Objective：To compare the efficacy of three conventional imaging tests，mammography（MG），ultrasonography（US），and magnetic resonance imaging（MRI），in assessing the lesion size of ductal carcinoma in situ（DCIS）or DCIS with microinvasion（DCISMI），and explore their potential value in guiding breast - conserving surgery. Methods：We retrospectively collected case files of patients with pathologically confirmed DCIS/DCIS -MI in our hospital，and compared the lesion sizes assessed by the three imaging modalities with the“gold standard”pathological size，and evaluated the accuracy and consistency of the different imaging modalities by using McNemar’s test and Bland-Altman’s method. Univariate and multivariate analyses were used to identify the clinicopathologic features that influenced the accuracy of the assessment，followed by subgroup analyses. Finally，the influencing factors leading to falsenegative -imaging results were explored and analyzed. Results：A total of 263 patients with DCIS/DCIS-MI were enrolled in this study. Regarding the measurement mean deviation，MRI mostly overestimated（+3.5 mm）lesions，while MG（-2.5 mm）and US（-1.4 mm） underestimated them. Consistency analysis suggested that MRI had the strongest correlation with pathological findings（r=0.853）and the narrowest range of 95% limits of agreement（95% LOA）（-1.73~2.44 cm），which was superior to MG（r=0.561）and US（r=0.614）. McNemar test indicated MRI’s superiority over US/MG（P ＜ 0.05），while combined US+MG achieved comparable accuracy to MRI（P= 0.921）. Logistic regression analysis showed that age ＞60 years at diagnosis（OR=0.322），tumor diameter 16-40 mm（OR=3.019），and ≥ 41 mm（OR=6.146）significantly affected the accuracy of MG assessment（all P ＜ 0.05）. Tumor diameters of 16-40 mm（OR=2.270） and ≥41 mm（OR=4.237）and ductal dilatation sign（OR=1.728）significantly increased the risk of US assessment error（all P ＜ 0.05）. Moderate -to - severe breast background parenchymal enhancement（OR=2.139）and non -mass -like foci of enhancement（OR=2.655） significantly increased the risk of MRI assessment error（all P ＜ 0.05）. Subgroup analyses suggested comparable performance for lesions ≤15 mm，US preference for 16-40 mm lesions，and MRI advantage for ≥41 mm lesions. In addition，the HER2 expression status （OR=0.100）and the Ki67 expression level（OR=0.297）were independent predictors for MG detection failure（all P ＜ 0.05）. Conclusion：Preoperative MRI is beneficial for guiding precise breast-conserving surgery in DCIS/DCIS-MI patients，particularly for lesions ≥41 mm. In clinical practice at primary hospitals，the combined use of US and MG can be promoted. For patients with low HER2 and Ki67 expression，the possibility of false-negative results should be considered during preoperative MG evaluation.