Dynamic multidetector CT of breast tumors: diagnostic features and comparison with conventional techniques

We sought to analyze the features of breast tumors as revealed on dynamic multidetector CT (MDCT), to develop descriptors for these features, and to compare the performance of MDCT with the performance of other techniques used in the depiction of tumors. SUBJECTS AND METHODS. MDCT was performed in 149 women with suspected breast tumors, and 173 breast lesions were detected. These breast lesions were classified as either mass or nonmass enhancing lesions. For mass lesions, the margin, shape, and enhancement patterns were evaluated. For nonmass enhancing lesions, the distribution of enhancement and the types of time-density curve patterns were evaluated. MDCT was compared with mammography and sonography as a method of revealing breast tumors.

Of the 173 breast lesions detected, 150 were mass lesions, 131 (87%) of which were malignant. Of the 23 nonmass enhancing lesions, 21 (91%) were malignant. The most highly predictive features for lesion malignancy were an irregular margin (100%), an irregular shape (99%), and rim enhancement (100%). Similar features were the most accurate signs of malignancy--a spiculated and irregular margin (90%). On time-density curves, the washout and plateau patterns showed high positive predictive value (93%) and sensitivity (91%) for malignancy. However, these patterns had low negative predictive value (42%) and specificity (48%). Seven breast lesions that could not be detected on mammography or sonography were identified on MDCT. MDCT more accurately revealed the margin of the tumor invasion in 11 breast tumors than did mammography or sonography.

The features revealed on MDCT can help to distinguish benign lesions from carcinomas. MDCT can add to the data obtained with mammography or sonography in patients with suspected breast tumors.