Breast cancer (BC) remains a global health challenge, demanding innovations in imaging to enhance early detection, precise characterization, and tailored treatment strategies. This aligns with the paradigm of personalized medicine (PM), delivering the right treatment to the right patient at the right time [1–4].

Contrast-enhanced mammography (CEM) has emerged as a promising second-level imaging technique, offering a cost-effective alternative to contrast-enhanced breast Magnetic Resonance (MR) [5–7]. By leveraging tumor neo-angiogenesis through intravenous iodinated contrast, CEM highlights hypervascularized regions, including neoplastic lesions [8,9]. Its main indications include preoperative staging, resolving screening concerns, and assessing response to neoadjuvant chemotherapy. CEM is also preferred for dense breasts, MR-indicated pathways, and patients with MR contraindications.

Radiomics, a rapidly evolving field, extracts quantitative imaging data (e.g., intensity, shape, texture) to improve cancer detection and characterization [9–12]. Combined with machine learning (ML) and deep learning (DL), these data can differentiate malignancies, assess genetics, predict treatment responses, and build integrated diagnostic models [10,12–17]. The convergence of CEM and radiomics may revolutionize breast cancer diagnostics and prognosis by offering cost-effective, high-sensitivity imaging with quantitative insights.

This study aims to evaluate radiomic features from tumor lesions and peritumoral backgrounds to identify correlations with prognostic factors. Specifically, it investigates whether peritumoral parenchyma exhibits features undetectable by the human eye that influence prognosis.