Marina Gorunescu

Neural network analysis of dynamic sequences of EUS elastography used for the differential diagnosis of chronic pancreatitis and pancreatic cancer

BACKGROUND EUS elastography is a newly developed imaging procedure that characterizes the differences of hardness and strain between diseased and normal tissue. OBJECTIVE To assess the accuracy of real-time EUS elastography in pancreatic lesions. DESIGN Cross-sectional feasibility study. PATIENTS The study group included, in total, 68 patients with normal pancreas (N = 22), chronic pancreatitis (N = 11), pancreatic adenocarcinoma (N = 32), and pancreatic neuroendocrine tumors (N = 3). A subgroup analysis of 43 cases with focal pancreatic masses was also performed. INTERVENTIONS A postprocessing software analysis was used to examine the EUS elastography movies by calculating hue histograms of each individual image, data that were further subjected to an extended neural network analysis to differentiate benign from malignant patterns. MAIN OUTCOME MEASUREMENTS To differentiate normal pancreas, chronic pancreatitis, pancreatic cancer, and neuroendocrine tumors. RESULTS Based on a cutoff of 175 for the mean hue histogram values recorded on the region of interest, the sensitivity, specificity, and accuracy of differentiation of benign and malignant masses were 91.4%, 87.9%, and 89.7%, respectively. The positive and negative predictive values were 88.9% and 90.6%, respectively. Multilayer perceptron neural networks with both one and two hidden layers of neurons (3-layer perceptron and 4-layer perceptron) were trained to learn how to classify cases as benign or malignant, and yielded an excellent testing performance of 95% on average, together with a high training performance that equaled 97% on average. LIMITATION A lack of the surgical standard in all cases. CONCLUSIONS EUS elastography is a promising method that allows characterization and differentiation of normal pancreas, chronic pancreatitis, and pancreatic cancer. The currently developed methodology, based on artificial neural network processing of EUS elastography digitalized movies, enabled an optimal prediction of the types of pancreatic lesions. Future multicentric, randomized studies with adequate power will have to establish the clinical impact of this procedure for the differential diagnosis of focal pancreatic masses.

A machine learning approach to keystroke dynamics based user authentication

The majority of computer systems employ a login ID and password as the principal method for access security. In stand-alone situations, this level of security may be adequate, but when computers are connected to the internet, the vulnerability to a security breach is increased. In order to reduce vulnerability to attack, biometric solutions have been employed. In this paper, we investigate the use of a behavioural biometric based on keystroke dynamics. Although there are several implementations of keystroke dynamics available, their effectiveness is variable and dependent on the data sample and its acquisition methodology. The results from this study indicate that the Equal Error Rate (EER) is significantly influenced by the attribute selection process and to a lesser extent on the authentication algorithm employed. Our results also provide evidence that a Probabilistic Neural Network (PNN) can be superior in terms of reduced training time and classification accuracy when compared with a typical MLFN back-propagation trained neural network.

A Breast Cancer Diagnosis System: A Combined Approach Using Rough Sets and Probabilistic Neural Networks

In this paper, we present a medical decision support system based on a hybrid approach utilizing rough sets and a probabilistic neural network. We utilized the ability of rough sets to perform dimensionality reduction to eliminate redundant attributes from a biomedical dataset. We then utilized a probabilistic neural network to perform supervised classification. Our results indicate that rough sets were able to reduce the number of attributes in the dataset by 67% without sacrificing classification accuracy. Our classification accuracy results yielded results on the order of 93%

Intelligent decision-making for liver fibrosis stadialization based on tandem feature selection and evolutionary-driven neural network

Hepatic fibrosis represents the principal pointer to the development of liver diseases. The correct evaluation of its degree, based on both recent non-invasive procedures and machine learning models, is of current major concern. One of the latest medical imaging methodologies for assessing it is the Fibroscan, supported by biochemical and clinical examinations. Since the complex interaction between the Fibroscan stiffness indicator and the biochemical and clinical results is hard to be manually managed towards the liver fibrosis stadialization, well-performing machine learning algorithms have been proposed to support an automatic diagnosis. We propose in this paper a tandem feature selection mechanism and evolutionary-driven neural network as a computer-based support for liver fibrosis stadialization in chronic hepatitis C. A synergetic system, based on both specific statistical tools and the sensitivity analysis provided by neural networks is used for reducing the dimension of the database from twenty-five to just six attributes. An evolutionary-trained neural network is developed afterwards for the classification of the liver fibrosis stages. The tandem approach is direct and simple, resulting from embedding the feature selection system into the method structure, in order to dynamically concentrate the search only on the most relevant attributes. Experimental results and a thorough statistical analysis clearly demonstrated the efficiency of the proposed intelligent system in comparison with other machine learning techniques reported in literature.

Competitive/collaborative neural computing system for medical diagnosis in pancreatic cancer detection

: The use of computer technology to support medical decisions is now widespread and pervasive across a broad range of medical areas. Accordingly, computer-aided diagnosis has become an increasingly important area for intelligent computational systems. This paper describes a competitive/collaborative neural computing system designed to support the medical decision process using medical imaging databases. A concrete example concerning an application to support the differential diagnosis of chronic pancreatitis and pancreatic cancer is also provided.

An evolutionary computational approach to probabilistic neural network with application to hepatic cancer diagnosis

The performance of a probabilistic neural network is strongly influenced by the smoothing parameter. This paper introduces an evolutionary approach based on genetic algorithm to optimise the search of the smoothing parameter in a modified probabilistic neural network. A Java implementation is introduced and the computational results showed the viability of this hybrid approach to determine the optimum diagnosis for hepatic diseases.

Statistical Comparison of a Probabilistic Neural Network Approach in Hepatic Cancer Diagnosis

Probabilistic neural network (PNN) may provide an alternative to establish predictive algorithms for the cancer early diagnosis. We trained a PNN, using three different techniques for searching the smoothing parameter, with a database of 299 patients. This paper deals with the comparison of the prediction capabilities of different PNN approaches used to assist the diagnosis process for hepatic diseases

Length of Stay-Based Clustering Methods for Patient Grouping

Length of stay (LOS) is often used as a proxy measure of a patient’ resource consumption because of the practical difficulties of directly measuring resource consumption and the easiness of calculating LOS. Grouping patient spells according to their LOS has proved to be a challenge in health care applications due to the inherent variability in the LOS distribution. Sound methods for LOS-based patient grouping should certainly lead to a better planning of bed allocation, and patient admission and discharge. Grouping patient spells according to their LOS in a computational efficient manner is still a research issue that has not been fully addressed. For instance, grouping patient spells according to LOS intervals (e.g. 0-3 days, 4-9 days, 10-21 days etc.), has previously been defined by non-algorithmic approaches using clinical judgement, visual inspection of the LOS distribution or according to the perceived casemix. The aim of this paper is to present a novel methodology of grouping patients according to their length of stay based on fitting Gaussian mixture models to LOS observations. This method was developed as part of an innovative prediction tool that helps identify groups of patients exhibiting similar resource consumption levels as these are approximated by patient LOS. As part of evaluating the approach, we also compare it to two alternative clustering approaches, K-means and the two-step algorithm. Computational results show the superiority of this method compared to alternative clustering approaches in terms of its ability to extract clinically meaningful patient groups as applied to a skewed LOS dataset.

A statistical evaluation of neural computing approaches to predict recurrent events in breast cancer

Breast cancer is considered to be the second leading cause of cancer deaths in women today. Sometimes, breast cancer can return after primary treatment. A medical diagnosis of recurrent cancer is often more challenging task than the initial one. In this paper we investigate the potential contribution of intelligent neural networks as a useful tool to support health professionals in diagnosing such events. The neural network algorithms are applied to the breast cancer dataset obtained from Ljubljana Oncology Institute. An extensive statistical analysis has been performed to verify our experiments. The results show that a simple network structure for both the multi-layer perception and radial basis function can produce equally good results, not all attributes are needed to train these algorithms and finally, the classification performances of both algorithms are statistically robust.

A statistical framework for evaluating neural networks to predict recurrent events in breast cancer

Breast cancer is the second leading cause of cancer deaths in women today. Sometimes, breast cancer can return after primary treatment. A medical diagnosis of recurrent cancer is often a more challenging task than the initial one. In this paper, we investigate the potential contribution of neural networks (NNs) to support health professionals in diagnosing such events. The NN algorithms are tested and applied to two different datasets. An extensive statistical analysis has been performed to verify our experiments. The results show that a simple network structure for both the multi-layer perceptron and radial basis function can produce equally good results, not all attributes are needed to train these algorithms and, finally, the classification performances of all algorithms are statistically robust. Moreover, we have shown that the best performing algorithm will strongly depend on the features of the datasets, and hence, there is not necessarily a single best classifier.