Evaluation and amelioration of computer-aided diagnosis with artificial neural networks utilizing small-sized sample sets

Concerns about the specificity and reliability of artificial neural networks (ANNs) impede further application of ANNs in medicine. This is particularly true when developing computer-aided diagnosis (CAD) tools using ANNs for orphan diseases and emerging research areas where only a small-sized sample set is available. It is unreasonable to claim one ANN's performance as better than another simply on the basis of a single output without considering possible output variability due to factors including data noise and ANN training protocols. In this paper, a bootstrap resampling method is proposed to quantitatively analyze ANN output reliability and changing performance as the sample data and training protocols are varied. The method is tested in the area of feature classification for analysis of masses detected on mammograms. Our experiments show that ANNs performance, measured in terms of the area under the receiver operating characteristic (ROC) curve, is not a fixed value, but follows a distribution function sensitive to many factors. We demonstrate that our approach to determining the bootstrap estimates of confidence intervals (CIs) and prediction intervals (PIs) can be used to assure optimal performance in terms of ANN model configuration. We also show that the unintentional inclusion of data noise, which biases ANN results in small task-specific databases, can be accurately detected via the bootstrap estimates.     

Temporal separation of Cerenkov radiation and scintillation using artificial neural networks in Clinical LINACs

The irradiation of scintillator-fiber optic dosimeters by clinical LINACs results in the measurement of scintillation and Cerenkov radiation. In scintillator-fiber optic dosimetry, the scintillation and Cerenkov radiation responses are separated to determine the dose deposited in the scintillator volume. Artificial neural networks (ANNs) were trained and applied in a novel single probe method for the temporal separation of scintillation and Cerenkov radiation. Six dose profiles were measured using the ANN, with the dose profiles compared to those measured using background subtraction and an ionisation chamber. The average dose discrepancy of the ANN measured dose was 2.2% with respect to the ionisation chamber dose and 1.2% with respect to the background subtraction measured dose, while the average dose discrepancy of the background subtraction dose was 1.6% with respect to the ionisation chamber dose. The ANNs performance was degraded when compared with background subtraction, arising from an inaccurate model used to synthesise ANN training data.     

Assessment of respiratory system mechanics by artificial neural networks: an exploratory study.

We evaluated 1) the performance of an artificial neural network (ANN)-based technology in assessing the respiratory system resistance (Rrs) and compliance (Crs) in a porcine model of acute lung injury and 2) the possibility of using, for ANN training, signals coming from an electrical analog (EA) of the lung. Two differently experienced ANNs were compared. One ANN (ANN(BIO)) was trained on tracings recorded at different time points after the administration of oleic acid in 10 anesthetized and paralyzed pigs during constant-flow mechanical ventilation. A second ANN (ANN(MOD)) was trained on EA simulations. Both ANNs were evaluated prospectively on data coming from four different pigs. Linear regression between ANN output and manually computed mechanics showed a regression coefficient (R) of 0.98 for both ANNs in assessing Crs. On Rrs, ANN(BIO) showed a performance expressed by R = 0.40 and ANN(MOD) by R = 0.61. These results suggest that ANNs can learn to assess the respiratory system mechanics during mechanical ventilation but that the assessment of resistance and compliance by ANNs may require different approaches.     

The 'subsequent artificial neural network' (SANN) approach might bring more classificatory power to ANN-based DNA microarray analyses

MOTIVATION: Human decisions often proceed in two steps. Initially those most preferred are chosen followed by a subsequent choice of these preferences. Applying one artificial neural network (ANN), a classification is limited to the preselection process. The final categorization is only possible by a subsequent ANN that distinguishes the pre-chosen classes. Existing strategies using coupled ANNs are discussed and a new approach particularly suited for multiclass classification problems is introduced ('Subsequent ANN', SANN). RESULTS: Evaluating a simulated data base comprising 3 classes, classification results of SANN were obviously superior to those achieved by ANN. To evaluate a real-world data base the microarray benchmark GCM (14 classes) was chosen. The ANN results reached 72%, comparable to previous results. Using SANN, up to 81% of the tumors were correctly classified. AVAILABILITY: Programs used in this work and numerical results are available upon request.     

String matching artificial neural networks.

Three artificial neural networks (ANNs) are proposed for solving a variety of on- and off-line string matching problems. The ANN structure employed as the building block of these ANNs is derived from the harmony theory (HT) ANN, whereby the resulting string matching ANNs are characterized by fast match-mismatch decisions, low computational complexity, and activation values of the ANN output nodes that can be used as indicators of substitution, insertion (addition) and deletion spelling errors.     

CTSLoc: an indoor localization method based on CNN by using time-series RSSI

Due to the complexities of indoor WiFi signal propagations, it is challenging to improve the performance of indoor fingerprint-based positioning techniques which is the main hot research in Internet of Things. Most existing methods have limited positioning accuracy, since they do not take the full advantage of the information available, i.e. timing information attached to the Received Signal Strength Indicator (RSSI) vector, and adopt the inappropriate training methods. This paper proposes an indoor localization method based on Convolutional Neural Network (CNN) by using time-series RSSI, termed CTSLoc, by taking into account the correlation among RSSI in time and space. A CNN model is used to extract the temporal fluctuation patterns of RSSI and learn the nonlinear mappings from the signal features with time and space to position coordinates. Finally the trained model is used to predict the user’s location. An extensive experiment has been carried out in a space with the size of nearly 1000 squared meters, and a comprehensive comparison with several existing methods indicates that CTSLoc attains a lower average localization error (i.e. 4.23 m) and more stable performance than those methods. The CTSLoc method performs relatively less dependent on the amount of data which also eliminates spatial ambiguity and reduces the effect of noise on localization.

     

The superior fault tolerance of artificial neural network training with a fault/noise injectionbased genetic algorithm

Artificial neural networks (ANNs) are powerful computational tools that are designed to replicate the human brain and adopted to solve a variety of problems in many different fields. Fault tolerance (FT), an important property of ANNs, ensures their reliability when significant portions of a network are lost. In this paper, a fault/noise injection-based (FIB) genetic algorithm (GA) is proposed to construct fault-tolerant ANNs. The FT performance of an FIB-GA was compared with that of a common genetic algorithm, the back-propagation algorithm, and the modification of weights algorithm. The FIB-GA showed a slower fitting speed when solving the exclusive OR (XOR) problem and the overlapping classification problem, but it significantly reduced the errors in cases of single or multiple faults in ANN weights or nodes. Further analysis revealed that the fit weights showed no correlation with the fitting errors in the ANNs constructed with the FIB-GA, suggesting a relatively even distribution of the various fitting parameters. In contrast, the output weights in the training of ANNs implemented with the use the other three algorithms demonstrated a positive correlation with the errors. Our findings therefore indicate that a combination of the fault/noise injection-based method and a GA is capable of introducing FT to ANNs and imply that the distributed ANNs demonstrate superior FT performance.     

Rapid phenotypic characterization of Salmonella enterica strains by pyrolysis metastable atom bombardment mass spectrometry with multivariate statistical and artificial neural network pattern recognition

Pyrolysis mass spectrometry was investigated for rapid characterization of bacteria. Spectra of Salmonella were compared to their serovars, pulsed-field gel electrophoresis (PFGE) patterns, antibiotic resistance profiles, and MIC values. Pyrolysis mass spectra generated via metastable atom bombardment were analyzed by multivariate principal component-discriminant analysis and artificial neural networks (ANNs). Spectral patterns developed by discriminant analysis and tested with Leave-One-Out (LOO) cross-validation distinguished Salmonella strains by serovar (97% correct) and by PFGE groups (49%). An ANN model of the same PFGE groups was cross-validated, using the LOO rule, with 92% agreement. Using an ANN, thirty previously unseen spectra were correctly classified by serotype (97%) and at the PFGE level (67%). Attempts by ANN to model spectra grouped by resistance profile-but ignoring PFGE or serotype-failed (10% correct), but ANNs differentiating ten samples of the same serotype/PFGE class were more successful. To assess the information content of PyMS data serendipitously associated with or directly related to resistance character, the ten isolates were grouped into four, three, or two categories. The four categories corresponded to four resistance profiles. The four class and three class ANNs showed much improved but insufficient modeling power. The two-class ANN and a corresponding multivariate model maximized inferential power for a coarse antibiotic-resistance-related distinction. They each cross-validated by LOO at 90%. This is the first direct correlation of pyrolysis metastable atom bombardment mass spectrometry with immunological (e.g. serology) or molecular biology (e.g. PFGE) based techniques.     

What are artificial neural networks and what they can do?

The artificial neural networks (ANN) are very often applied in many areas of toxicology for the solving of complex problems, such as the prediction of chemical compound properties and quantitative structure-activity relationship. The aim of this contribution is to give the basic knowledge about conception of ANN, theirs division and finally, the typical application of ANN will be discussed. Due to the diversity of architectures and adaptation algorithms, the ANNs are used in the broad spectrum of applications from the environmental processes modeling, through the optimization to quantitative structure-activity relationship (QSAR) methods. In addition, especially ANNs with Kohonen learning are very effective classification tool. The ANNs are mostly applied in cases, where the commonly used methods does not work.     

Estimation of biomass concentrations in fermentation processes for recombinant protein production

Online biomass estimation for bioprocess supervision and control purposes is addressed. As the biomass concentration cannot be measured online during the production to sufficient accuracy, indirect measurement techniques are required. Here we compare several possibilities for the concrete case of recombinant protein production with genetically modified Escherichia coli bacteria and perform a ranking. At normal process operation, the best estimates can be obtained with artificial neural networks (ANNs). When they cannot be employed, statistical correlation techniques can be used such as multivariate regression techniques. Simple model-based techniques, e.g., those based on the Luedeking/Piret-type are not as accurate as the ANN approach; however, they are very robust. Techniques based on principal component analysis can be used to recognize abnormal cultivation behavior. For the cases investigated, a complete ranking list of the methods is given in terms of the root-mean-square error of the estimates. All techniques examined are in line with the recommendations expressed in the process analytical technology (PAT)-initiative of the FDA.     

Trunk acceleration for neuroprosthetic control of standing: a pilot study

This pilot study investigated the potential of using trunk acceleration feedback control of center of pressure (COP) against postural disturbances with a standing neuroprosthesis following paralysis. Artificial neural networks (ANNs) were trained to use three-dimensional trunk acceleration as input to predict changes in COP for able-bodied subjects undergoing perturbations during bipedal stance. Correlation coefficients between ANN predictions and actual COP ranged from 0.67 to 0.77. An ANN trained across all subject-normalized data was used to drive feedback control of ankle muscle excitation levels for a computer model representing a standing neuroprosthesis user. Feedback control reduced average upper-body loading during perturbation onset and recovery by 42% and peak loading by 29% compared with optimal, constant excitation.     

The use of ladder particle swarm optimisation for quantitative structure–activity relationship analysis of human immunodeficiency virus-1 integrase inhibitors

This contribution focuses on the use of ladder particle swarm optimisation (LPSO) on modelling of oxadiazole- and triazole-substituted naphthyridines as human immunodeficiency virus-1 integrase inhibitors. Artificial neural network (ANN) and Monte Carlo cross-validation techniques were combined with LPSO to develop a quantitative structure–activity relationship model. The techniques of LPSO, ANN and sample set partitioning based on joint x–y distances were applied as feature selection, mapping and model evaluation, respectively. The variables selected by LPSO were used as inputs of Bayesian regularisation ANN. The statistical parameters of correlation of deterministic, R ², and root-mean-square error for the test set were 0.876 and 0.23, respectively. Robustness of the model was confirmed by Y-randomisation method. Comparison of the LPSO–ANN results with those of stepwise multiple linear regression (MLR), LPSO–MLR and LPSO–MLR–ANN showed the superiority of LPSO–ANN. Inspection of the selected variables indicated that atomic mass, molecular size and electronic structure of the molecules play a significant role in inhibitory behaviour of oxadiazole- and triazole-substituted naphthyridines.     

Artificial neural network modelling of a large-scale wastewater treatment plant operation

Artificial Neural Networks (ANNs), a method of artificial intelligence method, provide effective predictive models for complex processes. Three independent ANN models trained with back-propagation algorithm were developed to predict effluent chemical oxygen demand (COD), suspended solids (SS) and aeration tank mixed liquor suspended solids (MLSS) concentrations of the Ankara central wastewater treatment plant. The appropriate architecture of ANN models was determined through several steps of training and testing of the models. ANN models yielded satisfactory predictions. Results of the root mean square error, mean absolute error and mean absolute percentage error were 3.23, 2.41 mg/L and 5.03% for COD; 1.59, 1.21 mg/L and 17.10% for SS; 52.51, 44.91 mg/L and 3.77% for MLSS, respectively, indicating that the developed model could be efficiently used. The results overall also confirm that ANN modelling approach may have a great implementation potential for simulation, precise performance prediction and process control of wastewater treatment plants.     

Artificial neural networks for decision-making in urologic oncology

Artificial neural networks (ANNs) are computational methodologies that perform multifactorial analyses, inspired by networks of biological neurons. Like neural networks, ANNs contain layers of simple points (nodes) of data that interract through carefully weighted connection lines. ANNs are "trained" and balanced by having been previously fed data, which the ANN uses as the means for adjusting its interconnections. Studies have shown that novel and highly accurate ANNs significantly enhance the ability to detect prostate cancer early (high sensitivity) while avoiding a greater number of unnecessary tissue samplings (high specificity). The use of ANNs in prostate cancer is ideal because of 1) multiple predicting factors that influence outcome; 2) the desire to offer individual consulting based on various tests; 3) the fact that prior logistic regression analysis results have had serious limitations in application; and 4) the need for an up-to-date tool that can apply easily to everyone. An ANN should be seen as an important tool that is complementary to the physician's personal knowledge and judgment in making decisions.     

Pose estimation in automated visual inspection using ANN

In this paper, we present an accurate and robust pose estimator of rigid, polyhedral objects, based on Artificial Neural Networks (ANN), as suitable for Automated Visual Inspection (AVI) applications. The estimator is novel in the sense that it is trained with different poses of the objects having dimensional deviations within its tolerance range and is therefore robust with respect to within tolerance dimensional errors. The estimation accuracy is scalable and our computer simulation experiments in the existing configurations of ANNs have shown an accuracy better than 4% of the placement error. The ANN based pose estimator offers several advantages over the classical implementations.     

Artificial neural networks: fundamentals, computing, design, and application

Artificial neural networks (ANNs) are relatively new computational tools that have found extensive utilization in solving many complex real-world problems. The attractiveness of ANNs comes from their remarkable information processing characteristics pertinent mainly to nonlinearity, high parallelism, fault and noise tolerance, and learning and generalization capabilities. This paper aims to familiarize the reader with ANN-based computing (neurocomputing) and to serve as a useful companion practical guide and toolkit for the ANNs modeler along the course of ANN project development. The history of the evolution of neurocomputing and its relation to the field of neurobiology is briefly discussed. ANNs are compared to both expert systems and statistical regression and their advantages and limitations are outlined. A bird's eye review of the various types of ANNs and the related learning rules is presented, with special emphasis on backpropagation (BP) ANNs theory and design. A generalized methodology for developing successful ANNs projects from conceptualization, to design, to implementation, is described. The most common problems that BPANNs developers face during training are summarized in conjunction with possible causes and remedies. Finally, as a practical application, BPANNs were used to model the microbial growth curves of S. flexneri. The developed model was reasonably accurate in simulating both training and test time-dependent growth curves as affected by temperature and pH.     

Artificial neural networks trained to detect viral and phage structural proteins

Phages play critical roles in the survival and pathogenicity of their hosts, via lysogenic conversion factors, and in nutrient redistribution, via cell lysis. Analyses of phage- and viral-encoded genes in environmental samples provide insights into the physiological impact of viruses on microbial communities and human health. However, phage ORFs are extremely diverse of which over 70% of them are dissimilar to any genes with annotated functions in GenBank. Better identification of viruses would also aid in better detection and diagnosis of disease, in vaccine development, and generally in better understanding the physiological potential of any environment. In contrast to enzymes, viral structural protein function can be much more challenging to detect from sequence data because of low sequence conservation, few known conserved catalytic sites or sequence domains, and relatively limited experimental data. We have designed a method of predicting phage structural protein sequences that uses Artificial Neural Networks (ANNs). First, we trained ANNs to classify viral structural proteins using amino acid frequency; these correctly classify a large fraction of test cases with a high degree of specificity and sensitivity. Subsequently, we added estimates of protein isoelectric points as a feature to ANNs that classify specialized families of proteins, namely major capsid and tail proteins. As expected, these more specialized ANNs are more accurate than the structural ANNs. To experimentally validate the ANN predictions, several ORFs with no significant similarities to known sequences that are ANN-predicted structural proteins were examined by transmission electron microscopy. Some of these self-assembled into structures strongly resembling virion structures. Thus, our ANNs are new tools for identifying phage and potential prophage structural proteins that are difficult or impossible to detect by other bioinformatic analysis. The networks will be valuable when sequence is available but in vitro propagation of the phage may not be practical or possible.     

Application of an artificial neural network (ANN) model for predicting mosquito abundances in urban areas

The mosquito species is one of most important insect vectors of several diseases, namely, malaria, filariasis, Japanese encephalitis, dengue, and so on. In particular, in recent years, as the number of people who enjoy outdoor activities in urban areas continues to increase, information about mosquito activity is in demand. Furthermore, mosquito activity prediction is crucial for managing the safety and the health of humans. However, the estimation of mosquito abundances frequently involves uncertainty because of high spatial and temporal variations, which hinders the accuracy of general mechanistic models of mosquito abundances. For this reason, it is necessary to develop a simpler and lighter mosquito abundance prediction model. In this study, we tested the efficacy of the artificial neural network (ANN), which is a popular empirical model, for mosquito abundance prediction. For comparison, we also developed a multiple linear regression (MLR) model. Both the ANN and the MLR models were applied to estimate mosquito abundances in 2-year observations in Yeongdeungpo-gu, Seoul, conducted using the Digital Mosquito Monitoring System (DMS). As input variables, we used meteorological data, including temperature, wind speed, humidity, and precipitation. The results showed that performances of the ANN model and the MLR model are almost same in terms of R and root mean square error (RMSE). The ANN model was able to predict the high variability as compared to MLR. A sensitivity analysis of the ANN model showed that the relationships between input variables and mosquito abundances were well explained. In conclusion, ANNs have the potential to predict fluctuations in mosquito numbers (especially the extreme values), and can do so better than traditional statistical techniques. But, much more work needs to be conducted to assess meaningful time delays in environmental variables and mosquito numbers.