Sixty-two patients were included, 45 treated with primary surgery and 17 treated with neoadjuvantsystemic therapy. When categorizing patients based on imaging findings, 17 (27.4%) had only suspiciouscalcifications, 10 (16.1%) had only NME, and 17 (27.4%) showed both findings (Figure 2). In 18 of the 62patients (29.0%), neither calcifications nor NME were detected.When comparing the imaging and histopathological characteristics of patients with detected DCIS (n=44)to undetected DCIS (n=18), patients with presence of comedonecrosis and calcifications inhistopathology were more often detected, in 80.9% and 85.0% of patients respectively, compared topatients without comedonecrosis and calcifications in histopathology (40% and 45.5% detectedrespectively). An example of imaging and histopathology of a detected and undetected DCIS componentare shown in Figure 2 and 3, respectively.Of the 45 patients treated with primary surgery, 35 had a DCIS component present outside of theinvasive tumor. The detection rate of this DCIS component, based on suspicious calcifications and/orNME, was 77.1% (27/35). The DCIS component outside of the invasive tumor that was not detected(n=8) was significantly smaller compared to the detected DCIS (median size 12mm (IQR 9.25-23.75) vs40mm (IQR 20-70), p < 0.001). The DCIS/IBC size ratio was also significantly lower compared to the DCIScomponent that was detected (median 1.06 vs 3.13, p = 0.001).In the 45 patients that underwent primary surgery, the size of the different imaging findings wascompared to the size of the DCIS component in histopathology (Table 1). For the 22 patients withsuspicious calcifications, the percentage size concordance (error within margin of 10mm) was 59.1%(13/22). The intraclass correlation coefficient for suspicious calcifications was 0.43. For the 21 patientswith non-mass enhancement, 47.6% (10/21) of patients had a size difference of ≤10 mm. The intraclasscorrelation coefficient for NME was 0.65. Suspicious calcifications more often underestimated the DCIScomponent size in pathology, while NME both over- and underestimated the DCIS size.

The inter-observer reliability was investigated in all patients, on both detection and measurement ofimaging findings. In 90.3% (56/62) of cases, the independent radiologists agreed on the presence orabsence of suspicious calcifications, with a kappa of 0.81. The ICC for measured diameter of suspiciouscalcifications was 0.89, in the 30 patients in which both measurements were performed. In 74.2%(46/62) of cases, the independent radiologists agreed on the presence or absence of NME, with a kappaof 0.47. The ICC for measured diameter of NME was 0.80, in the 17 patients in which bothmeasurements were performed.