PDF

Keywords

Discriminate Analysis
neural networks
Medical Diagnosis of Oral Cancer

Abstract

Abstract The current research aims at comparing the classification of: the discriminate analysis and the neural networks as a technique of computer sciences and artificial intelligence. Then comparing their results with the actual medical diagnosis of oral cancer that has been done on a sample consisted of (37) outpatient to Al – Jumhoori and Al – Salam Hospitals. In order to achieve the aims of the study, the data - base have been programmed via visual basic (v.6). The process includes all the medical data sheets for diagnosis (attached 1). In the discriminate analysis, three standards consisted of discriminate function. Finding the cut of point, the (rate of error) on the samples that includes (14), cases diagnosed medically (23), cases out of disease. The research demonstrated the discriminate function that can be used to distinct among with and without infected through knowing, cut of point amounted (-21.3), additionally knowing the rate of error referred in the end of the research that give a power of the discriminate function. A simplified perception neural network has been used as a successful procedure in classifying the cases of infected and non infected patients. By using (θ = 0.75, & (η = 0.020) as a threshold value and training rate respectively. In comparing what has been done in discriminate analysis and neural networks perception via input a number new cases then classifying them according to the medical diagnosis done, showed a great coincidence in results .
https://doi.org/10.33899/tanra.2008.161736
  PDF