Computer-aided detection (CAD) is applied during screening mammography for millions of women each year. Despite its popularity, several large studies have observed no benefit in breast cancer detection for practices that use CAD. This lack of benefit may be driven by how CAD information is conveyed to the radiologist. In the current study, we examined this possibility in an artificial task modeled after screening mammography. Prior work at high (50%) target prevalence suggested that CAD marks might disrupt visual attention: Targets that are missed by the CAD system are more likely to be missed by the user. However, targets are much less common in screening mammography. Moreover, the prior work on this topic has focused on simple binary CAD systems that place marks on likely locations, but some modern CAD systems employ interactive CAD (iCAD) systems that may mitigate the previously observed costs. Here, we examined the effects of target prevalence and CAD system. We found that the costs of binary CAD were exacerbated at low prevalence. Meanwhile, iCAD did not lead to a cost on unmarked targets, which suggests that this sort of CAD implementation may be superior to more traditional binary CAD implementations when targets occur infrequently. (PsycInfo Database Record (c) 2020 APA, all rights reserved).

中文翻译：

---

现实生活中更糟：对 CAD 成本低的眼动追踪检查。

计算机辅助检测 (CAD) 用于每年对数百万女性进行乳房 X 光检查。尽管它很受欢迎，但几项大型研究发现使用 CAD 的做法在乳腺癌检测中没有任何好处。这种缺乏好处可能是由 CAD 信息如何传达给放射科医师造成的。在当前的研究中，我们在筛选乳房 X 光检查后建模的人工任务中检查了这种可能性。之前在高 (50%) 目标流行率方面的工作表明 CAD 标记可能会干扰视觉注意力：CAD 系统错过的目标更有可能被用户错过。然而，目标在筛查乳房 X 光检查中并不常见。此外，关于此主题的先前工作集中在简单的二进制 CAD 系统上，这些系统在可能的位置上放置标记，但一些现代 CAD 系统采用交互式 CAD (iCAD) 系统，可以降低之前观察到的成本。在这里，我们检查了目标流行率和 CAD 系统的影响。我们发现二元 CAD 的成本在低流行率时会加剧。同时，iCAD 不会导致未标记目标的成本，这表明当目标不经常出现时，这种 CAD 实施可能优于更传统的二进制 CAD 实施。（PsycInfo 数据库记录 (c) 2020 APA，保留所有权利）。这表明，当目标不经常出现时，这种 CAD 实现可能优于更传统的二进制 CAD 实现。（PsycInfo 数据库记录 (c) 2020 APA，保留所有权利）。这表明，当目标不经常出现时，这种 CAD 实现可能优于更传统的二进制 CAD 实现。（PsycInfo 数据库记录 (c) 2020 APA，保留所有权利）。