MRI and ADC measurement techniques

All scans were performed on 1.5- or 3-T MRI scanners, using dedicated breast coils with the patients placed in prone position. All scans were performed using protocols that were standardized within each study sample following international guidelines, and included a T2-weighted sequence and native and CE T1-weighted sequences [6, 9]. All DWI scans were performed using echo planar imaging sequences and complied with the recommendations of the European Society of Breast Imaging (EUSOBI) [7]. ADC maps were calculated by each scanner’s integrated software using monoexponential fitting. Details on hardware, DWI scanning parameters, and image postprocessing used for the different patient collectives are displayed in Table ​Table22.

Hardware and sequence parameters as used for the different studies included in this retrospective analysis

Cx-Px center X-population X, SS-EPI single-shot echoplanar imaging, IR inversion recovery, SPAIR spectral adiabatic inversion recovery, MF monoexponential fit

All ADC measurements were performed using 2-dimensional regions of interest (ROIs) covering the darkest part of the lesion identified visually on the ADC map, while using the high-b-value DWI and CE T1-weighted images to avoid necrotic areas or low-signal areas caused by T2 blackout effects of fat suppression, according to recommendations of EUSOBI and a recent meta-analysis [7, 10, 11]. All measurements were performed independently by one or more radiologists blinded to histological outcome on clinical workstations. The radiologists had different levels of experience at breast MRI interpretation, ranging between 3 and 25 years (Supplemental Table 1).