自动乳腺超声诊断系统结合计算机辅助检测乳腺恶性肿瘤敏感度的影响因素分析

目的评估影响计算机辅助检测（CAD）识别自动乳腺超声诊断系统（ABUS）乳腺恶性肿瘤敏感度的因素。 方法收集自2016年1月至2017年2月于空军军医大学西京医院行ABUS检查并经外科手术或组织学活检病理证实的乳腺恶性肿瘤患者232例,共240个恶性病灶。所有病例均经CAD软件检测,统计CAD对病灶的总敏感度,并统计分析病灶组织学类型、最大径、距乳头距离、距皮肤距离及象限等因素与CAD敏感度之间的关系。以外科手术或组织学活检病理结果为诊断"金标准",采用χ²检验分析病灶组织学类型、最大径、距乳头距离、距皮肤距离、象限、病灶边缘特征等因素与CAD敏感度的关系。 结果CAD对恶性病灶的总敏感度为85%（204/240）,对不同病理学类型的敏感度分别为：浸润性导管癌89.0%（186/209）、导管原位癌53.9%（14/29）、黏液癌75.0%（3/4）、恶性叶状肿瘤100%（1/1）,差异有统计学意义（χ²=18.836,P<0.001）。病灶最大径、距乳头距离、距皮肤距离及象限均与CAD敏感度之间比较,差异无统计学意义（P>0.05）。病灶距皮肤距离、病灶边缘特征与CAD对浸润性导管癌的敏感度之间比较,差异有统计学意义（P<0.05）。 结论CAD对恶性病灶的敏感度较高（85.0%）,尤其是对浸润性导管癌的检出（89.0%）,医师在借助CAD读图时,应注意是否有遗漏的导管原位癌、位置深或边缘模糊的浸润性导管癌。

Factors influencing sensitivity of computer-aided detection system in detection of breast cancer based on an automated breast ultrasound system

ObjectiveTo evaluate the factors affecting the sensitivity of computer-aided detection (CAD) system in detection of breast cancer based on an automated breast ultrasound system (ABUS). MethodsABUS images from 232 women with 240 histologically proven malignant lesions were collected from January 2016 to February 2017 in this retrospective study. The CAD system (QView Medical, USA) was used to evaluate ABUS images. The total sensitivity for breast cancer detection and its associations with histological type, maximum diameter, distance from the nipple, distance from the skin, and the quadrant of tumor were evaluated. ResultsThe total sensitivity of the CAD based on the ABUS for detection of breast cancer was 85.0% (204/240), and the sensitivities for detecting invasive ductal carcinoma (IDC), ductal carcinoma in situ (DCIS), mucinous carcinoma, and malignant phyllodes tumor were 89.0% (186/209), 53.9% (14/29), 75.0% (3/4), and 100% (1/1), respectively (χ²=18.836, P=0.000). The sensitivity of CAD had no significant association with the maximum diameter of lesion, the distance from the nipple, the distance from the skin, and the quadrant of tumor (P>0.05). For IDC, there was a significant association between the sensitivity of CAD and the distance from lesion to the skin and the margin of lesion (P<0.05). ConclusionsABUS-CAD has a high sensitivity (85.0%) for detecting breast cancer, especially for IDC (89.0%). When doctors interpret images with the assistance of CAD, attention should be paid to DCIS and IDC with a deep location or indistinct.