Background <p>HER2 positive directly determines whether to use targeted drugs, but the existing methods are still lack of noninvasive methods. This study aims to predict HER2-positive status in breast cancer patients by quantifying Intratumoral heterogeneity (ITH) using pre-treatment MRI and to explore the association between ITH scores and HER2-positive response.</p> Materials and methods <p>A multicenter retrospective study enrolled breast cancer patients who underwent pretreatment MRI at two hospitals, regardless of molecular subtype. ITH scores were derived from MRI radiomic features. Pearson correlation analysis was used to assess the relationship between ITH scores and HER2 status. The predictive value of ITH scores and age for HER2 positivity was also evaluated. Subgroup analyses by age and hospital were conducted to validate the association between HER2 positivity and related factors.</p> Results <p>A total of 526 patients were included, with 294 from the first hospital and 232 from the second. The AUC for predicting HER2-positive status using ITH was 0.898, with a sensitivity of 0.827 and a specificity of 0.913. Additionally, an ITH + age combined model was constructed, which achieved an AUC of 0.898, sensitivity of 0.812, and specificity of 0.913 for HER2 positivity prediction in the validation set. Subgroup analysis revealed stable predictive efficacy across hospitals and improved discriminability for patients aged over 40.</p> Conclusions <p>The ITH score is significantly associated with HER2-positive status in breast cancer and has potential as a non-invasive biomarker for predicting HER2-positive status. As an adjunct to pathological examination, it may support clinical decision-making but is not intended to replace tissue biopsy.</p>