Performance Comparison of Conventional Imaging Methods for Adjunctive Diagnosis of Breast Cancer
Feng ZHANG,Li-Min LUO,Xu-Dong BAO
DOI: https://doi.org/10.3969/j.issn.0258-8021.2012.02.017
2012-01-01
Abstract:To avoid presences of benign biopsies, there is a urgent need to introduce adjunctive diagnostic methods for suspicious lesions classified as BI-RADS 3-5 detected in mammography. However, diagnostic performances of different conventional imaging modalities, such as ultrasound (US), magnetic resonance imaging (MRI), scintimammography (SMM) and electrical impedance scanning (EIS), for this specific clinical application are less known. In this paper, related articles published between 2000 and 2011 were searched, and the diagnostic performances of each imaging modality based on a relative large size of breast lesions were demonstrated by meta-analysis. The sensitivity with 95% confidence interval of US, MRI, SMM and EIS are 81.3% (0.794, 0.831), 90.0% (0.883, 0.916), 84.4% (0.821, 0.866) and 81.0% (0.778, 0.839); the specificity are 93.0% (0.921, 0.938), 82.9% (0.811, 0.846), 83.6% (0.816, 0.855) and 67.9% (0.651, 0.707); the accuracy, as the area under summary receive operation curve, are 94.7% (0.924, 0.969), 93.4% (0.894, 0.973), 92.8% (0.900, 0.955) and 76.9% (0.691, 0.846). Statistic analysis indicated that MRI was the most sensitive one and US was of the most specificity of four conventional imaging modalities. There were no significant differences among US, MRI and SMM in accuracy. EIS had a relative high sensitivity, but its specificity was lower. Adjunctive diagnosis according the characteristic of each imaging modality, adopting single (US/MRI/SMM/EIS) or combing imaging modalities (MRI followed by US), could reduce benign biopsies rate from 80% to 25.7%/43.0%/43.8%/61.2% and even 6.4%, with a rate of missed diagnosis of 10.0%-27.0%. In application, short-interval follow-up should be taken for benign patients diagnosed by the adjunctive imaging method to decrease the blight of possible missed diagnosis.