Antoniya Georgieva

Global optimization based on novel heuristics, low-discrepancy sequences and genetic algorithms

In this paper a new heuristic hybrid technique for bound-constrained global optimization is proposed. We developed iterative algorithm called GLP[tau]S that uses genetic algorithms, LP[tau] low-discrepancy sequences of points and heuristic rules to find regions of attraction when searching a global minimum of an objective function. Subsequently Nelder-Mead Simplex local search technique is used to refine the solution. The combination of the three techniques (Genetic algorithms, LP[tau]O Low-discrepancy search and Simplex search) provides a powerful hybrid heuristic optimization method which is tested on a number of benchmark multimodal functions with 10-150 dimensions, and the method properties - applicability, convergence, consistency and stability are discussed in detail.

Artificial neural networks applied to fetal monitoring in labour

Birth asphyxia can result in death or permanent brain damage. To prevent it, the fetal heart rate (FHR) is recorded in labour on a paper strip. In clinical practice, the complicated FHR patterns are assessed by eye, which is error-prone, inconsistent and unreliable. Objective alternatives are needed and thus we investigated the applicability of feed-forward artificial neural networks (ANNs) for FHR analysis. Six FHR features were extracted and combined with six clinical parameters to form a feature space of 12 dimensions. The feature space was reduced to six dimensions by principal component analysis. Subsequently, a network committee of ten ANNs was trained with the data of 124 patients (a balanced set of 62 adverse, coded 1, and 62 normal outcomes, coded 0). The ANN committee was tested on another balanced set of 252 patients obtaining misclassification rate of 36%. Finally, the committee was tested on a large dataset of 7,568 patients (non-balanced). As the committee output continuously increased from 0 to 1, there was a consistent growth of the adverse outcome rate (from 0.26 to 5.3%) and the low umbilical pH rate (from 2.6 to 16.7%.) Based on this correlation between the committee output and the risk of compromise, we concluded that ANNs can be successfully applied to FHR monitoring in labour. However, extensive further work is necessary, for which we outline our plans. To our knowledge, this is the first time that an automated method for FHR diagnostic analysis has been tested on a database of this size.

Phase-rectified signal averaging for intrapartum electronic fetal heart rate monitoring is related to acidaemia at birth.

Recent studies suggest that phase‐rectified signal averaging (PRSA), measured in antepartum fetal heart rate (FHR) traces, may sensitively indicate fetal status; however, its value has not been assessed during labour. We determined whether PRSA relates to acidaemia in labour, and compare its performance to short‐term variation (STV), a related computerised FHR feature.

Neural Network Learning With Global Heuristic Search

A novel hybrid global optimization (GO) algorithm applied for feedforward neural networks (NNs) supervised learning is investigated. The network weights are determined by minimizing the traditional mean square error function. The optimization technique, called LPtau NM, combines a novel global heuristic search based on LPtau low-discrepancy sequences of points, and a simplex local search. The proposed method is initially tested on multimodal mathematical functions and subsequently applied for training moderate size NNs for solving popular benchmark problems. Finally, the results are analyzed, discussed, and compared with such as from backpropagation (BP) (Levenberg-Marquardt) and differential evolution methods

A hybrid meta-heuristic for global optimisation using low-discrepancy sequences of points

A hybrid novel meta-heuristic technique for bound-constrained global optimisation (GO) is proposed in this paper. We have developed an iterative algorithm called LP@tOptimisation(LP@tO) that uses low-discrepancy sequences of points and meta-heuristic knowledge to find regions of attraction when searching for a global minimum of an objective function. Subsequently, the well-known Nelder-Mead (NM) simplex local search is used to refine the solution found by the LP@tO method. The combination of the two techniques (LP@tO and NM) provides a powerful hybrid optimisation technique, which we call LP@tNM. Its properties-applicability, convergence, consistency and stability are discussed here in detail. The LP@tNM is tested on a number of benchmark multimodal mathematical functions from 2 to 20 dimensions and compared with results from other stochastic heuristic methods.

Feature selection using genetic algorithms for fetal heart rate analysis

The fetal heart rate (FHR) is monitored on a paper strip (cardiotocogram) during labour to assess fetal health. If necessary, clinicians can intervene and assist with a prompt delivery of the baby. Data-driven computerized FHR analysis could help clinicians in the decision-making process. However, selecting the best computerized FHR features that relate to labour outcome is a pressing research problem. The objective of this study is to apply genetic algorithms (GA) as a feature selection method to select the best feature subset from 64 FHR features and to integrate these best features to recognize unfavourable FHR patterns. The GA was trained on 404 cases and tested on 106 cases (both balanced datasets) using three classifiers, respectively. Regularization methods and backward selection were used to optimize the GA. Reasonable classification performance is shown on the testing set for the best feature subset (Cohens kappa values of 0.45 to 0.49 using different classifiers). This is, to our knowledge, the first time that a feature selection method for FHR analysis has been developed on a database of this size. This study indicates that different FHR features, when integrated, can show good performance in predicting labour outcome. It also gives the importance of each feature, which will be a valuable reference point for further studies.

Self-organizing maps for texture classification

A further investigation of our intelligent machine vision system for pattern recognition and texture image classification is discussed in this paper. A data set of 335 texture images is to be classified into several classes, based on their texture similarities, while no a priori human vision expert knowledge about the classes is available. Hence, unsupervised learning and self-organizing maps (SOM) neural networks are used for solving the classification problem. Nevertheless, in some of the experiments, a supervised texture analysis method is also considered for comparison purposes. Four major experiments are conducted: in the first one, classifiers are trained using all the extracted features without any statistical preprocessing; in the second simulation, the available features are normalized before being fed to a classifier; in the third experiment, the trained classifiers use linear transformations of the original features, received after preprocessing with principal component analysis; and in the last one, transforms of the features obtained after applying linear discriminant analysis are used. During the simulation, each test is performed 50 times implementing the proposed algorithm. Results from the employed unsupervised learning, after training, testing, and validation of the SOMs, are analyzed and critically compared with results from other authors.

Computerized fetal heart rate analysis in labor: detection of intervals with un-assignable baseline

The fetal heart rate (FHR) is monitored during labor to assess fetal health. Both visual and computerized interpretations of the FHR depend on assigning a baseline to detect key features such as accelerations or decelerations. However, it is sometimes impossible to assign a baseline reliably, by eye or by numerical methods. To address this issue, we used the Oxford Intrapartum FHR Database to derive an algorithm based on the distribution of the FHR that detects heart rate intervals without a clear baseline. We aimed to recognize when a fetus cannot maintain its heart rate baseline and use this to assist computerized FHR analysis. Twenty-three FHR windows (15 min long) were used to develop the method. The algorithm was then validated by comparison with experts who classified 50 FHR windows into two groups: baseline assignable or un-assignable. The average agreement between experts (κ = 0.76) was comparable to the agreement between method and experts (κ = 0.67). The algorithm was used in 22 559 patients with intrapartum FHR records to retrospectively determine the incidence of intervals (defined as 15 min windows) that had un-assignable baselines. Sixty-six percent had one or more such episodes at some stage, most commonly after the onset of pushing (55%) and least commonly pre-labor (16%). These episodes are therefore relatively common. Their detection should improve the reliability of computerized analysis and allow further studies of what they signify clinically.

Supervised Neural Network Training with a Hybrid Global Optimization Technique

A novel hybrid global optimization method applied for feedforward neural networks (NN) supervised learning is investigated. The network weights are determined by minimizing the traditional mean-square error function. The optimization technique, called GLPtauS is a combination of novel global optimization heuristic search based on low-discrepancy sequences of points, called LPtau Optimization (LPtauO), a Genetic Algorithm, and a Simplex local search. The proposed method is initially tested on 10 multimodal mathematical functions of 30 and 100 dimensions. Subsequently, it is applied for training moderate size NN for function fitting and solving benchmark classification problems, such as the parity problem (XOR and 4-Parity), Iris dataset, and a medical diagnosis problem (Diabetes). The investigated technique is also tested on predicting continuous output of a mechanical system dataset (Servo). Finally, the results are analysed, discussed, and compared with others.

Computerized intrapartum electronic fetal monitoring: Analysis of the decision to deliver for fetal distress

We applied computerized methods to assess the Electronic Fetal Monitoring (EFM) in labor. We analyzed retrospectively the last hour of EFM for 1,370 babies, delivered by emergency Cesarean sections before the onset of pushing (data collected at the John Radcliffe Hospital, Oxford, UK). There were two cohorts according to the reason for intervention: (a) fetal distress, n1 = 524 and (b) failure to progress and/or malpresentation, n2 = 846. The cohorts were compared in terms of classical EFM features (baseline, decelerations, variability and accelerations), computed by a dedicated Oxford system for automated analysis — OxSys. In addition, OxSys was employed to simulate current clinical guidelines for the classification of fetal monitoring, i.e. providing in real time a three-tier grading system of the EFM (normal, indeterminate, or abnormal). The computerized features and the simulated guidelines corresponded well to the clinical management and to the actual labor outcome (measured by umbilical arterial pH).