Neural Network Pattern Recognition Research Articles

Prediction of Successful Surgery Outcome in Lumbar Disc Herniation Based on Artificial Neural Networks

Introduction The aim of this study, was to develop an artificial neural networks (ANNs) model for predicting successful surgery outcome in lumbar disc herniation (LDH). Materials and Methods An ANN model and a logistic regression (LR) model were used to predict outcomes. The age, gender, duration of symptoms, smoking status, surgical level, visual analog scale of leg/back pain, the Zung depression scale, and the Japanese Orthopaedic Association score, were determined as the input variables for the established ANN model. The Macnab classification was used for outcome assessment. ANNs on data from LDH patients, who had surgery, were trained to predict 2-year successful discectomy using several input variables. Sensitivity analysis to the established ANN model was used to identify the relevant variables. For evaluating the two models, the area under a receiver operating characteristic curve, accuracy rate of predicting, and Hosmer-Lemeshow (H-L) statistics were considered. Results A total of 203 (96 male, 107 female, mean age 48.3 ± 9.8 years) patients were categorized into training, testing, and validation datasets consisting of 101, 51, and 51 cases, respectively. Surgical successful outcome was categorized as: excellent, 32.0%; good, 40.9%; fair, 20.7%; and poor, 6.4% at 2-year follow-up. Compared with the LR model, the ANN model showed better results: accuracy rate, 95.8%; H-L statistic, 41.5%; and AUC, 0.82% of patients, respectively. Conclusion The findings show that ANNs can predict successful surgery outcome with a high level of accuracy in LDH patients. Such information is of use in the clinical decision-making process. Disclosure of Interest None declared References Fardon DF, Milette PC; Combined Task Forces of the North American Spine Society, American Society of Spine Radiology, and American Society of Neuroradiology. Nomenclature and classification of lumbar disc pathology. Recommendations of the Combined task Forces of the North American Spine Society, American Society of Spine Radiology, and American Society of Neuroradiology. Spine 2001;26(5):E93-E113 Li YC, Liu L, Chiu WT, Jian WS. Neural network modeling for surgical decisions on traumatic brain injury patients. Int J Med Inform 2PO.069;57(1):1–9 Bishop CM. Neural Networks for Pattern Recognition. Oxford, UK: Clarendon; 1995 Azimi P, Benzel EC, Shahzadi S, Azhari S, Mohammadi HR. Use of artificial neural networks to predict surgical satisfaction in patients with lumbar spinal canal stenosis. J Neurosurg Spine 2014;20(3):300–305 Price DD, McGrath PA, Rafii A, Buckingham B. The validation of visual analogue scales as ratio scale measures for chronic and experimental pain. Pain 1983;17(1):45–56 Azimi P, Mohammadi HR, Montazeri A. An outcome measure of functionality and pain in patients with lumbar disc herniation: a validation study of the Japanese Orthopedic Association (JOA) score. J Orthop Sci 2012;17(4):341–345 Zung WW, Richards CB, Short MJ. Self-rating depression scale in an outpatient clinic. Further validation of the SDS. Arch Gen Psychiatry 1965;13(6):508–515 Macnab I. Chapter 14. Pain and disability in degenerative disc disease. Clin Neurosurg 1973;20:193–196 Cross SS, Harrison RF, Kennedy RL. Introduction to neural networks. Lancet 1995;346(8982):1075–1079 Rughani AI, Dumont TM, Lu Z, et al. Use of an artificial neural network to predict head injury outcome. J Neurosurg 2010;113(3):585–590 Finneson BE. A lumbar disc surgery predictive score card. Spine 1978;3(2):186–188

Construction of Artificial Neural Networks for Pattern Recognition Using a Successive Geometric Segmentation Method

This work presents a new method for artificial neural networks construction and training based on successive geometric segmentation (SGSM). This methodology is capable of generating both the network topology and neurons weights without parameters’ specification. The SGSM groups the data of each class into hyper-boxes (HBs) aligned in accordance with the largest axis of its points’ distribution. If the HBs are linearly separable, then a separating hyperplane may be identified resulting in a neuron. If it is not, then a segmentation technique is applied to divide the data into smaller classes for new HBs. For each subdivision, new neurons are added to the network. The tests show a rapid method with a high success rate.

Design and Comparative Analysis of SVM and PR Neural Network for Classification of Brain Tumour in MRI

grams (EEGs) or MRI are progressively emerging as a significant measure of brain activity and they possess immense potential for the diagnosis and treatment of mental and brain diseases and abnormalities. This research paper presents an automated system for efficient classification of brain tumours in MRI images using Support Vector Machine (SVM). It works on the principle of a vector which supports in training of classifier using train images. The results obtained from the above method are compared with those obtained from Artificial Neural Networks (ANNs). This method uses pattern recognition algorithms for the classification of tumour in MR images. First, a pattern recognition neural network is created and then trained. The trained pattern recognition neural network when fed with a test MR image, effectively classifies the images. The performance analysis shows that qualitative results obtained from the proposed model are comparable with those obtained by ANN. The experimental results demonstrate the effectiveness of the proposed system in classification of brain tumour in MR images. The implementation is done using MATLAB R2012b using Artificial Neural Network and Image Processing Toolbox.

Construction of Artificial Neural Networks for Pattern Recognition Using a Successive Geometric Segmentation Method

Construction of Artificial Neural Networks for Pattern Recognition Using a Successive Geometric Segmentation Method

Study on Artificial Neural Networks to Identify Sedimentary Microfacies

Based on a number of stratigraphic sedimentary information included in log data, application of the Artificial Neural Network to identify sedimentary microfacies from well logging data can complete the series auto-interpreting. The application can improve the auto-interpreting accuracy and make us get more satisfied results. Ten parameters from the well logging curves are selected for to describing their shape characteristics when the deposition patterns of 8 in gas-bearing formation of Upper Paleozoic group, Ordos basin are studied. Effective parameters were selected on the basis of cores, and based on artificial neural network pattern recognition technique; the sedimentary microfacies of well cross section were auto-interpreted. About 300 wells and the results were interpreted by using the software. The software will be fit for the researchers who have the experiences of geological interpretation and some backgrounds of local geology.

Complicated Graphics Model based on Neural Network of Particle Swarm

The past decades have seen the great progress pattern recognition and image understanding, motivated by a wide range of real world applications. The previous approaches mostly based on the feature extraction and recognition methods, and usually suffer from the problem of weak discrimination power. In this paper, we propose a complex model based on particle swarm and neural network pattern recognition methods. The proposed approach uses the class label of the input sample as the actual output of the neural network which has the maximum node corresponds to the class. To evaluate the effectiveness of the proposed approach, we experimentally compare the particle swarm neural network based approach with and SVM classification algorithm. The experimental results show that, both the recognition accuracy and training time of the particle swarm and neural network based approach, are better than that of SVM. The particle swarm neural network pattern recognition in complex graphics showed a great advantage in application.

Automatic seizure detection based on Teager Energy Cepstrum and pattern recognition neural networks

About 1–3% of the world population suffers from epilepsy. A long term inpatient/ambulatory electroencephalogram (EEG), lasting from a few hours to several days, which definitely contains hallmarks of epilepsy, is required clinically to diagnose, monitor and localize the epileptogenic zone. The traditional method relies on well-trained neurophysiologists who visually inspect the entire lengthy EEG signals, which is tedious and time-consuming. Therefore, many automated epileptic detection systems have been developed and such systems significantly reduce the time taken to review off-line the long-term EEG recordings and facilitate the neurologist to diagnose and treat more patients in a given time. There are not many studies that have explored, to a sufficient depth, the features used in other domains of signal processing, for example, the Teager energy cepstrum (TECEP), attempted use in seizure detection. Epileptic seizures are abnormal sudden discharges in the brain with signatures manifesting in the EEG recordings by frequency changes and increased amplitudes. These changes, in this work, are captured through static and dynamic features derived from TECEP. We compared the performance of the baseline TECEP and its two composite vectors with those of the traditional cepstrum (CEP) and its corresponding two composite vectors, in EEG epileptic seizure detection. The first composite vector includes velocity cepstral coefficients and the second includes velocity, as well as, acceleration cepstral coefficients. The comparison is tried on eight different classification problems which encompass all the possible discriminations in the medical field related to epilepsy, using pattern recognition neural network (PRNN). In this case, it is found that the overall performance of both the Teager energy composite vectors excels those of traditional cepstral composite vectors. The static and dynamic features derived from TECEP outperform those derived from CEP, suggesting their suitability in epilepsy seizure detection.

Comparison between BP and RBF Neural Network Pattern Recognition Process Applied in the Droplet Analyzer

Among various pattern recognition methods used for liquid identification, the method based on neural network has the advantages of robustness and fault tolerance, which can study and adapt to the uncertain system. The waveform analysis is exploited for feature extraction of the liquid droplet fingerprint (LDF) in this paper, and the liquid identification is carried out by means of BP and RBF neural network. The experimental results proved that the recognition rate is excellent in both of these two methods. In condition that the training data is limited, RBF network is better than BP network in recognition speed and rate.

Pattern Recognition Neural Networks in Classification Groups of Partial Removable Dentures Lifetime

Classification is one of the most common research work associated with neural networks. This paper cast lighton the use of this technique as a robust statistical procedure for modeling data regarding problem of classification groups of partial removable dentures lifetime. Groups are either 1 which refers to lifetime exceeds average, and 0 which refers to lifetime less than or equal average lifetime. The main purpose of this work is to provide a new technique of classification as well as to stimulate further research interests and efforts in this topic.

Image Segmentation with Artificial Neural Networs Alongwith Updated Jseg Algorithm

Image segmentation plays a very crucial role in computer vision. Computational methods based on JSEG algorithm is used to provide the classification and characterization along with artificial neural networks for pattern recognition. It is possible to run simulations and carry out analyses of the performance of JSEG image segmentation algorithm and artificial neural networks in terms of computational time. A Simulink is created in Matlab software using Neural Network toolbox in order to study the performance of the system. In this paper, four windows of size 9*9, 17*17, 33*33 and 65*65 has been used. Then the corresponding performance of these windows is compared with ANN in terms of their computational time. Image segmentation is a process of separating images into small section that can be used as meaningful objects in computing and searching techniques. These segments are developed by using a number of processing techniques that are clustering, J-fractal image calculation. In Bottom Up approach techniques like JSEG, pixels of image are processed and divided into classes. These classes are used for indentifying image segments, once pixels are labeled using classes. The classes are clustered to divide image pixels and for image segments. Existing JSEG image segmentation technique uses K-Means for clustering. But there are a number of clustering variations of K-means available which can be used for improving results of image segmentation. K means clustering algorithm suffers from problem of over segmentation which is a big problem in process of segmentation in case of JSEG. We will be experimenting with available clustering techniques for building image segmentation. Many of existing segmentation techniques such as direct clustering methods in color space work well on homogeneous color regions. Natural scenes are rich in color and texture. Parameter estimation is a difficult problem and requires good homogeneous region for robust estimation. A new approach called JSEG is proposed towards this goal. This approach does not tend to estimate a specific model for a texture region. It test for the homogeneity of a given color pattern, which is computationally more feasible method. JSEG algorithm can be improved by adding enhanced classification and modified K means clustering.

Thin Film Coated QCM-Sensors and Pattern Recognition Methods for Discrimination of VOCs

Abstract discrimination and quantification of volatile organic compounds (VOCs) using a non-selective sensor requires a combination of sensors followed by pattern recognition methods. Based on this concept, this paper deals with the discrimination of gas from the responses of several gas sensors coated with different type of polymer. Quartz crystal microbalance (QCM) electrodes were coated from hexamethyldisiloxane (HMDSO), hexamethyldisilazane (HMDSN) and tetraethoxysilane (TEOS) for the elaboration of gas sensors with different chemical affinity towards VOC molecules. The sensitivity of the elaborated QCM-based sensors was evaluated by monitoring the frequency shifts of the quartz exposed to different concentrations of volatile organic compounds, such as; ethanol, benzene and chloroform. The sensors responses data have been used for the identification and quantification of VOCs. The principal component analysis (PCA) and the neural-network (NNs) pattern recognition analysis were used for the discrimination of gas species and concentrations. Good separation among gases has been obtained using the principal component analysis. The feed-forward multilayer neural network (MLNNs) with a hidden layer and trained by Broyden Fletcher Goldfarb Shanno (BFGS) Quasi Newton algorithm has been implemented in order to identify and quantify the VOCs. By increasing the number of the neuron in the hidden layer, the precision of the estimate concentration increases. The approach is standard, however its application on the elaborated sensors have not been studied in depth so far.

Label free detection of pseudorabies virus infection in Vero cells using laser force analysis

The rapid and robust identification of viral infections has broad implications for a number of fields, including medicine, biotechnology and biodefense. Most detection systems rely on specific molecules, such as nucleic acids or proteins, to identify the target(s) of interest. These molecules afford great specificity, but are often expensive, labor-intensive, labile and limited in scope. Label free detection methods seek to overcome these limitations by instead using detection methods that rely on intrinsic properties as a basis for identifying and separating species of interest and thus do not rely on specific prior knowledge of the target. Optical chromatography, one such technique, uses the balance between optical and fluidic drag forces within a microfluidic channel to determine the optical force on cells or particles. Here we present the application of individual optical force measurements as a means of investigating pseudorabies virus infection in Vero cells. Optical force differences are seen between cells from uninfected and infected populations at a multiplicity of infection as low as 0.001 and as soon as 2 hours post infection, demonstrating the potential of this technique as a means of detecting viral infection. Through the application of a pattern recognition neural network, individual cell size data are combined with optical force as a means of classifying cell populations. Potential applications include the early detection of bloodborne pathogens for the prevention of sepsis and other diseases as well as the detection of biological threat agents.

Neural Network Pattern Recognition and its Application

many functions are possessed by the neural network such as parallel processing, self-learning and self-adapting. It could approximate any nonlinear function with any precision. A very effective way is provided by the neural network to deal with complex control problems, such as nonlinear, multivariable and uncertain ones etc. Therefore, the neural network is widely used in many aspects: pattern recognition, system identification and control fields and so on.It is developed in the paper about the application of neural networks pattern recognition and system identification. With MATLAB 6.1 and Visual Basic 6.0 design platform and developing tool, for some application instances, implement modeling, simulation and systematic test tasks of the neural networks pattern recognition and system identification. The above research and instances indicate that the neural networks pattern recognition and system identification based on MATLAB have better application prospects.

Fault Diagnosis on Transmission System of Wind Turbines Based on Wavelet Packet Transform and RBF Neural Networks

Wind turbine transmission system with abundant fault feature and variable types, the vibration signal was a carrier of fault features and it can reflect most of the fault information in the wind turbine transmission system. As there were a large number of transient and non-stationary signals accompany with the vibration signals, so wavelet packet transform was adopted for feature extraction. As RBF Neural network has a strong nonlinear mapping ability and self-adaptability, so it was introduced to the diagnosis system for network training, the neural networks structure and learning algorithm was presented, which could enhance the accuracy of diagnosis. The two-level neural networks recognition method was proposed, first level for fault classification and second level for fault diagnosis. The example shows that this method can be effectively applied to transmission system of wind turbine fault diagnosis with wavelet packet algorithm for fault feature extraction and RBF neural network for pattern recognition.

From evolving artificial gene regulatory networks to evolving spiking neural networks for pattern recognition

The GReaNs platform (which stands for Gene Regulatory evolutionary artificial Networks) is an artificial life software platform that has been previously shown to allow the evolution of networks for signal processing and animat control [1]. The structure of the resulting network in this model is encoded in a linear genome, without imposing any restrictions on the size of the genome or the size of the network. We extended this model to encode spiking neural networks [4]. In previous work using GReaNs [1], each node in the artificial network has been considered to be an analog of a biological transcriptional unit. However, the nodes can also be seen as artificial neurons. Our extension has introduced to GReaNs two models of spiking neurons (LIF: leaky integrate and fire neurons with a fixed threshold [2] and AdEx: adaptive-exponential integrate and fire [3]). Here we use GReaNs for the evolution of spiking networks for pattern recognition. The genetic algorithm used in GReaNS is based on evolving the number of the neurons and the weights of the connections, but the synaptic delays are constant. We evolved a network that generates spikes at a maximum firing rate when the network is presented with one spike from five input neurons in a specific order, and that stays silent when the order is reversed. We also used three inputs, each with one spike, requiring the network to fire at am aximum rate for a specific temporal order of spikes and to remain silent for all the other permutations. In further work, we plan to investigate the generalization abilities of the evolved networks and the robustness of their behaviour to noise. GReaNs can export a PyNN script, which allows the simulation of an evolved network on different backends, including neuromorphic real-time hardware for real time simulation.

Improving the inhibitory activity of arylidenaminoguanidine compounds at the N-methyl-d-aspartate receptor complex from a recursive computational-experimental structure–activity relationship study

Using a combination of both the partial least squares (PLS) and back-propagation artificial neural network (ANN) pattern recognition methods, several models have been developed to predict the activity of a series of arylidenaminoguanidine analogs as inhibitory modulators of the N-methyl-d-aspartate receptor complex. This was done by correlating structural and physicochemical descriptors obtained from computation software with the experimentally observed [3H]MK-801 displacement ability of a small library of synthesized and in vitro screened arylidenaminoguanidines. Results for the generated PLS model were r2=0.814, rmsd=0.208, rCV2=0.714, loormsd=0.261. The ANN model was created utilizing the eleven descriptors from the PLS model for comparison. The quality of the ANN model (r2=0.828, rmsd=0.200, rCV2=0.721, loormsd=0.257) is similar to the PLS model, and indicates that the feature between the inputs and the output is majorly linear. These computational models were able to predict inhibition of the NMDA receptor complex by this series of compounds in silico, affording a predictive structure-based ‘pre-screening’ paradigm for the arylideneaminoguanidine analogs.

A Journal of Real Peak Recognition of Electrocardiogram (ECG) Signals Using Neural Network

This paper describes about the analysis of electrocardiogram (ECG) signals using neural network approach. Heart structure is a unique system that can generate ECG signals independently via heart contraction. Basically, an ECG signal consists of PQRST wave. All these waves are represented respective heart functions. Normal healthy heart can be simply recognized by normal ECG signal while heart disorder or arrhythmias signals contain differences in terms of features and morphological attributes in their corresponding ECG waveform. Some major important features will be extracted from ECG signals such as amplitude, duration, pre-gradient, post-gradient and so on. These features will then be fed as an input to neural network system. The target output represented real peaks of the signals is also being defined using a binary number. Result obtained showing that neural network pattern recognition is able to classify and recognize the real peaks accordingly with overall accuracy of 81.6% although there might be limitations and misclassification happened. Future recommendations have been highlighted to improve network’s performance in order to get better and more accurate result.

Face Recognition System Based on Different Artificial Neural Networks Models and Training Algorithms

Face recognition is one of the biometric methods that is used to identify any given face image using the main features of this face. In this research, a face recognition system was suggested based on four Artificial Neural Network (ANN) models separately: feed forward backpropagation neural network (FFBPNN), cascade forward backpropagation neural network (CFBPNN), function fitting neural network (FitNet) and pattern recognition neural network (PatternNet). Each model was constructed separately with 7 layers (input layer, 5 hidden layers each with 15 hidden units and output layer). Six ANN training algorithms (TRAINLM, TRAINBFG, TRAINBR, TRAINCGF, TRAINGD, and TRAINGD) were used to train each model separately. Many experiments were conducted for each one of the four models based on 6 different training algorithms. The performance results of these models were compared according to mean square error and recognition rate to identify the best ANN model. The results showed that the PatternNet model was the best model used. Finally, comparisons between the used training algorithms were performed. Comparison results showed that TrainLM was the best training algorithm for the face recognition system.

Successful combination of computationally inexpensive GIAO 13C NMR calculations and artificial neural network pattern recognition: a new strategy for simple and rapid detection of structural misassignments

GIAO NMR chemical shift calculations coupled with trained artificial neural networks (ANNs) have been shown to provide a powerful strategy for simple, rapid and reliable identification of structural misassignments of organic compounds using only one set of both computational and experimental data. The geometry optimization, usually the most time-consuming step in the overall procedure, was carried out using computationally inexpensive methods (MM+, AM1 or HF/3-21G) and the NMR shielding constants at the affordable mPW1PW91/6-31G(d) level of theory. As low quality NMR prediction is typically obtained with such protocols, the decision making was foreseen as a problem of pattern recognition. Thus, given a set of statistical parameters computed after correlation between experimental and calculated chemical shifts the classification was done using the knowledge derived from trained ANNs. The training process was carried out with a set of 200 molecules chosen to provide a wide array of chemical functionalities and molecular complexity, and the results were validated with a set of 26 natural products that had been incorrectly assigned along with their 26 revised structures. The high prediction effectiveness observed makes this method a suitable test for rapid identification of structural misassignments, preventing not only the publication of wrong structures but also avoiding the consequences of such a mistake.

Multiple Sensor Monitoring in Nickel Alloy Turning for Tool Wear Assessment via Sensor Fusion

A multiple sensor monitoring system comprising cutting force, acoustic emission and vibration sensing units was employed for tool state assessment during turning of Inconel 718 nickel alloy. Feature extraction was realised by processing the detected sensor signals in order to reduce the high dimensionality of the sensorial data. The extracted features were fused to realise a sensor fusion methodology based on neural network pattern recognition for decision making on tool wear condition.