Breast cancer remains the most prevalent malignancy among women worldwide, and lymph node metastasis is considered one of the earliest and most important pathways of disease dissemination. Among these, the axillary lymph node (ALN) is typically the first site of node metastasis. Studies have demonstrated a clear association between ALN metastasis and tumor T stage, with increasing T stages corresponding to high metastatic rates [1].

Although the incidence of ALN metastasis in patients with clinical T1-T2 breast cancer is relatively low, the presence and extent of nodal involvement are critical for treatment planning and prognosis [2]. The results of the Z0011 trial conducted by the American Society of Clinical Oncology [3] have shifted clinical practice significantly: patients with T1-T2 primary invasive breast cancer and only 1-2 positive sentinel lymph nodes may safely avoid axillary lymph node dissection (ALND), thereby reducing postoperative complications [4]. This evolution in surgical management has increased the demand for accurate, noninvasive preoperative assessment of ALN status.

Conventional imaging-based evaluations rely primarily on morphological characteristics of the lymph nodes, such as size, cortical thickness, margin irregularity, and the absence of fatty hilum [5]. However, these criteria are often subjective and dependent on the radiologist's level of experience, which may result in limited sensitivity, particularly in patients with minimal metastatic burden. The pooled sensitivity of ultrasound combined with needle biopsy has been reported to be only around 55.2%, underscoring the need for more reliable assessment tools [6].

Currently, ultrasound-guided lymph node biopsy or sentinel lymph node biopsy (SLNB) remains the clinical standard for determining ALN involvement. However, both are invasive procedures and may carry procedural risks. In this context, radiomics has emerged as a promising technique that enables the extraction of high-dimensional quantitative features from medical images, thereby facilitating objective tumor characterization, staging, and prognostic evaluation [[7], [8], [9]].

Ultrasound, being cost-effective, noninvasive, and widely available, is an ideal modality for radiomics analysis in routine practice. Building upon these strengths, this study investigates whether texture features derived from two-dimensional ultrasound images of primary breast tumors can predict ALN metastasis in patients with T1-T2 breast cancer. By doing so, we aim to provide a novel and practical tool to assist in early, noninvasive identification of lymph node involvement, potentially guiding treatment decisions and reducing the reliance on invasive procedures.