Data Mining Technology In The Diagnosis Of Breast X-ray

(mammography,data mining technology R4)The Discussion of Application with Data Mining Technology in MammographyObjective:To discuss the value of Data Mining technology in mammography screenning and diagnosis.Methods:We collected 1264 pathologically confirmed cases.Radiologist-extracted mammographic features,pathologic result were used as input features for Decision tree and Logoistic Regression models to predict benign or malignant.Mammographic findings using BI-RADS for description.Then we compared the sensitivity and specificity between two models,Results:The correct rate had significant differences between two models in mass diagnosis(89.18%, 91.78%;p=0.135) and had same capacity as the doctor.While they had no significant differences in calcification diagnosis(80.11%,83.21%;p=0.337),still equally with doctors.To all kinds of mammographic features,there were no significant differences between two models(87.25%,88.54%,p=0.138) and showed lower capacity than doctor.For the category4 cases,two models showed lower correct rate,and had no significant differences(71.91%,75.02%;p=0.387).Conclusion:Results showed that data mining technology can play a guiding role in mammography and has potential application value.The questions founded in the process should be further studied.