Zarb Francis, Mizzi Deborah, Bezzina Paul, Galea Leanne
Department of Radiography, Faculty of Health Sciences, University of Malta, MSD2080 Msida, Malta.
Diagnostics (Basel). 2025 Jun 12;15(12):1497. doi: 10.3390/diagnostics15121497.
Abbreviated breast MRI protocols have been proposed as a faster and more cost-effective alternative to standard full protocols for breast cancer detection. This study aimed to compare the diagnostic accuracy of an abbreviated protocol with that of a full protocol in identifying lobular breast carcinoma using Breast Imaging Reporting and Data System (BI-RADS) classification. The diagnostic performance was evaluated against a gold standard comprising biopsy-proven lobular carcinoma or negative follow-up imaging, using Receiver Operating Characteristic (ROC) analysis and performance metrics such as sensitivity and specificity. A retrospective analysis was conducted on 35 breast MRI examinations performed between January 2019 and December 2021. Of these, 20 cases had biopsy-confirmed lobular carcinoma, and 15 were determined to be normal based on at least 12 months of negative follow-up imaging. Two radiologists independently reviewed the images using only the abbreviated protocol, blinded to the original reports. Their findings were then compared with the initial full-protocol MRI reports. BI-RADS categories 1 and 2 were considered negative for malignancy, while BI-RADS categories 3, 4, and 5 were considered positive. The area under the ROC curve (AUC) was 1.0 for the full protocol and 0.920 and 0.922 for Radiologists A and B, respectively, using the abbreviated protocol. All malignant lesions were correctly identified by both radiologists across both protocols, resulting in a sensitivity of 100%. However, the abbreviated protocol demonstrated significantly lower specificity (73.3% for Radiologist A and 53.5% for Radiologist B) compared to 100% specificity with the full protocol ( < 0.05). Lymph node involvement was correctly identified in 6-7 of 7 cases, though Radiologist A reported four false positives. Lesion laterality and count matched histopathology in 75-90% of cancer cases depending on protocol. Lesion localization was accurate in 60-80% of cases using the abbreviated protocol, though size comparisons were limited due to the incomplete radiological documentation of dimensions. While the abbreviated MRI protocol achieved diagnostic accuracy and sensitivity comparably to the full protocol, it demonstrated reduced specificity. These findings suggest that abbreviated MRI breast protocol may be a viable screening tool, although the higher false-positive rate should be considered in clinical decision-making.
简化的乳腺MRI方案已被提议作为一种比标准全流程方案更快且更具成本效益的乳腺癌检测替代方案。本研究旨在使用乳腺影像报告和数据系统（BI-RADS）分类，比较简化方案与全流程方案在识别小叶型乳腺癌方面的诊断准确性。通过接受者操作特征（ROC）分析以及敏感性和特异性等性能指标，对照由活检证实的小叶癌或阴性随访影像组成的金标准来评估诊断性能。对2019年1月至2021年12月期间进行的35例乳腺MRI检查进行了回顾性分析。其中，20例经活检证实为小叶癌，15例根据至少12个月的阴性随访影像确定为正常。两位放射科医生仅使用简化方案独立审查图像，对原始报告不知情。然后将他们的结果与最初的全流程MRI报告进行比较。BI-RADS 1类和2类被视为恶性阴性，而BI-RADS 3类、4类和5类被视为阳性。全流程方案的ROC曲线下面积（AUC）为1.0，使用简化方案时，放射科医生A和B的AUC分别为0.920和0.922。两位放射科医生在两种方案下均正确识别了所有恶性病变，敏感性为100%。然而，与全流程方案的100%特异性相比，简化方案的特异性显著降低（放射科医生A为73.3%，放射科医生B为53.5%）（P<0.05）。7例中有6 - 7例正确识别了淋巴结受累情况，不过放射科医生A报告了4例假阳性。根据方案不同，75 - 90%的癌症病例中病变的侧别和数量与组织病理学相符。使用简化方案时，60 - 80%的病例中病变定位准确，不过由于尺寸的放射学记录不完整，大小比较受限。虽然简化的MRI方案在诊断准确性和敏感性方面与全流程方案相当，但其特异性降低。这些发现表明，简化的乳腺MRI方案可能是一种可行的筛查工具，尽管在临床决策中应考虑较高的假阳性率。
