Improving diagnostic accuracy using a hierarchical neural network to model decision subtasks

Combined neural network model employing lyapunov exponents: internal carotid arterial disorders detection case

This paper illustrates the use of combined neural network models to guide model selection for diagnosis of internal carotid arterial disorders. The method presented in this study was directly based on the consideration that internal carotid arterial Doppler signals are chaotic signals. This consideration was tested successfully using the nonlinear dynamics tools, like the computation of Lyapunov exponents. Statistics were used over the set of the Lyapunov exponents in order to reduce the dimensionality of the extracted feature vectors. The first level networks were implemented for the diagnosis of internal carotid arterial disorders using the selected Lyapunov exponents as inputs. To improve diagnostic accuracy, the second level network was trained using the outputs of the first level networks as input data. The combined neural network models achieved accuracy rates which were higher than that of the stand-alone neural network models.

Analysis of doppler ultrasound signals: ophthalmic arterial disorders detection case

The paper includes illustrative and detailed information about implementation of automated diagnostic systems and feature extraction/selection from signals recorded from ophthalmic arteries. The major objective of the paper is to be a guide for the readers, who want to develop an automated diagnostic systems for detection of arterial disorders. Toward achieving this objective, this paper present the techniques which should be considered in developing automated diagnostic systems. The paper will assist to the people in gaining a better understanding of the techniques in the detection of arterial disorders.

Combining neural network models for automated diagnostic systems

This paper illustrates the use of combined neural network (CNN) models to guide model selection for diagnosis of internal carotid arterial (ICA) disorders. The ICA Doppler signals were decomposed into time-frequency representations using discrete wavelet transform and statistical features were calculated to depict their distribution. The first level networks were implemented for the diagnosis of ICA disorders using the statistical features as inputs. To improve diagnostic accuracy, the second level network was trained using the outputs of the first level networks as input data. The CNN models achieved accuracy rates which were higher than that of the stand-alone neural network models.

Automated diagnostic systems with diverse and composite features for Doppler ultrasound signals

In this paper, we present the automated diagnostic systems for Doppler ultrasound signals classification with diverse and composite features and determine their accuracies. We compared the classification accuracies of six different classifiers, namely multilayer perceptron neural network (MLP), combined neural network (CNN), mixture of experts (ME), modified mixture of experts (MME), probabilistic neural network (PNN), and support vector machine (SVM), which were trained on diverse or composite features. The present study was conducted with the purpose of answering the question of whether the automated diagnostic systems improve the capability of classification of ophthalmic arterial (OA) and internal carotid arterial (ICA) Doppler signals. Our research demonstrated that the SVM trained on composite feature and the MME trained on diverse features achieved accuracy rates which were higher than that of the other automated diagnostic systems.

Feature extraction from Doppler ultrasound signals for automated diagnostic systems

This paper presented the assessment of feature extraction methods used in automated diagnosis of arterial diseases. Since classification is more accurate when the pattern is simplified through representation by important features, feature extraction and selection play an important role in classifying systems such as neural networks. Different feature extraction methods were used to obtain feature vectors from ophthalmic and internal carotid arterial Doppler signals. In addition to this, the problem of selecting relevant features among the features available for the purpose of classification of Doppler signals was dealt with. Multilayer perceptron neural networks (MLPNNs) with different inputs (feature vectors) were used for diagnosis of ophthalmic and internal carotid arterial diseases. The assessment of feature extraction methods was performed by taking into consideration of performances of the MLPNNs. The performances of the MLPNNs were evaluated by the convergence rates (number of training epochs) and the total classification accuracies. Finally, some conclusions were drawn concerning the efficiency of discrete wavelet transform as a feature extraction method used for the diagnosis of ophthalmic and internal carotid arterial diseases.

Improving the classification of multiple disorders with problem decomposition

Differential diagnosis of multiple disorders is a challenging problem in clinical medicine. According to the divide-and-conquer principle, this problem can be handled more effectively through decomposing it into a number of simpler sub-problems, each solved separately. We demonstrate the advantages of this approach using abductive network classifiers on the 6-class standard dermatology dataset. Three problem decomposition scenarios are investigated, including class decomposition and two hierarchical approaches based on clinical practice and class separability properties. Two-stage classification schemes based on hierarchical decomposition boost the classification accuracy from 91% for the single-classifier monolithic approach to 99%, matching the theoretical upper limit reported in the literature for the accuracy of classifying the dataset. Such models are also simpler, achieving up to 47% reduction in the number of input variables required, thus reducing the cost and improving the convenience of performing the medical diagnostic tests required. Automatic selection of only relevant inputs by the simpler abductive network models synthesized provides greater insight into the diagnosis problem and the diagnostic value of various disease markers. The problem decomposition approach helps plan more efficient diagnostic tests and provides improved support for the decision-making process. Findings are compared with established guidelines of clinical practice, results of data analysis, and outcomes of previous informatics-based studies on the dataset.

Correlation of pneumonia score with electronic nose signature: A prospective study

OBJECTIVES

Ventilator-associated pneumonia (VAP) is a frequent complication in patients in surgical intensive care units. Pneumonia scores, chest radiography, and bronchoscopy are all employed, but there is no gold standard test for the diagnosis of VAP. The electronic nose, a sensor of volatile molecules, is well suited to testing the breath of mechanically ventilated patients. Our objective was to determine the potential use of an electronic nose as a diagnostic adjunct in the detection of VAP.

METHODS

We performed a prospective study of mechanically ventilated patients in a surgical intensive care unit. Clinical data, including temperature, white blood cell count, character and quantity of tracheal secretions, ratio of partial pressure of arterial oxygen to fraction of inspired oxygen, and chest radiographs, were collected, and a pneumonia score between 0 and 10 was calculated. Exhaled gas was sampled from the expiratory limb of the ventilator circuit. The gases were assayed with a commercially available electronic nose. Multidimensional data reduction analysis was used to analyze the results.

RESULTS

Forty-four patients were studied. Fifteen patients had pneumonia scores of 7 or greater, and 29 patients had scores of 6 or less. With Fisher discriminant analysis and K-nearest neighbor analysis, the electronic nose was able to discriminate between the two groups.

CONCLUSIONS

The electronic nose is a new technology that is inexpensive, noninvasive, and portable. We demonstrate its ability to predict pneumonia, based on a well-recognized scoring system. This technology promises to serve as a diagnostic adjunct in the management of VAP.

Improving medical diagnostic accuracy of ultrasound Doppler signals by combining neural network models

There are a number of different quantitative models that can be used in a medical diagnostic decision support system including parametric methods (linear discriminant analysis or logistic regression), nonparametric models (k nearest neighbor or kernel density) and several neural network models. The complexity of the diagnostic task is thought to be one of the prime determinants of model selection. Unfortunately, there is no theory available to guide model selection. This paper illustrates the use of combined neural network models to guide model selection for diagnosis of ophthalmic and internal carotid arterial disorders. The ophthalmic and internal carotid arterial Doppler signals were decomposed into time-frequency representations using discrete wavelet transform and statistical features were calculated to depict their distribution. The first-level networks were implemented for the diagnosis of ophthalmic and internal carotid arterial disorders using the statistical features as inputs. To improve diagnostic accuracy, the second-level networks were trained using the outputs of the first-level networks as input data. The combined neural network models achieved accuracy rates which were higher than that of the stand-alone neural network models.

Improved artificial neural networks in prediction of malignancy of lesions in contrast-enhanced MR-mammography

The aim of this study was to evaluate the capability of improved artificial neural networks (ANN) and additional novel training methods in distinguishing between benign and malignant breast lesions in contrast-enhanced magnetic resonance-mammography (MRM). A total of 604 histologically proven cases of contrast-enhanced lesions of the female breast at MRI were analyzed. Morphological, dynamic and clinical parameters were collected and stored in a database. The data set was divided into several groups using random or experimental methods [Training & Testing (T&T) algorithm] to train and test different ANNs. An additional novel computer program for input variable selection was applied. Sensitivity and specificity were calculated and compared with a statistical method and an expert radiologist. After optimization of the distribution of cases among the training and testing sets by the T & T algorithm and the reduction of input variables by the Input Selection procedure a highly sophisticated ANN achieved a sensitivity of 93.6% and a specificity of 91.9% in predicting malignancy of lesions within an independent prediction sample set. The best statistical method reached a sensitivity of 90.5% and a specificity of 68.9%. An expert radiologist performed better than the statistical method but worse than the ANN (sensitivity 92.1%, specificity 85.6%). Features extracted out of dynamic contrast-enhanced MRM and additional clinical data can be successfully analyzed by advanced ANNs. The quality of the resulting network strongly depends on the training methods, which are improved by the use of novel training tools. The best results of an improved ANN outperform expert radiologists.