Computer-aided interpretation of dynamic magnetic resonance imaging reflects histopathology of invasive breast cancer.

OBJECTIVE: To perform a semiautomated software-based comparison of invasive breast carcinoma dynamic enhancement patterns in MR mammography with histological prognostic factors considering whole lesion volumes. MATERIAL AND METHODS: A total of 128 patients with 145 invasive breast carcinomas underwent dynamic MR mammography. Kinetic features from the invasive breast lesions were obtained using commercially available software to automatically assess volume enhancement characteristics of a manually chosen lesion. Findings were compared with histological factors determining tumour aggressiveness (lymph node status, LN; oestrogen/progesterone receptor (ER/PR) status; HER-2/neu status; tumour grade) by using nonparametric rank tests and binary logistic regression analysis (BLRA). RESULTS: Volume enhancement characteristics were significantly influenced by LN, ER/PR and HER-2/neu status (P < 0.05). BLRA implied that total lesion and plateau voxel volume were independent predictors of ER/PR and HER-2/neu status. Strongest initial enhancement predicted negative ER/PR, and time to peak of the most suspect curve was inversely correlated with positive LN status. On the other hand, no statistical significance could be observed between histological tumour grading and kinetic features. CONCLUSION: Histopathological criteria associated with poor prognosis lead to significantly more aggressive dynamic enhancement patterns in MR mammography. In this study, higher lesion volumes as well as higher and earlier initial enhancement were independent covariates predicting higher tumour aggressiveness.

Clinical MR-mammography: are computer-assisted methods superior to visual or manual measurements for curve type analysis? A systematic approach.

RATIONALE AND OBJECTIVES: Enhancement characteristics after administration of a contrast agent are regarded as a major criterion for differential diagnosis in magnetic resonance mammography (MRM). However, no consensus exists about the best measurement method to assess contrast enhancement kinetics. This systematic investigation was performed to compare visual estimation with manual region of interest (ROI) and computer-aided diagnosis (CAD) analysis for time curve measurements in MRM. MATERIALS AND METHODS: A total of 329 patients undergoing surgery after MRM (1.5 T) were analyzed prospectively. Dynamic data were measured using visual estimation, including ROI as well as CAD methods, and classified depending on initial signal increase and delayed enhancement. RESULTS: Pathology revealed 469 lesions (279 malignant, 190 benign). Kappa agreement between the methods ranged from 0.78 to 0.81. Diagnostic accuracies of 74.4% (visual), 75.7% (ROI), and 76.6% (CAD) were found without statistical significant differences. CONCLUSIONS: According to our results, curve type measurements are useful as a diagnostic criterion in breast lesions irrespective of the method used.