Abstract
Purpose: To determine the diagnostic accuracy of targeted electrical impedance imaging in characterizing breast lesions, and to evaluate whether lesion size, depth and histopathology affect the diagnosis.
Material and Methods: A total of 137 women with 145 lesions (79 malignant and 66 benign) found by palpation or mammography were prospectively enrolled in this study. The patients were examined by means of clinical breast examination, mammography, ultrasonography, and electrical impedance imaging with TransScan TS2000. A level of suspicion (LOS) post-processing algorithm (v2.67) was used for TS2000 lesion assessment. Imaging findings were correlated with cytologic (n = 54) and histologic diagnoses (n = 91). Patients with benign lesions were followed up for a mean of 36 months.
Results: TS2000 showed a high sensitivity (86%) which did not differ significantly from that of mammography (87%) and ultrasonography (US) (75%). The specificity of TS2000 (49%) was significantly lower compared to mammography (97%, P<0.0001) and US (100%, P<0.0001). The additive use of TS2000 to mammography and US yielded no significant increase in sensitivity (97%), but the decrease in specificity was significant (46%, P<0.0001). Diagnostic effectiveness of TS2000 (Az = 0.68), as measured by the area under the ROC curve, was significantly lower than for mammography (Az = 0.93, P<0.0001) and for US (Az = 0.91, P<0.0001). When using TS2000 in addition to mammography and US (Az = 0.86), a significant impairment was found (P = 0.0003).
Conclusion: The role of targeted electrical impedance imaging as an adjunct to mammography and ultrasonography in the diagnosis of breast lesions is not justified by the result of this study.