Original Article

Case Based Reasoning Application to Detect Malignant lesions on Breast Mammograms

Abstract

• Breast cancer is among the most common women cancer and major cause of cancer produced women death around the globe. Mammograms have a high rate of missed tumours, or "false negatives." More than 10 percent of malignant tumours in mammograms can not be detected in women over 50. Case-based reasoning (CBR) is method which use solutions of already solved problems to devise solution of coming problem. We used CBR to differentiate benign lesions from malignant of breast cancer. CBR reported precision and recall for benign cases equal to 0.87 and 0.7 respectively. For Classification of malignant cases, CBR reported a precision of 0.75 and a recall of 0.9. In an effort to increase the precision and recall, principal component analysis was applied to the data and cases were prepared using the transformed data. When CBR was implemented, precision increased by 20% and recall by 11% for malignant cases. Precision increased by 15% and recall by 28.5% for benign cases. The results obtained by CBR was compared with different classification techniques, these techniques include BaysNet, NaiveBayes, RBFNetwork, VotedPerception, Bagging, AdaboostM1, ADTree, J48 and Conjunctive Rule. The CBR based classification produced better results as compared to all the other methods in terms of precision, recall , false negative rate, true positive rate and F-measure for malignant cases.
  
  

• Keywords: Mammography, breast cancer, Case-based reasoning Similarity measure,Early detection ,Recall

Back to All Articles