You are on page 1of 8

Neural Virtual Physician

Shivani Joshi
Astt. Professor, GNIM

Abstract
This paper discusses about the neural networks which is somewhat able to think like human
brain from the patterns and help in various fields. This paper elaborates the neural network
capabilities in biological system and tells how it helps diagnosing the chronic diseases of the
patients at early stage and helps the doctors , experts & practitioners to get the solution of the
disease by the case history of the patient
1. Introduction

A neural network or neural computing, computer architecture modeled upon the human
brain’s interconnected system of neurons. Neural networks imitate the brain's ability to sort
out patterns and learn from trial and error, discerning and extracting the relationships that
underlie the data with which it is presented. Most neural networks are software simulations
run on conventional computers. In neural computers, transistor circuits serve as the neurons
and variable resistors act as the interconnection between axons and dendrites. As human brain
consists of ten billion nerves so is the neural network consisting of integrated circuits, with
1,024 silicon "neurons," has also been developed. Each neuron in the network has one or
more inputs and produces an output; each input has a weighting factor, which modifies the
value entering the neuron. The neuron mathematically manipulates the inputs, and outputs the
result. The neural network is simply neurons joined together, with the output from one neuron
becoming input to others until the final output is reached.

Neural networks are good at providing very fast, very close approximations of the correct
answer. Neural networks recognize shapes or patterns, learns from experience, or sorting
relevant data from irrelevant. Their applications can be categorized into classification,
recognition and identification, assessment, monitoring and control, and forecasting and
prediction. They are also well suited for handwriting recognition, foreign language
translation, process control, financial forecasting, medical data interpretation, artificial
intelligence research, and parallel processing implementations of conventional processing
tasks. In an ironic reversal, neural networks are being used to model disorders of the brain in
an effort to discover better therapeutic strategies.

Discovery
A computer program operates in a manner analogous to the natural neural network in the
brain. The neural networks was developed in 1943 by the neurophysiologist Warren
McCulloch and the mathematician Walter Pitts. In 1954 Belmont Farley and Wesley Clark
succeeded in running the first simple neural network. The primary goal of neural networks is
their ability to emulate the brain’s pattern-recognition skills. In business sector neural
networks have been used to make investment decisions, recognize handwriting, and even
detect bombs.
2 Medical Diagnosis

The major problem in medical field is to diagnose disease. Human being always make
mistake and because of their limitation, diagnosis would give the major issue of human
expertise. One of the most important problems of medical diagnosis, in general, is the
subjectivity of the specialist. It can be noted, in particular in pattern recognition activities,
that the experience of the professional is closely related to the final diagnosis. This is due to
the fact that the result does not depend on a systematized solution but on the interpretation of
the patient's signal (Lanzarini and Giusti, 1999).
Brause (2001) highlighted that almost all the physicians are confronted during their formation
by the task of learning to diagnose. Here, they have to solve the problem of deducing
certain diseases or formulating a treatment based on more or less specified observations
and knowledge. For this task, certain basic difficulties have to be taken into account:-

• The basis for a valid diagnosis, a sufficient number of experienced cases, is reached
only in the middle of a physician’s career and is therefore not yet present at the end of
the academic formation.

• This is especially true for rare or new diseases where also experienced physicians are
in the same situation as newcomers.

• Principally, humans do not resemble statistic computers but pattern recognition


systems. Humans can recognize patterns or objects very easily but fail when
probabilities have to be assigned to observations.

Brause(2001) also give an example of a study in the year 1971 showed these basic facts
in the medical area. This study had shown that human have many limitations in diagnosis.
The results of this experiment were as follows:-

• Best human diagnosis (most experienced physician): 79.7%


• Computer with expert data base: 82.2%
• Computer with 600 patient data: 91.1%

From this result we can see that humans can not ad hoc analyze complex data without errors.

3.0 Neural Network Capabilities

Here it is discussed how Neural Network approach can diagnose disease using patient
medical data such as breast cancer, heart failure, medical images, acidosis diseases, and lung
cancer.

3.1 Diagnose Breast Cancer

Breast cancer is the second largest cause of cancer deaths among women. The automatic
diagnosis of breast cancer is an important, real-world medical problem. A major class of
problems in medical science involves the diagnosis of disease, based upon various tests
performed upon the patient. When several tests are involved, the ultimate diagnosis may be
difficult to obtain, even for a medical expert. This has given rise, over the past few decades,
to computerized diagnostic tools, intended to aid the physician in making sense out of the
confusion of data (Kiyan and Yildirim, 2003).
Neural network have been applied to breast cancer diagnosis.(Kiyan and Yildirim, 2003)
employed Radial Basis Function, General Regression Neural Network and Probabilistic
Neural Network in order to get the suitable result. From overall results, it is seen that the
most suitable neural network model for classifying Wisconsin Breast Cancer data is General
Regression Neural Network. This work also indicates that statistical neural networks can be
effectively used for breast cancer diagnosis to help oncologists.

Figure 1. Development of cancer

3.2 Diagnose heart failure

Making prognosis for patients with congestive heart failure is difficult due to the complex
nature of this multisystem disease. No single criterion helps to identify patients at risk, and a
combination of several prognostic parameters is recommended (Cowburn et al., 1998).
Neural networks are associative self learning techniques with the ability to identify
multidimensional relationships and perform pattern recognition in non-linear domains.
Atieza et al., (2003) identified that classification is the best result for this cases.

3.4 Diagnose Acidosis Diseases

Ultsch et al. (1995) used the capability of neural network to diagnose acidosis diseases by
using knowledge based system in their hybrid system. The data set consists of 11 attributes
originating from the blood analysis. Several classification methods according to (Deichsel
and Trampisch, 1985) were used to explain these data. The Neural Network together with the
UMatrix method was able to classify the data into the subcategories healthy, lacacidemia,
metabolical acidosis, respiratory acidosis and one patient with cerebral deficiency. They used
rule generation module to extracted rules out of the Neural Network, which were described
by 4 or 5 attributes resembling more closely the decisions made by medical experts (Ultsch
and Li, 1993).

3.5 Diagnose Lung Cancer

Lung cancer is another diseases that commonly known as a deadly disease in the world.
Many patients suffer from this disease. Early detection of this disease is very important to
prevent this disease. Expertise have to measures for early stage lung cancer diagnosis mainly
includes those utilizing X-ray chest films, CT, MRI, isotope, bronchoscopy, and needle
biopsies. According to Zhou et al. (2001) at present, the specimens of needle biopsies are
usually analyzed by experienced pathologists. Since senior pathologists are rare, reliable
pathological diagnosis is not always available.
Zhou et al. (2001) named his an automatic pathological diagnosis procedure named Neural
Ensemble based Detection (NED). It is proposed and realized in an early stage Lung Cancer
Diagnosis System (LCDS). NED utilizes an artificial neural network ensemble to identify
cancer cells in the images of the specimens of needle biopsies obtained from the bodies of the
subjects to be diagnosed. A fast adaptive neural network classifier has been used to identify
the lung cancer cell. Zhou et al. (2001) stated that a fast adaptive neural classifier that
performs one-pass incremental learning with fast speed and high accuracy and does not
require the user manually set up the number of hidden units.

Diagnose by Medical Images

Neural networks are extremely useful, since not only are they capable of recognizing patterns
with the aid of the expert, but also of generalizing the information contained in the input data,
thus showing relations which are a prior complex. (Laura and Armando, 1999) combined the
processing of digital image and neural network to carry out the required recognition and
classification. As a result, the solution to the problem can be divided in two parts: the
segmentation of different elements, and their subsequent classification. In this case good
results good results have been obtained thanks to the definition of a new clustering algorithm
based on e re-definition of the input image. As for the classification stage, different solutions
using neural networks have been compared, the results obtained being correct, with an error
smaller than 10%.
Neural Networks are used in pattern recognition because of their ability to learn and to store
knowledge. Because of their 'parallel' nature can achieve, Neural network can achieve very
high computation rates which is vital in application like telemedicine (Siganos, 1995).

4.0 Applications of Neural Network in Medicine

In this paper the discussion of applications of neural network in medical applications is


divided into several domain that are applications in basic sciences, clinical medicine, signal
processing and interpretation and medical image processing.

4.1 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 for example
can help clinician to control the diabetes mellitus. By using feed forward NN for predicting
the time course of blood glucose levels from the complex interaction of glucose counter
regulatory hormones and insulin.
Multi-Layer Perceptron (MLP) with sigmoidal Feed-Forward and standard Back-Propagation
(BP) learning algorithm was employed as a forecaster for bacteria-antibiotic interactions of
infectious diseases The conclusion is that one-month forecaster produces output correct to
within occurrences of sensitivity. However, predictions for the two-month and three-month
are less accurate.

4.2 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 clinic pathological 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 predicting length of stay
(LOS) more than seven days for acute pancreatitis patients’. Their study indicates that NN
achieve the highest sensitivity (75%) for predicting LOS more than seven. 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).

4.3 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 (Self-Organizing 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 improved 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 had 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.
Figure 2. Signal processing

4.4 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. 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.
Ahmed and Farag (1998) uses two self-organizing maps (SOM) in two stages, self-organizing
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.

Figure 3. Image Processing

4.5 Applications in Haematology

There is an attempt to create an expert system to diagnose classes of anaemia and report
presumptive diagnoses directly on the haematology form (Birndort et al., 1996). The purpose
is to simulate the processes of human experts that can reliably achieve diagnostic separability
by pattern analysis. In doing this, they constructed a hybrid expert system combining rule-
based and artificial neural network (ANN) models to evaluate microcytic anaemia in a 3-
layered program using haematocrit (HCT), mean corpuscular volume (MCV), and coefficient
of variation of cell distribution width (RDWcv) as inputs. These measurements are available
as standard output on most haematology analyzers. Three categories of microcytic anaemia
were considered, iron deficiency (IDA), haemo globinopathy (HEM), and anaemia of chronic
disease (ACD). The performance of the model was evaluated with actual case data. The
results show that the model was successful in correctly classifying 96.5% of 473 documented
cases of microcytic anaemia and anaemia of chronic disease. This result exhibits sufficient
accuracy to be considered for use in reporting microcytic anaemia diagnoses on haematology
forms.

The leukocyte-vessel wall interactions are studied in post capillary vessels by intravital video
microscopy during in vivo animal experiments (Egmont-Petersen et al., 2000). Sequences of
video images are obtained and digitized with a frame grabber. A method for automatic
detection and characterization of leukocytes in the video images is developed. Individual
leukocytes are detected using a neural network that is trained with synthetic leukocyte images
generated using a novel stochastic model. This model makes it feasible to generate images of
leukocytes with different shapes and sizes under various lighting conditions. Experiments
indicate that neural networks trained with the synthetic leukocyte images perform better than
networks trained with images of manually detected leukocytes. The best performing neural
network trained with synthetic leukocyte images resulted in an 18% larger area under the
ROC curve than the best performing neural network trained with manually detected
leukocytes.

Conclusion

Neural network has been proven of their capabilities in many domains such as medical
application. Neural network become powerful tool in medical diagnosis. Neural network
show that experience from expertise is not enough in diagnosis. Nowadays, physicians
combined this opportunity that give by neural network and their expertise to detect early
stage of patients disease.
NN have been successfully implemented in many applications including medicine. NN,
which simulates the function of human biological neuron, has potential of ease
implementation in many applications domain. The main consideration of NN implementation
is the input data. Once the network is train, the knowledge could be applied to all cases
including the new cases in the domain. Studies have shown that NN predictive capability is a
useful capability in medical application. Such capability could be used to predict patient
condition based on the history cases. The prediction could help doctor to plan for a better
medication and provide the patient with early diagnosis.

Reference

1. Abidi S. S. R., and Goh, A. (1998). Neural Network Based Forecasting of Bacteria-
Antibiotic Interactions for Infectious Disease Control. In 9th World Congress on Medical
Informatics (MedInfo’98), Seoul August 18-22.
2 Ahmed, M. N., and Farag, A. A. (1998). Two-stage Neural Network for Medical Volume
Segmentation. Accepted for Publication in the Journal of Pattern Recognition Letters,
1998.
3. Bottaci, L., and Drew, P. J. (1997). Artificial Neural Networks Applied to Outcome
Prediction for Colorectal Cancer Patients in Separate Institutions. Lancet, Vol. 350,
Issue 9076, pp. 469-473.
4. Dybowski, R. (2000). Neural Computation in Medicine: Perspective and Prospects. In
Malmgren, H., Borga, M., Niklasson, L. (eds.) Proceedings of the ANNIMAB-1
Conference (Artificial Neural Networks in Medicine and Biology), Goteborg, 13-16 May
2000. Springer, pp. 26-36.
5. Fausett, L. (1994). Fundamentals of Neural Network: Architectures, Algorithms and
Applications. Prentice Hall; Englewood Cliffs.
6. Janet, F. (1997). Artificial Neural Networks Improve Diagnosis of Acute Myocardial
Infarction. Lancet, Vol. 350, Issue 9082, pp. 935.
7. Kemsley, D. H., and Martinez, T. R. (1992). A Survey of Neural Network Research and
Fielded Applications. In International Journal of Neural Networks: Research and
Applications, Vol. 2, No. 2/3/4, pp. 123-133.
8. Lagerholm, M., Peterson, C., Braccini, G., Edenbrandt, L., and Sörnmo, L. (2000).
Clustering ECG Complexes Using Hermite Functions and Self-Organizing Maps. IEEE
Transactions on Biomedical Engineering 47, 838-848
9. Ohno-Machado, L. (1996). Medical Applications of Artificial Neural Networks:
Connectionist Model of Survival. Ph.D Dissertation. Stanford University.
10. Partridge, D., Abidi, S. S. R., and Goh, A. (1996). Neural Network Applications in
Medicine. Proceedings of National Conference on Research and Development in
Computer Science and Its Applications (REDECS’96), University Pertanian Malaysia:
Kuala Lumpur, pp. 20 - 23.