保乳术后乳腺X线监测:数字乳腺断层合成与人工智能计算机辅助检测的影响
该回顾性研究纳入314例保乳术后接受数字乳腺断层合成监测的女性,比较数字乳腺X线摄影(DM)、DM联合DBT、DM联合AI-CAD三种判读模式的召回率与诊断效能。结果显示,联合AI-CAD或DBT均可降低同侧和对侧乳腺召回率并提高准确率;在同侧乳腺,AI-CAD较DBT召回率更低、准确率更高,但敏感度显著低于DM。
要点
- 保乳术后监测中,DM联合AI-CAD可显著降低同侧和对侧乳腺召回率并提高特异度。
- 同侧乳腺AI-CAD的敏感度(22.2%)显著低于DM(66.7%),阅片时需警惕漏诊。
- AI-CAD在同侧乳腺的召回率和准确率优于附加DBT,但敏感度与DBT无差异。
核心结论
- 保乳术后,附加DBT或AI-CAD较DM均降低同侧和对侧乳腺召回率并提高准确率。
- 同侧乳腺中,附加AI-CAD较附加DBT召回率更低、准确率更高,但敏感度显著低于DM。
研究设计与样本
- 回顾性研究,纳入314例保乳术后接受DBT监测的女性(平均年龄53.3±10.6岁,4例双侧乳腺癌)。
- 手术至DBT平均间隔15.2±15.4个月;3名乳腺放射科医生独立判读DM、DM+DBT、DM+AI-CAD三种模式。
诊断标准与关键数值
- 同侧乳腺召回率:DM+AI-CAD 1.9%,DM+DBT 4.1%,DM 11.2%(p<.001)。
- 对侧乳腺召回率:DM+AI-CAD 1.5%,DM+DBT 2.7%,DM 6.6%(AI-CAD vs DM p<.001;AI-CAD vs DBT p=.08)。
- 同侧乳腺准确率:DM+AI-CAD 97.0%,DM+DBT 94.8%,DM 88.5%(p<.05);特异度:98.3% vs 96.1% vs 89.3%(p<.05);敏感度:22.2% vs 22.2% vs 66.7%(AI-CAD vs DM p=.03)。
- 对侧乳腺准确率:DM+AI-CAD 97.1%,DM+DBT 96.1%,DM 92.5%(AI-CAD vs DM p<.001);特异度:98.6% vs 97.5% vs 93.7%(AI-CAD vs DM p<.001);敏感度无差异(33.3% vs 22.2% vs 11.1%,p>.05)。
临床应用与阅片要点
- 保乳术后监测可考虑附加AI-CAD以降低召回率、提高特异度,但需注意其同侧乳腺敏感度较低,避免漏诊。
- 同侧乳腺附加AI-CAD的召回率和准确率优于附加DBT,但敏感度与DBT相当,判读时应结合临床。
局限性
- 回顾性单中心设计,样本量有限(314例),复发事件少(6例),敏感度估计可能不稳定。
- AI-CAD与DBT的判读模式为独立 sessions,未评估联合使用效果。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
英文原文(摘要)· 点击展开
BACKGROUND. Postoperative mammograms present interpretive challenges due to postoperative distortion and hematomas. The application of digital breast tomosyn-thesis (DBT) and artificial intelligence-based computer-aided detection (AI-CAD) after breast-conserving therapy (BCT) has not been widely investigated. OBJECTIVE. The purpose of our study was to assess the impact of additional DBT or AI-CAD on recall rate and diagnostic performance in women undergoing mammographic surveillance after BCT. METHODS. This retrospective study included 314 women (mean age, 53.3 ± 10.6 [SD] years; four with bilateral breast cancer) who underwent BCT followed by DBT (mean interval from surgery to DBT, 15.2 ± 15.4 months). Three breast radiologists independently reviewed images in three sessions: digital mammography (DM), DM with DBT (DM plus DBT), and DM with AI-CAD (DM plus AI-CAD). Recall rates and diagnostic performance were compared between DM, DM plus DBT, and DM plus AI-CAD using the readers' mean results. RESULTS. Of the 314 women, six breast recurrences (three ipsilateral and three contralateral) had developed at the time of surveillance mammography. The ipsilateral breast recall rate was lower for DM plus AI-CAD (1.9%) than for DM (11.2%) or DM plus DBT (4.1%) (p < .001). The contralateral breast recall rate was significantly lower for DM plus AI-CAD (1.5%, p < .001) than for DM (6.6%) but for not DM plus DBT (2.7%, p = .08). In the ipsilateral breast, accuracy was higher for DM plus AI-CAD (97.0%) than for DM (88.5%) or DM plus DBT (94.8%) (p < .05); specificity was higher for DM plus AI-CAD (98.3%) than for DM (89.3%) or DM plus DBT (96.1%) (p < .05); sensitivity was significantly lower for DM plus AI-CAD (22.2%) than for DM (66.7%, p = .03) but not DM plus DBT (22.2%, p > .99). In the contralateral breast, accuracy was significantly higher for DM plus AI-CAD (97.1%) than for DM (92.5%, p < .001) but not DM plus DBT (96.1%, p = .25); specificity was significantly higher for DM plus AI-CAD (98.6%) than for DM (93.7%, p < .001) but not DM plus DBT (97.5%) (p = .09); sensitivity was not different between DM (33.3%), DM plus DBT (22.2%), and DM plus AI-CAD (11.1%) (p > .05). CONCLUSION. After BCT, adjunct DBT or AI-CAD reduced recall rates and improved accuracy in the ipsilateral and contralateral breasts compared with DM. In the ipsilateral breast, the addition of AI-CAD resulted in a lower recall rate and higher accuracy than the addition of DBT. CLINICAL IMPACT. AI-CAD may help address the challenges of interpreting post-BCT surveillance mammograms.
引用
- Yoon JH, Kim EK, Kim GR, et al. Mammographic Surveillance After Breast-Conserving Therapy: Impact of Digital Breast Tomosynthesis and Artificial Intelligence-Based Computer-Aided Detection. AJR Am J Roentgenol. 2021. PMID: 34378399. doi:10.2214/AJR.21.26506
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。