Expert Systems in Medicine
Passent Mohamed El-Kafrawy;
Abstract
The thesis is devoted to test that fuzzy subsets will improve the performance of a Neural Network (NN) model. The thesis also tries to increase the efficiency of Computers as diagnostic tools for the problem of: liver diseases; the most significant issue for the Egyptian national health.
Computerized Ultrasound tissue characterization has become an objective means for diagnosis of liver diseases. Tissue characterization is a'
hot subject in this era because of liver cirrhosis and hepatitis. Ultrasound imaging is still a subjective matter that is insensitive to the elasticity of the tissues. In other words, it cannot differentiate between soft and hard tissues. Computerized systems add a quantitative measurement to the diagnosis process.
Diagnosis of diffuse liver diseases, in this thesis, includes differentiating between Cirrhotic and Fatty liver. The visual criterion for differentiating diffused diseases is rather confusing and highly dependent upon the sonographer experience, this often causes a bias effect in the diagnostic procedure and limits its objectivity and reproducibility. The need for a computerized tissue characterization is thus justified to assist quantitatively the sonographer for the accurate differentiation and to minimize the degree of risk from erroneous interpretation.
In this work we present the use of fuzzy subsets in conjunction with NN in medical diagnosis. NN and fuzzy theory are hot research subjects nowadays, their importance in real life applications appears to be significant; in contrast to traditional AI and ES. Both have also some common features that are more natural and closer to human language and thoughts. Thus they are new tools for efficient pattern recognition and classification problems.
The NN model we used in this thesis is the Kohonen model, for its advantages over traditional NN models. Actually, the Kohonen system is an unsupervised technique, that learns from unknown examples. In the liver diagnostic problem, data samples are scarce, very divergent in nature, inhomogenous and the number of examples of a certain class are limited and expensive to acquire. Thus, supervised NN models cannot help effectively in diagnosis. '
Computerized Ultrasound tissue characterization has become an objective means for diagnosis of liver diseases. Tissue characterization is a'
hot subject in this era because of liver cirrhosis and hepatitis. Ultrasound imaging is still a subjective matter that is insensitive to the elasticity of the tissues. In other words, it cannot differentiate between soft and hard tissues. Computerized systems add a quantitative measurement to the diagnosis process.
Diagnosis of diffuse liver diseases, in this thesis, includes differentiating between Cirrhotic and Fatty liver. The visual criterion for differentiating diffused diseases is rather confusing and highly dependent upon the sonographer experience, this often causes a bias effect in the diagnostic procedure and limits its objectivity and reproducibility. The need for a computerized tissue characterization is thus justified to assist quantitatively the sonographer for the accurate differentiation and to minimize the degree of risk from erroneous interpretation.
In this work we present the use of fuzzy subsets in conjunction with NN in medical diagnosis. NN and fuzzy theory are hot research subjects nowadays, their importance in real life applications appears to be significant; in contrast to traditional AI and ES. Both have also some common features that are more natural and closer to human language and thoughts. Thus they are new tools for efficient pattern recognition and classification problems.
The NN model we used in this thesis is the Kohonen model, for its advantages over traditional NN models. Actually, the Kohonen system is an unsupervised technique, that learns from unknown examples. In the liver diagnostic problem, data samples are scarce, very divergent in nature, inhomogenous and the number of examples of a certain class are limited and expensive to acquire. Thus, supervised NN models cannot help effectively in diagnosis. '
Other data
| Title | Expert Systems in Medicine | Other Titles | النظم الخبيرة في الطب | Authors | Passent Mohamed El-Kafrawy | Issue Date | 1996 |
Recommend this item
Similar Items from Core Recommender Database
Items in Ain Shams Scholar are protected by copyright, with all rights reserved, unless otherwise indicated.