Cardoso Maria João, Magalhães André, Almeida Teresa, Costa Susy, Vrieling Conny, Christie David, Johansen Jørgen, Cardoso Jaime S
Department of Surgery, Faculdade de Medicina do Porto, Porto, Portugal.
Breast Cancer Res Treat. 2008 Dec;112(3):565-8. doi: 10.1007/s10549-008-9896-5. Epub 2008 Jan 18.
The breast cancer conservative treatment. cosmetic results (BCCT.core) is a new software tool created for the automatic and objective evaluation of the aesthetic result of BCCT. It makes use of a face-only photographic view of each patient and might thus have been considered insufficient for an accurate evaluation, as others have used multiple views of each patient. The purpose of this work is to compare the performance of the BCCT.core (using face-only views) with a subjective expert analysis using both the face-only and four-view assessment. Photographs in four-views of 150 patients, were evaluated by a panel of experts and a consensus classification was obtained. The agreement between the consensus and the BCCT.core (face-only view) was calculated using the kappa (k) and weighted kappa (wk) statistics. Face-only views, of the same 150 patients, were subsequently sorted out in a different order and sent for individual evaluation by three specialists from the previous panel of experts. The individual agreement between the face-only view and the four-view evaluation by each of the three experts and the consensus was calculated using the same methods. Obtained results were compared to the BCCT.core performance. The software obtained a moderate agreement with the consensus (k = 0.57; wk = 0.68). The highest value of agreement, from the three experts, between the four-view evaluation and the consensus was identical to the software agreement (k = 0.55; wk = 0.67). In the face-only view experiment, the highest value of agreement between the experts and the consensus was only fair (k = 0.37; wk = 0.54). Performance of the software was thus considered equal to that obtained by experts using a four-view evaluation.
乳腺癌保守治疗的美容效果（BCCT.core）是一款新的软件工具，用于对乳腺癌保守治疗的美学效果进行自动、客观评估。它仅使用每位患者的面部照片视图，因此可能被认为不足以进行准确评估，因为其他方法使用了每位患者的多个视图。这项工作的目的是将BCCT.core（仅使用面部视图）的性能与使用面部视图和四视图评估的主观专家分析进行比较。150名患者的四视图照片由一组专家进行评估，并获得了共识分类。使用kappa（k）和加权kappa（wk）统计量计算共识与BCCT.core（仅面部视图）之间的一致性。随后，对同一150名患者的仅面部视图以不同顺序进行整理，并由前一组专家中的三位专家进行单独评估。使用相同方法计算三位专家各自的仅面部视图与四视图评估之间以及与共识之间的个体一致性。将获得的结果与BCCT.core的性能进行比较。该软件与共识达成了中等一致性（k = 0.57；wk = 0.68）。三位专家中，四视图评估与共识之间的最高一致性值与软件的一致性相同（k = 0.55；wk = 0.67）。在仅面部视图实验中，专家与共识之间的最高一致性值仅为一般（k = 0.37；wk = 0.54）。因此，该软件的性能被认为与专家使用四视图评估获得的性能相当。
