Clinical value of radiomics and machine learning in breast ultrasound: a multicenter study for differential diagnosis of benign and malignant lesions.

OBJECTIVES: We aimed to assess the performance of radiomics and machine learning (ML) for classification of non-cystic benign and malignant breast lesions on ultrasound images, compare ML's accuracy with that of a breast radiologist, and verify if the radiologist's performance is improved by using ML. METHODS: Our retrospective study included patients from two institutions. A total of 135 lesions from Institution 1 were used to train and test the ML model with cross-validation. Radiomic features were extracted from manually annotated images and underwent a multistep feature selection process. Not reproducible, low variance, and highly intercorrelated features were removed from the dataset. Then, 66 lesions from Institution 2 were used as an external test set for ML and to assess the performance of a radiologist without and with the aid of ML, using McNemar's test. RESULTS: After feature selection, 10 of the 520 features extracted were employed to train a random forest algorithm. Its accuracy in the training set was 82% (standard deviation, SD, ± 6%), with an AUC of 0.90 (SD ± 0.06), while the performance on the test set was 82% (95% confidence intervals (CI) = 70-90%) with an AUC of 0.82 (95% CI = 0.70-0.93). It resulted in being significantly better than the baseline reference (p = 0.0098), but not different from the radiologist (79.4%, p = 0.815). The radiologist's performance improved when using ML (80.2%), but not significantly (p = 0.508). CONCLUSIONS: A radiomic analysis combined with ML showed promising results to differentiate benign from malignant breast lesions on ultrasound images. KEY POINTS: • Machine learning showed good accuracy in discriminating benign from malignant breast lesions • The machine learning classifier's performance was comparable to that of a breast radiologist • The radiologist's accuracy improved with machine learning, but not significantly.

Preliminary Results of a Simplified Breast MRI Protocol to Characterize Breast Lesions: Comparison with a Full Diagnostic Protocol and a Review of the Current Literature.

RATIONALE AND OBJECTIVES: This study aimed to investigate whether a simplified breast magnetic resonance imaging (MRI) protocol consisting of a localizer, one precontrast sequence, and three time-point postcontrast sequences (at 28 seconds, 84 seconds and 252 seconds after the contrast agent administration) is suitable for the characterization of breast lesions as compared to a full diagnostic protocol (FDP). This study also aimed to review the current literature concerning abbreviated breast MRI protocols and offer an alternative protocol. MATERIALS AND METHODS: Breast magnetic resonance (MR) examinations with detected breast lesions of 98 patients were retrospectively evaluated. Two expert radiologists in consensus reviewed the simplified breast protocol (SBP) first and only thereafter the regular FDP, recording a diagnosis for each detected lesion for both protocols. Receiver operating characteristic curve analysis was performed to determine the diagnostic performance of the SBP compared to the standard FDP. A revision of the previously reported abbreviated breast magnetic resonance protocols was also carried out. RESULTS: A total of 180 lesions were identified; of these, 110 (61%) were malignant and 70 (39%) were benign. Of the 110 malignant lesions, 86 (78%) were invasive ductal carcinoma, 18 (16%) were invasive lobular carcinoma, and 6 (6%) were ductal carcinoma in situ. Areas under the curve for the receiver operating characteristic curves for the SBP vs the FDP were equivalent (0.98 vs 0.99, respectively; P = 0.76). The SBP could be performed in approximately 6 minutes and 58 seconds, compared to 14 minutes and 48 seconds for the FDP. CONCLUSIONS: An SBP protocol including a late postcontrast time point is accurate for the characterization of breast lesions and was comparable to the standard FDP protocol, allowing a potential reduction of the total acquisition and interpretation times.