Neural Network in Medical Application: a Review

Artificial Neural Networks or widely known as Neural networks (ANNs or NNs) is a computational paradigm that comprises of mathematical, statistical, biological sciences and philosophy. These paradigms formulate a formula to form a brain like function, called artificial neuron. Artificial neuron comprises of large number of computational processing elements called units, nodes or cells. Analogously, these processing elements mimic the processing elements of biological neuron. This paper discusses neural network as a powerful tool to enhance current medical prognostic techniques. One of the most popular learning algorithms that are backpropagation algorithm is discussed. Several applications of neural network in medical application are reviewed. 1.0 INTRODUCTION Artificial Intelligence (AI) techniques have been shown well performing in many applications. AI techniques such as Neural Network (NN), expert system (ES), Fuzzy Logic (FL), Data Mining (DM), Genetic Algorithm (GA), Intelligent Agent (IA) and Machine Learning (ML), are among the most popular techniques implemented in various applications. These techniques utilizing their intelligence ability to learn from experience, learn and provide meaning for fuzzy data, quick response to a new situation, able to use the knowledge, reasons and manipulates the environment. Neural Networks (NNs) is one of the most popular AI technique implemented in medical application. Sarle (1994) describe the usage of NN in three main ways, typically, as models of biological nervous systems and “intelligence”, as real-time adaptive signal processors or controllers implemented in hardware and as methods for data analysis. Passold et al. (1996) summarized the benefits of NN as follows:  Ability to process a massive of input data  Simulation of diffuse medical reasoning  Higher performances when compared with statistical approaches  Self-organizing ability-learning capability  Easy knowledge base updating NN is a powerful tool to enhance current medical diagnostic techniques. Partridge et al. (1996) listed several potentials of NN over conventional computation and manual analysis in medical application:  Implementation using data instead of possibly ill defined rules.  Noise and novel situations are handled automatically via data generalization. Wan Hussain Wan Ishak, Fadzilah Siraj, Abu Talib Othman (2001). Neural Network in Medical Application: A Review. Simposium Kedua Penyelidikan Pengurusan Siswazah (16-17 Oct. 2001). Universiti Utara Malaysia  Predictability of future indicator values based on past data and trend recognition.  Automated real-time analysis and diagnosis.  Enables rapid identification and classification of input data.  Eliminates error associated with human fatigue and habituation. This paper discuses the potential of NN in several medical applications. Backpropagation algorithm, which is one of the most popular learning algorithm in NN is also presented. This algorithm has been used by many medical researches and has demonstrated NN’s predictive capability. 2.0 MEDICAL APPLICATIONS OF NEURAL NETWORK Neural Network (NN) in medicine has attracted many researchers. A simple search by Machado (1996) in Medline for articles about computer-based NN between 1982 and 1994 resulted with more than 600 citations. Another search by Dybowski (2000) in the same database yields 473 publications in 1998. According to Dybowski, NN in medicine is subjected to increase, as the numbers of experts are limited while interpretation work at clinical laboratories is subjected to mounting. Furthermore, the complexity of patient related data could easily overlooked even by the specialist. NNs have been implemented in many medical applications such as medical basic sciences, clinical medicine, signal processing and interpretation and image processing. Applications in Basic sciences In basic sciences, NN helps clinician to investigate the impact of parameter after certain conditions or treatments. It supplies clinicians with information about the risk or incoming circumstances regarding the domain. Learning the time course of blood glucose (Prank et al., 1998) for example can help clinician to control the diabetes mellitus. They used feedforward NN for predicting the time course of blood glucose levels from the complex interaction of glucose counterregulatory hormones and insulin. Multi-Layer Perceptron (MLP) with sigmoidal Feed-Forward and standard BackPropagation (BP) learning algorithm was employed as a forecaster for bacteriaantibiotic interactions of infectious diseases (Abidi and Goh, 1998). They conclude that the 1-month forecaster produces output correct to within 1  occurrences of sensitivity. However, predictions for the 2-month and 3-month are less accurate. Applications in Clinical Medicine Patient who hospitalize for having high-risk diseases required special monitoring as the disease might spread in no time. NN has been used as a tool for patient diagnosis and prognosis to determine patients’ survival. Bottaci and Drew (1997) investigate fully connected feed forward MLP and BP learning rule, were able to predict patients with colorectal cancer more accurately than clinicopathological methods. They indicate that NN predict the patients’ survival and death very well compared to the surgeons. Pofahl et al. (1998) compare the performance of NN, Ranson criteria and Acute Physiology and Chronic Health Evaluation (APACHE II) scoring system for Wan Hussain Wan Ishak, Fadzilah Siraj, Abu Talib Othman (2001). Neural Network in Medical Application: A Review. Simposium Kedua Penyelidikan Pengurusan Siswazah (16-17 Oct. 2001). Universiti Utara Malaysia predicting length of stay (LOS) greater than 7 days for acute pancreatitis patient. Their study indicates that NN achieve the highest sensitivity (75%) for predicting LOS greater than 7 days. Ohlsson et al. (1999) presents their study for the diagnosis of Acute Myocardial Infarction. In their study NN with 10 hidden nodes and one output neuron have been used as the classifier to classified whether the patient suffered from Acute Myocardial Infarction (1) or not (0). The results show that NN performance is 0.84 and 0.85 under receiver-operating characteristics (ROC). Applications in Signal Processing and Interpretation Signal processing and interpretation in medicine involve a complex analysis of signals, graphic representations, and pattern classification. Consequently, even experienced surgeon could misinterpret or overlooked the data (Janet, 1997; Dybowski, 2000). In electrocardiographic (ECG) analysis for example, the complexity of the ECG readings of acute myocardial infarction could be misjudged even by experienced cardiologist (Janet, 1997). Accordingly the difficulty faced in ECG patient monitoring is the variability in morphology and timing across patients and within patients, of normal and ventricular beats (Waltrous and Towell, 1995). (Lagerholm et al., 2000) employed Self-Organizing Neural Networks (SelfOrganizing Maps or SOMs) in conjunction with Hermite Basis function for the purpose of beat clustering to identify and classify ECG complexes in arrhythmia. SOMs topological structure is a benefit in interpreting the data. The experimental results were claimed to outperform other supervised learning method that uses the same data. Analysis of NN as ECG analyzer also proves that NN is capable to deal with ambiguous nature of ECG signal (Silipo and Marchesi, 1998). Silipo and Marchesi use static and recurrent neural network (RNN) architectures for the classification tasks in ECG analysis for arrhythmia, myocardial ischemia and chronic alterations. Feedforward network with 8-24-14-1 architecture was employed as a classifier for ECG patient monitoring (Waltrous and Towell, 1995). The analysis indicated that the performance of the patient-adapted network was improve due to the ability of the modulated classifier to adjust the boundaries between classes, even though the distributions of beats were different for different patients. Multi layer RNN performance with 15-3-2 architecture have been studied and the performance of NN is compared with conventional algorithms for recognizing fetal heart rate abnormality (Lee et al., 1999). The study reveals that the performance of NN is exceptional compared to conventional systems even with adjusted thresholds. Applications in Medical Image Processing Image processing is one of the important applications in medicine as most of decision-making is made by looking at the images (Horsch et al., 1997). In general the segmentation of medical images is to find regions, which represent single anatomical structures (Poli and Valli, 1995). Poli and Valli employed Hopfield neural network for optimum segmentation of 2-D and 3-D medical images. The networks have been tested on synthetic images and on real tomographic and X-ray images. Wan Hussain Wan Ishak, Fadzilah Siraj, Abu Talib Othman (2001). Neural Network in Medical Application: A Review. Simposium Kedua Penyelidikan Pengurusan Siswazah (16-17 Oct. 2001). Universiti Utara Malaysia Ahmed and Farag (1998) uses two self-organizing maps (SOM) in two stages, selforganizing principal components analysis (SOPCA) and self-organizing feature map (SOFM) for automatic volume segmentation of medical images. They performed a statistical comparison of the performance of the SOFM with Hopfield network and ISODATA algorithm. The results indicate that the accuracy of SOFM is superior compare to both networks. In addition, SOFM was claimed to have advantage of ease implementation and guaranteed convergence. 3.0 BACKPROPAGATION LEARNING ALGORITHM Backpropagation (or backprop) algorithm is one of the well known algorithm in neural networks. Backpropagation algorithm has been popularized by Rumelhart, Hinton, and Williams in 1980s as a euphemism for generalized delta rule. Backpropagation of errors or generalized delta rule is a decent method to minimize the total squared error of the output computed by the net (Fausett, 1994). The introduction of backprop algorithm has overcome the drawback of previous NN algorithm in 1970s where single layer perceptron fail to solve a simple XOR problem. The aim of backpropagation algorithm is to train the net to achieve a balance between the ability to respond correctly to the input patterns that are used for training (memorization) and the ability to give reasonable (good) responses to input that is similar, but not identical, to that used in training (generalization) (Fausett, 1994). Sarle (1997) describes backpropagation algorithm as follows;  method for computing the gradient of the case-wise error function with respect to the weights for a feedforward network.  a training method that uses backpropagation to compute the gradient.  a feedforward network trained by backpropagation.