Diagnosis of Breast Cancer using Decision Tree

• Main Authors: Hsin-Ju Wu, 吳欣儒
• Other Authors: Ja-chih Fu
• Format: Others
• Language: zh-TW
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/61349441866949192457

碩士 === 國立雲林科技大學 === 工業工程與管理系 === 102 === Abstract Applying Decision science in medicine developed rapidly in this decade, such as diagnosis of disease with feature and lab test that are imprecise and uncertain same as thinking of human. Many researches applied decision tree for discovering knowledge has been earned relatively achievement. Physician need to consider relation inter-factors when make decision for disease of multi-factors. Decision Tree is a excellent decision support tool with characteristics of easy comprehension, explanation, and to do with ordinary and nominal simultaneously. Breast cancer is the most common female cancer and became top cancer in the world. Women with breast cancer rate increasing by year. Women should examine breast once a year to detect whether has breast cancer for more earlier to find threat of cancer due to life style changed and female late marriage. Mammography data has readable numerical and nominal data suit for analyzing breast cancer with decision tree. This research collect 300 positive mass traced patterns as training data for decision tree, there were 11 rules were extracted and a fold 83 patterns as test dataset for pruned decision tree to validate accuracy. At result, decision tree achieved 93.1% accuracy and back propagation artificial neural network achieved 88% accuracy. Our research proves decision tree has excellent classification capability, and the extracted rules known by human to assist physician making decision of disease.