Department of Medical Imaging, Princess Margaret Hospital, University of Health Network and Mount Sinai Hospital, University of Toronto, Toronto, Ontario, Canada.
Can Assoc Radiol J. 2010 Jun;61(3):162-9. doi: 10.1016/j.carj.2009.11.010. Epub 2010 Feb 4.
The purpose of this study was to evaluate the sensitivity of a direct computer-aided detection (CAD) system (d-CAD) in full-field digital mammography (FFDM) for the detection of microcalcifications not associated with mass or architectural distortion.
A database search of 1063 consecutive stereotactic core biopsies performed between 2002 and 2005 identified 196 patients with Breast Imaging-Reporting and Data System (BI-RADS) 4 and 5 microcalcifications not associated with mass or distortion detected exclusively by bilateral FFDM. A commercially available CAD system (Second Look, version 7.2) was retrospectively applied to the craniocaudal and mediolateral oblique views in these patients (mean age, 59 years; range, 35-84 years). Breast density, location and mammographic size of the lesion, distribution, and tumour histology were recorded and analysed by using chi(2), Fisher exact, or McNemar tests, when applicable.
When using d-CAD, 71 of 74 malignant microcalcification cases (96%) and 101 of 122 benign microcalcifications (83%) were identified. There was a significant difference (P < .05) between CAD sensitivity on the craniocaudal view, 91% (68 of 75), vs CAD sensitivity on the mediolateral oblique view, 80% (60 of 75). The d-CAD sensitivity for dense breast tissue (American College of Radiology [ACR] density 3 and 4) was higher (97%) than d-CAD sensitivity (95%) for nondense tissue (ACR density 1 and 2), but the difference was not statically significant. All 28 malignant calcifications larger than 10 mm were detected by CAD, whereas the sensitivity for lesions small than or equal to 10 mm was 94%.
D-CAD had a high sensitivity in the depiction of asymptomatic breast cancers, which were seen as microcalcifications on FFDM screening, with a sensitivity of d-CAD on the craniocaudal view being significantly better. All malignant microcalcifications larger than 10 mm were detected by d-CAD.
本研究旨在评估直接计算机辅助检测（CAD）系统（d-CAD）在全数字化乳腺摄影（FFDM）中检测不伴肿块或结构扭曲的微钙化的敏感性。
通过对 2002 年至 2005 年间进行的 1063 例立体定向核心活检的数据库搜索，确定了 196 例乳腺影像报告和数据系统（BI-RADS）4 级和 5 级微钙化患者，这些患者的微钙化仅通过双侧 FFDM 检测到，且不伴肿块或结构扭曲。回顾性地将一种商业上可用的 CAD 系统（Second Look，版本 7.2）应用于这些患者的头尾位和内外斜位（平均年龄 59 岁；范围 35-84 岁）。记录并分析了乳房密度、病变位置和乳腺大小、分布以及肿瘤组织学，应用卡方检验、Fisher 确切检验或 McNemar 检验，当适用时。
当使用 d-CAD 时，74 例恶性微钙化病例中的 71 例（96%）和 122 例良性微钙化病例中的 101 例（83%）被识别。CAD 在头尾位上的敏感性（91%，68 例/75 例）与 CAD 在内外斜位上的敏感性（80%，60 例/75 例）之间存在显著差异（P <.05）。致密乳腺组织（美国放射学院[ACR]密度 3 和 4）的 d-CAD 敏感性（97%）高于非致密乳腺组织（ACR 密度 1 和 2）的 d-CAD 敏感性（95%），但差异无统计学意义。所有 28 例大于 10mm 的恶性钙化均被 CAD 检测到，而对于小于或等于 10mm 的病变，其敏感性为 94%。
d-CAD 在检测无症状乳腺癌方面具有很高的敏感性，这些乳腺癌在 FFDM 筛查中表现为微钙化，其在头尾位上的敏感性明显更好。所有大于 10mm 的恶性微钙化均被 d-CAD 检测到。
