Massachusetts General Hospital, Boston, MA, USA.
Harvard Medical School, Radiology, Boston, MA, USA.
J Natl Cancer Inst. 2022 Oct 6;114(10):1355-1363. doi: 10.1093/jnci/djac142.
Deep learning breast cancer risk models demonstrate improved accuracy compared with traditional risk models but have not been prospectively tested. We compared the accuracy of a deep learning risk score derived from the patient's prior mammogram to traditional risk scores to prospectively identify patients with cancer in a cohort due for screening.
We collected data on 119 139 bilateral screening mammograms in 57 617 consecutive patients screened at 5 facilities between September 18, 2017, and February 1, 2021. Patient demographics were retrieved from electronic medical records, cancer outcomes determined through regional tumor registry linkage, and comparisons made across risk models using Wilcoxon and Pearson χ2 2-sided tests. Deep learning, Tyrer-Cuzick, and National Cancer Institute Breast Cancer Risk Assessment Tool (NCI BCRAT) risk models were compared with respect to performance metrics and area under the receiver operating characteristic curves.
Cancers detected per thousand patients screened were higher in patients at increased risk by the deep learning model (8.6, 95% confidence interval [CI] = 7.9 to 9.4) compared with Tyrer-Cuzick (4.4, 95% CI = 3.9 to 4.9) and NCI BCRAT (3.8, 95% CI = 3.3 to 4.3) models (P < .001). Area under the receiver operating characteristic curves of the deep learning model (0.68, 95% CI = 0.66 to 0.70) was higher compared with Tyrer-Cuzick (0.57, 95% CI = 0.54 to 0.60) and NCI BCRAT (0.57, 95% CI = 0.54 to 0.60) models. Simulated screening of the top 50th percentile risk by the deep learning model captured statistically significantly more patients with cancer compared with Tyrer-Cuzick and NCI BCRAT models (P < .001).
A deep learning model to assess breast cancer risk can support feasible and effective risk-based screening and is superior to traditional models to identify patients destined to develop cancer in large screening cohorts.
与传统风险模型相比，深度学习乳腺癌风险模型的准确性有所提高，但尚未经过前瞻性测试。我们比较了从患者之前的乳房 X 光片中得出的深度学习风险评分与传统风险评分的准确性，以前瞻性地识别在筛查队列中患有癌症的患者。
我们收集了 2017 年 9 月 18 日至 2021 年 2 月 1 日期间在 5 个设施接受筛查的 57617 例连续患者的 119139 例双侧筛查乳房 X 光片的数据。患者的人口统计学数据从电子病历中检索，癌症结果通过区域肿瘤登记处链接确定，并使用 Wilcoxon 和 Pearson χ2 双尾检验比较风险模型。比较了深度学习、Tyrer-Cuzick 和美国国家癌症研究所乳腺癌风险评估工具（NCI BCRAT）风险模型在性能指标和受试者工作特征曲线下面积方面的表现。
按深度学习模型评估的高危患者每千名筛查患者的癌症检出率高于 Tyrer-Cuzick 模型（8.6，95%置信区间[CI] = 7.9 至 9.4）和 NCI BCRAT 模型（3.8，95% CI = 3.3 至 4.3）（P < 0.001）。深度学习模型的受试者工作特征曲线下面积（0.68，95% CI = 0.66 至 0.70）高于 Tyrer-Cuzick 模型（0.57，95% CI = 0.54 至 0.60）和 NCI BCRAT 模型（0.57，95% CI = 0.54 至 0.60）。模拟使用深度学习模型对前 50%风险的筛查可以比 Tyrer-Cuzick 和 NCI BCRAT 模型更准确地捕获统计学上更多的癌症患者（P < 0.001）。
评估乳腺癌风险的深度学习模型可以支持可行且有效的基于风险的筛查，并且比传统模型更能识别出在大型筛查队列中注定要患癌症的患者。
