Department of Radiology, Breast Imaging Service, Memorial Sloan Kettering Cancer Center, New York, New York, USA.
Department of Radiology, Neuroradiology Service, Memorial Sloan Kettering Cancer Center, New York, New York, USA.
J Magn Reson Imaging. 2022 Oct;56(4):1068-1076. doi: 10.1002/jmri.28111. Epub 2022 Feb 15.
Background parenchymal enhancement (BPE) is assessed on breast MRI reports as mandated by the Breast Imaging Reporting and Data System (BI-RADS) but is prone to inter and intrareader variation. Semiautomated and fully automated BPE assessment tools have been developed but none has surpassed radiologist BPE designations.
To develop a deep learning model for automated BPE classification and to compare its performance with current standard-of-care radiology report BPE designations.
Retrospective.
Consecutive high-risk patients (i.e. >20% lifetime risk of breast cancer) who underwent contrast-enhanced screening breast MRI from October 2013 to January 2019. The study included 5224 breast MRIs, divided into 3998 training, 444 validation, and 782 testing exams. On radiology reports, 1286 exams were categorized as high BPE (i.e., marked or moderate) and 3938 as low BPE (i.e., mild or minimal).
FIELD STRENGTH/SEQUENCE: A 1.5 T or 3 T system; one precontrast and three postcontrast phases of fat-saturated T1-weighted dynamic contrast-enhanced imaging.
Breast MRIs were used to develop two deep learning models (Slab artificial intelligence (AI); maximum intensity projection [MIP] AI) for BPE categorization using radiology report BPE labels. Models were tested on a heldout test sets using radiology report BPE and three-reader averaged consensus as the reference standards.
Model performance was assessed using receiver operating characteristic curve analysis. Associations between high BPE and BI-RADS assessments were evaluated using McNemar's chi-square test (α* = 0.025).
The Slab AI model significantly outperformed the MIP AI model across the full test set (area under the curve of 0.84 vs. 0.79) using the radiology report reference standard. Using three-reader consensus BPE labels reference standard, our AI model significantly outperformed radiology report BPE labels. Finally, the AI model was significantly more likely than the radiologist to assign "high BPE" to suspicious breast MRIs and significantly less likely than the radiologist to assign "high BPE" to negative breast MRIs.
Fully automated BPE assessments for breast MRIs could be more accurate than BPE assessments from radiology reports.
4 TECHNICAL EFFICACY STAGE: 3.
乳腺磁共振成像报告中要求对背景实质强化（BPE）进行评估，这是乳腺成像报告和数据系统（BI-RADS）规定的，但BPE 评估容易受到读者间和读者内的差异影响。已经开发出半自动化和全自动化的 BPE 评估工具，但没有一种工具能够超越放射科医生的 BPE 分级。
开发一种用于自动 BPE 分类的深度学习模型，并将其性能与当前的标准护理放射学报告 BPE 分级进行比较。
回顾性。
连续的高危患者（即乳腺癌终生风险>20%），他们在 2013 年 10 月至 2019 年 1 月期间接受了对比增强筛查性乳腺 MRI 检查。研究包括 5224 例乳腺 MRI，分为 3998 例训练集、444 例验证集和 782 例测试集。在放射学报告中，1286 例被归类为高 BPE（即明显或中度），3938 例为低 BPE（即轻度或轻微）。
磁场强度/序列：1.5T 或 3T 系统；一个预对比和三个后对比的脂肪饱和 T1 加权动态对比增强成像相位。
使用放射学报告的 BPE 标签，开发了两种深度学习模型（Slab 人工智能（AI）；最大强度投影[MIP] AI）用于 BPE 分类。使用测试集中的保留测试集，使用放射学报告的 BPE 和三位读者的平均共识作为参考标准，对模型进行了测试。
使用受试者工作特征曲线分析评估模型性能。使用 McNemar 的卡方检验（α*=0.025）评估高 BPE 与 BI-RADS 评估之间的相关性。
使用放射学报告参考标准，Slab AI 模型在整个测试集中显著优于 MIP AI 模型（曲线下面积为 0.84 与 0.79）。使用三位读者共识 BPE 标签参考标准，我们的 AI 模型显著优于放射学报告的 BPE 标签。最后，人工智能模型比放射科医生更有可能将“高 BPE”分配给可疑的乳腺 MRI，而比放射科医生更不可能将“高 BPE”分配给阴性乳腺 MRI。
乳腺 MRI 的全自动 BPE 评估可能比放射学报告的 BPE 评估更准确。
4 级 技术功效分期：3 级。
