Interobserver reliability of automated breast volume scanner (ABVS) interpretation and agreement of ABVS findings with hand held breast ultrasound (HHUS), mammography and pathology results

Abstract

Objectives: Handheld breast ultrasound (HHUS) lacks standardization and reproducibility. The automated breast volume scanner (ABVS) could overcome this limitation. To analyze the interobserver reliability of ABVS and the agreement with HHUS, mammography and pathology is the aim of this study.

Methods: All 42 study participants (=84 breasts) received an ABVS examination in addition to the conventional breast diagnostic work-up. 25 breasts (30%) showed at least one lesion. The scans were interpreted by six breast diagnostic specialists blinded to results of breast imaging and medical history. 32 lesions received histological work-up: 20 cancers were detected. We used kappa statistics to interpret agreement between examiners and diagnostic instruments.

Results: On the basis of the Breast Imaging Reporting and Data System (BI-RADS) classification of the 84 breasts an agreement (defined as ≥4 of 6 examiners) was achieved in 63 cases (75%) (mk=0.35) and even improved when dichotomizing the interpretation in benign (BI-RADS 1, 2) and suspicious (BI-RADS 4, 5) to 98% (mk=0.52). Agreement of ABVS examination to HHUS, mammography and pathology was fair to substantial depending on the specific analysis.

Conclusions: The development of an ABVS seems to be a promising diagnostic method with a good interobserver reliability, as well as a comparable good test criteria as HHUS.