Wang Yao-Kuan, Klanecek Zan, Wagner Tobias, Cockmartin Lesley, Marshall Nicholas, Studen Andrej, Jeraj Robert, Bosmans Hilde
Department of Imaging and Pathology, University Hospital Leuven, Herestraat 49, Box 7003, 3000 Leuven, Belgium.
Faculty of Mathematics and Physics, University of Ljubljana, Ljubljana, Slovenia.
Radiol Artif Intell. 2025 Sep 3:e240417. doi: 10.1148/ryai.240417.
Purpose To evaluate whether features extracted by Mirai can be aligned with mammographic observations, and contribute meaningfully to the prediction. Materials and Methods This retrospective study examined the correlation of 512 Mirai features with mammographic observations in terms of receptive field and anatomic location. A total of 29,374 screening examinations with mammograms (10,415 women, mean age at examination 60 [SD: 11] years) from the EMBED Dataset (2013-2020) were used to evaluate feature importance using a feature-centric explainable AI pipeline. Risk prediction was evaluated using only calcification features (CalcMirai) or mass features (MassMirai) against Mirai. Performance was assessed in screening and screen-negative (time-to-cancer > 6 months) populations using the area under the receiver operating characteristic curve (AUC). Results Eighteen calcification features and 18 mass features were selected for CalcMirai and MassMirai, respectively. Both CalcMirai and MassMirai had lower performance than Mirai in lesion detection (screening population, 1-year AUC: Mirai, 0.81 [95% CI: 0.78, 0.84], CalcMirai, 0.76 [95% CI: 0.73, 0.80]; MassMirai, 0.74 [95% CI: 0.71, 0.78]; values < 0.001). In risk prediction, there was no evidence of a difference in performance between CalcMirai and Mirai (screen-negative population, 5-year AUC: Mirai, 0.66 [95% CI: 0.63, 0.69], CalcMirai, 0.66 [95% CI: 0.64, 0.69]; value: 0.71); however, MassMirai achieved lower performance than Mirai (AUC, 0.57 [95% CI: 0.54, 0.60]; value < .001). Radiologist review of calcification features confirmed Mirai's use of benign calcification in risk prediction. Conclusion The explainable AI pipeline demonstrated that Mirai implicitly learned to identify mammographic lesion features, particularly calcifications, for lesion detection and risk prediction. ©RSNA, 2025.
目的 评估由Mirai提取的特征是否能与乳腺钼靶观察结果相匹配，并对预测有显著贡献。材料与方法 这项回顾性研究从感受野和解剖位置方面，考察了512个Mirai特征与乳腺钼靶观察结果的相关性。利用来自EMBED数据集（2013 - 2020年）的29374例有乳腺钼靶检查的筛查病例（10415名女性，检查时平均年龄60岁[标准差：11岁]），通过以特征为中心的可解释人工智能流程评估特征重要性。仅使用钙化特征（CalcMirai）或肿块特征（MassMirai）与Mirai进行风险预测比较。使用受试者操作特征曲线下面积（AUC）在筛查人群和筛查阴性（癌症发生时间>6个月）人群中评估性能。结果 分别为CalcMirai和MassMirai选择了18个钙化特征和18个肿块特征。在病变检测方面，CalcMirai和MassMirai的性能均低于Mirai（筛查人群，1年AUC：Mirai为0.81[95%可信区间：0.78，0.84]，CalcMirai为0.76[95%可信区间：0.73，0.80]；MassMirai为0.74[95%可信区间：0.71，0.78]；P值<0.001）。在风险预测中，没有证据表明CalcMirai和Mirai在性能上存在差异（筛查阴性人群，5年AUC：Mirai为0.66[95%可信区间：0.63，0.69]，CalcMirai为0.66[95%可信区间：0.64，0.69]；P值：0.71）；然而，MassMirai的性能低于Mirai（AUC为0.57[95%可信区间：0.54，0.60]；P值<0.001）。放射科医生对钙化特征的审查证实了Mirai在风险预测中对良性钙化的应用。结论 可解释人工智能流程表明，Mirai隐含地学会了识别乳腺钼靶病变特征，尤其是钙化特征，用于病变检测和风险预测。©RSNA，2025年
