BI-RADS Category 5 Assessments at Diagnostic Breast Imaging:Outcomes Analysis Based on Lesion Descriptors

Improving Computer-aided Detection for Digital Breast Tomosynthesis by Incorporating Temporal Change

Purpose To develop a deep learning algorithm that uses temporal information to improve the performance of a previously published framework of cancer lesion detection for digital breast tomosynthesis. Materials and Methods This retrospective study analyzed the current and the 1-year-prior Hologic digital breast tomosynthesis screening examinations from eight different institutions between 2016 and 2020. The dataset contained 973 cancer and 7123 noncancer cases. The front end of this algorithm was an existing deep learning framework that performed single-view lesion detection followed by ipsilateral view matching. For this study, PriorNet was implemented as a cascaded deep learning module that used the additional growth information to refine the final probability of malignancy. Data from seven of the eight sites were used for training and validation, while the eighth site was reserved for external testing. Model performance was evaluated using localization receiver operating characteristic curves. Results On the validation set, PriorNet showed an area under the receiver operating characteristic curve (AUC) of 0.931 (95% CI: 0.930, 0.931), which outperformed both baseline models using single-view detection (AUC, 0.892 [95% CI: 0.891, 0.892]; P < .001) and ipsilateral matching (AUC, 0.915 [95% CI: 0.914, 0.915]; P < .001). On the external test set, PriorNet achieved an AUC of 0.896 (95% CI: 0.885, 0.896), outperforming both baselines (AUC, 0.846 [95% CI: 0.846, 0.847]; P < .001 and AUC, 0.865 [95% CI: 0.865, 0.866]; P < .001, respectively). In the high sensitivity range of 0.9 to 1.0, the partial AUC of PriorNet was significantly higher (P < .001) relative to both baselines. Conclusion PriorNet using temporal information further improved the breast cancer detection performance of an existing digital breast tomosynthesis cancer detection framework. Keywords: Digital Breast Tomosynthesis, Computer-aided Detection, Breast Cancer, Deep Learning © RSNA, 2024 See also commentary by Lee in this issue.

A New Practical Decision Rule to Better Differentiate BI-RADS 3 or 4 Breast Masses on Breast Ultrasound

OBJECTIVES

The BI-RADS classification provides a standardized way to describe ultrasound findings in breast cancer diagnostics. However, there is little information regarding which BI-RADS descriptors are most strongly associated with malignancy, to better distinguish BI-RADS 3 (follow-up imaging) and 4 (diagnostic biopsy) breast masses.

METHODS

Patients were recruited as part of an international, multicenter trial (NCT02638935). The trial enrolled 1294 women (6 excluded) categorized as BI-RADS 3 or 4 upon routine B-mode ultrasound examination. Ultrasound images were evaluated by three expert physicians according to BI-RADS. All patients underwent histopathological confirmation (reference standard). We performed univariate and multivariate analyses (chi-square test, logistic regression, and Krippendorff's alpha).

RESULTS

Histopathologic evaluation showed malignancy in 368 of 1288 masses (28.6%). Upon performing multivariate analysis, the following descriptors were significantly associated with malignancy (P < .05): age ≥50 years (OR 8.99), non-circumscribed indistinct (OR 4.05) and microlobulated margin (OR 2.95), nonparallel orientation (OR 2.69), and calcification (OR 2.64). A clinical decision rule informed by these results demonstrated a 97% sensitivity and missed fewer cancers compared to three physician experts (range of sensitivity 79-95%) and a previous decision rule (sensitivity 59%). Specificity was 44% versus 22-83%, respectively. The inter-reader reliability of the BI-RADS descriptors and of the final BI-RADS score was fair-moderate.

CONCLUSIONS

A patient should undergo a diagnostic biopsy (BI-RADS 4) instead of follow-up imaging (BI-RADS 3) if the patient is 50 years or older or exhibits at least one of the following features: calcification, nonparallel orientation of mass, non-circumscribed margin, or posterior shadowing.

A National Chinese Survey on Ultrasound Feature Interpretation and Risk Assessment of Breast Masses Under ACR BI-RADS

Purpose

Through this nationwide survey on ACR BI-RADS including ultrasound images of 10 selected breast lesions, we aimed to learn about consistency in feature interpretation and assessment categories and to identify factors that might contribute to inconsistencies, thereby promoting the application of BI-RADS in China.

Materials and Methods

The survey was delivered through a self-developed website about blinded image interpretation and was released to the public through online platforms and social media. A total of 10 representative lesions were selected by an experienced radiologist to gather information about the general practice of BI-RADS lexicons and categories. The Kappa statistic, the chi-squared test, and descriptive statistics were used for data analysis.

Results

Nine hundred ultrasound workers completed the questionnaire, coming from all provinces and major cities in China. They had different positions, grades of work organization, and seniority. The interrater agreement of BI-RADS features was fair to substantial (kappa value: 0.37–0.66). For BI-RADS categories, the highest agreement was observed in the typical benign group (average constituent rate = 74.78%), and generally lower agreement was observed in the typical malignant (average constituent rate = 36.03%) and suspicious groups (average constituent rate = 39.02%).

Conclusion

We found inconsistencies in BI-RADS applications, providing direction for image feature research using big data. Therefore, we call for more efforts to improve the consistency of BI-RADS application and provide an evidence-based basis for identifying benign and malignant lesions by sonographic features.

Application of MRI Radiomics-Based Machine Learning Model to Improve Contralateral BI-RADS 4 Lesion Assessment

Objective

This study aimed to explore the potential of magnetic resonance imaging (MRI) radiomics-based machine learning to improve assessment and diagnosis of contralateral Breast Imaging Reporting and Data System (BI-RADS) category 4 lesions in women with primary breast cancer.

Materials and Methods

A total of 178 contralateral BI-RADS 4 lesions (97 malignant and 81 benign) collected from 178 breast cancer patients were involved in our retrospective dataset. T1 + C and T2 weighted images were used for radiomics analysis. These lesions were randomly assigned to the training (n = 124) dataset and an independent testing dataset (n = 54). A three-dimensional semi-automatic segmentation method was performed to segment lesions depicted on T2 and T1 + C images, 1,046 radiomic features were extracted from each segmented region, and a least absolute shrinkage and operator feature selection method reduced feature dimensionality. Three support vector machine (SVM) classifiers were trained to build classification models based on the T2, T1 + C, and fusion image features, respectively. The diagnostic performance of each model was evaluated and tested using the independent testing dataset. The area under the receiver operating characteristic curve (AUC) was used as a performance metric.

Results

The T1+C image feature-based model and T2 image feature-based model yielded AUCs of 0.71 ± 0.07 and 0.69 ± 0.07 respectively, and the difference between them was not significant (P > 0.05). After fusing T1 + C and T2 imaging features, the proposed model’s AUC significantly improved to 0.77 ± 0.06 (P < 0.001). The fusion model yielded an accuracy of 74.1%, which was higher than that of the T1 + C (66.7%) and T2 (59.3%) image feature-based models.

Conclusion

The MRI radiomics-based machine learning model is a feasible method to assess contralateral BI-RADS 4 lesions. T2 and T1 + C image features provide complementary information in discriminating benign and malignant contralateral BI-RADS 4 lesions.