Maier Anna, Heil Joerg, Lauer Anna, Harcos Aba, Schaefgen Benedikt, von Au Alexandra, Spratte Julia, Riedel Fabian, Rauch Geraldine, Hennigs André, Domschke Christoph, Schott Sarah, Rom Joachim, Schuetz Florian, Sohn Christof, Golatta Michael
University Breast Unit, Department of Gynecology and Obstetrics, University of Heidelberg, Im Neuenheimer Feld 440, 69120, Heidelberg, Germany.
Institute of Medical Biometry and Informatics, University of Heidelberg, Heidelberg, Germany.
Arch Gynecol Obstet. 2017 Sep;296(3):571-582. doi: 10.1007/s00404-017-4473-y. Epub 2017 Jul 26.
Breast ultrasound could be a valuable tool complementary to mammography in breast cancer screening. Automated 3D breast ultrasound (ABUS) addresses challenges of hand-held ultrasound and could allow double reading analysis of ultrasound images. This trial assesses the inter-rater reliability and double reading analysis of an ABUS system.
To assess the reproducibility and diagnostic validity of the ABUS system, SomoV™, a blinded double reading analysis, was performed in 1019 patients (2038 breasts) by two examiners (examiner A/B) and compared to single reading results, as well as to the reference standard regarding its diagnostic validity. Cohen's kappa coefficients were calculated to measure the inter-rater reliability and agreement of the different diagnostic modalities. Patient comfort and time consumption for image acquisition and reading were analyzed descriptively as secondary objectives.
Analysis of inter-rater reliability yielded agreement in 81.6% (κ = 0.37; p < 0.0001) showing fair agreement. Single reading analysis of SomoV™ exams (examiner A/examiner B) compared to reference standard showed good specificity (examiner A: 88.3%/examiner B: 84.5%), fair inter-rater agreement (examiner A: κ = 0.31/examiner B: κ = 0.31), and adequate sensitivity (examiner A: 53.1%/examiner B: 64.2%). Double reading analysis yielded good sensitivity and specificity (73.7 and 77.7%). Mammography (n = 1911) alone detected 160 of 176 carcinomas (sensitivity 90.1%). Adding SomoV™ to mammography would have detected 12 additional carcinomas, resulting in a higher sensitivity of 97.7%.
SomoV™ is a promising technique with good sensitivity, high patient comfort, and fair inter-examiner reliability. It allows double reading analysis that, in combination with mammography, could increase detection rates in breast cancer screening.
乳腺超声可能是乳腺癌筛查中与乳腺X线摄影互补的一项有价值的工具。自动三维乳腺超声（ABUS）解决了手持超声的难题，并可实现超声图像的双人阅片分析。本试验评估了一种ABUS系统的阅片者间可靠性及双人阅片分析。
为评估ABUS系统SomoV™的可重复性及诊断有效性，由两名检查者（检查者A/B）对1019例患者（2038侧乳房）进行了盲法双人阅片分析，并将结果与单人阅片结果以及关于其诊断有效性的参考标准进行比较。计算Cohen's kappa系数以衡量阅片者间的可靠性及不同诊断方式的一致性。作为次要目标，对患者的舒适度以及图像采集和阅片的时间消耗进行了描述性分析。
阅片者间可靠性分析显示一致性为81.6%（κ = 0.37；p < 0.0001），表明一致性尚可。SomoV™检查的单人阅片分析（检查者A/检查者B）与参考标准相比显示出良好的特异性（检查者A：88.3%/检查者B：84.5%）、尚可的阅片者间一致性（检查者A：κ = 0.31/检查者B：κ = 0.31）以及足够的敏感性（检查者A：53.1%/检查者B：64.2%）。双人阅片分析得出了良好的敏感性和特异性（分别为73.7%和77.7%）。单独的乳腺X线摄影（n = 1911）检测出了176例癌症中的160例（敏感性90.1%）。将SomoV™添加到乳腺X线摄影中可额外检测出12例癌症，从而使敏感性提高到97.7%。
SomoV™是一项很有前景的技术，具有良好的敏感性、较高的患者舒适度和尚可的检查者间可靠性。它允许进行双人阅片分析，与乳腺X线摄影相结合可提高乳腺癌筛查的检出率。
