Differentiation breast mass lesions between benign and malignancy by contrast-enhanced dynamic MRI classification

Objective: The purpose of this paper is to grade hypervascularized mass lesions in CE MR imaging of the breast by analyzing morphology and dynamic characteristics. Methods: Criteria for the evaluation of breast MR findings were used, including the analysis of shape, margin, internal enhancement characteristics, initial enhancement phase, and delayed phase enhancement pattern. A total of 188 hypervascularized mass lesions were analyzed, and scores were given from 1 to 10 points and then the mass was classified into one of five categories. Next, the score results and pathology results were used to calculate the ROC curve and to find out the optimal benign and malignant diagnosis points. According to these results, the lesions were classified as benign or malignant. The sensitivity and specificity of this classification system were also calculated. Results: Out of the 188 hypervascularized mass lesions, 91 lesions were confirmed to be malignant by pathologic diagnosis and 97 to be benign. The area under the ROC curve was 0.938±0.016. This curve was used to calculate the optimal point of differentiation diagnosis, and it was found to be a score of 5. The 188 lesions were classified into grade Ⅱ of 24 lesions, grade Ⅲ of 72, grade Ⅳ of 54, and grade V of 38. Sensitivity for the detection of malignancy using this classification system was 87.91% and specificity was 87.62%. If the three cases with false negative results of ductal carcinoma in situ of' grade Ⅲ were excluded, the specificity increased to 90.90% . Conclusion: The classification based on a multifactorial analysis is very helpful in the objective interpretation of breast CE MRI. (authors)