Using Neural Network Algorithms in Prediction of Mean Glandular Dose Based on the Measurable Parameters in Mammography
SUMMARY. In this paper we were investigate possibility of using neural network algorithms in prediction of mean glandular dose (MGD), based on the measurement of the compressed breast thickness (CBT) in patients population between 40 – 65 years. According to the available information’s this is the first time that is someone explored this possibility of using neural networks in prediction of MGD based on the information of CBT. The primary aim of this method is reducing unnecessary overdose of X–ray exposure to patients. The study population consisted of 63 patients (234 screens) from 40 to 64 year during routine mammographic control. The best results were achieved with Levenberg-Marquardt learning algorithm where correlation factor between neural network outputs and targets was R=0.845 (71.4%).