual bandwidths, while at the same time, provides competitive function approximation accuracy in relation to existing methods.     

Gait recognition: highly unique dynamic plantar pressure patterns among 104 individuals

Everyone''s walking style is unique, and it has been shown that both humans and computers are very good at recognizing known gait patterns. It is therefore unsurprising that dynamic foot pressure patterns, which indirectly reflect the accelerations of all body parts, are also unique, and that previous studies have achieved moderate-to-high classification rates (CRs) using foot pressure variables. However, these studies are limited by small sample sizes (n < 30), moderate CRs (CR ≃ 90%), or both. Here we show, using relatively simple image processing and feature extraction, that dynamic foot pressures can be used to identify n = 104 subjects with a CR of 99.6 per cent. Our key innovation was improved and automated spatial alignment which, by itself, improved CR to over 98 per cent, a finding that pointedly emphasizes inter-subject pressure pattern uniqueness. We also found that automated dimensionality reduction invariably improved CRs. As dynamic pressure data are immediately usable, with little or no pre-processing required, and as they may be collected discreetly during uninterrupted gait using in-floor systems, foot pressure-based identification appears to have wide potential for both the security and health industries.     

Incorporating Gradient Estimations in a Circle-Finding Probabilistic Hough Transform

In this paper we present a novel Probabilistic Hough Transform algorithm to detect circles. While other Probabilistic Hough Transforms reduce the generation of redundant evidence by sampling point-triples, the proposed algorithm achieves a much higher reduction in two ways. First, by using the gradient information, it allows point-pairs to define circles, and consequently decreases the sampling complexity from O(N 3 )to O(N 2 ). Secondly, the transformation is conditional, i.e. not all the pairs are eligible to vote. The evidence is gathered in a very sparse parameter space, so that peak recovery is performed readily. The result is high speed, increased accuracy and very low memory resources. Illustrative examples demonstrate the detection accuracy of the algorithm. Received: 20 April 1998?Received in revised form: 27 January 1999?Accepted: 9 February 1999     

Towards collaborative feature extraction for face recognition

Principal components analysis has become a popular preprocessing method to avoid the small sample size problem for most of the supervised graph embedding methods. Nevertheless, there is potential loss of relevant information when projecting the data onto the space defined by the principal Eigenfaces when the number of individuals in the gallery is large. This paper introduces a new collaborative feature extraction method based on projection pursuit, as a robust preprocessing for supervised embedding methods. A previously proposed projection index was adopted as a measure of interestingness, based on a weighted sum of six state of the art indices. We compare our collaborative feature extraction technique against principal component analysis as preprocessing stage for Laplacianfaces. For completeness, results for Eigenfaces and Fisherfaces are included. Experimental results to demonstrate the robustness of our approach against changes in facial expression and lighting are presented.