Axillary lymph node metastasis (ALNM) is one of the most important prognostic factors determining survival in breast cancer. 1 The status of the axilla determines the need for axillary lymph node dissection (ALND), axillary radiotherapy, neoadjuvant or adjuvant chemotherapy. 2 Accurate determination of the axillary status before treatment is critical in determining individualized treatment options. 3 Age, tumor size, tumor quadrant, multifocality, histological grade, pathological type, receptor status, molecular subtype are associated with ALNM. 456789 Radiomic analysis aims to contribute to diagnosis, treatment and follow-up processes by extracting specific quantitative information about diseases that the human eye cannot perceive from medical images. With radiomic analysis, the aim is to maximize the information obtained from images by obtaining quantitative data about signal intensities and spatial distribution of inter-pixel relationships. 10 Recently, radiomics has attracted considerable attention in the medical field, especially in oncology.

Successful results were also obtained in the diagnosis, treatment and classification of breast cancer. 11 Also, promising results were acquired in the differentiation of malignant and benign breast masses, in the estimation of the grade and receptor status and subtypes of malignant tumors using radiomic features extracted from magnetic resonance imaging (MRI). This method is also promising in the prediction of neoadjuvant chemotherapy response in breast cancer. In studies performed with MRI, it was found that radiomics successfully helped even in the prediction of recurrence of breast cancer. 1213141516 Radiomics features of axillary lymph nodes from T2W MR images were not successful in predicting ALNM. 17 However, ALNM could be predicted with radiomic features obtained from T2-weighted (T2W), diffusion-weighted (DW) and T1+C images of the primary tumor. 18192021222324 Our aim in this study was to investigate the performance of radiomic features obtained from short tau inversion recovery (STIR) sequences of primary tumor in predicting axillary metastasis.

Patients diagnosed with invasive breast cancer by core biopsy between August 2017 and February 2023 were evaluated retrospectively. Patients who underwent sentinel lymph node biopsy (SLNB) or ALND and patients who underwent MRI before treatment were included in the study. Patients who received neoadjuvant chemotherapy, or had unknown pathology, another malignancy, a recurrent disease, or artifacts in MRI images were excluded from the study. Manual segmentation was performed independently by two radiologists with 8 and 10 years of experience in breast imaging, blinded to the axillary condition of the patients. Segmentation was performed from all axial STIR sequences with tumor. A total of 851 texture features, including first-order, second-order and wavelet-based features were extracted with Slicer-Radiomics (PyRadiomics v3.0.1) (Figure 1). One month later, 30 randomly selected patients were independently segmented by the same two radiologists and radiomic features were extracted again. Thus, interobserver agreement was evaluated. Machine Learning Analysis Python 3.11 (Jupyter Notebook, Pycaret Library) was used for data processing and machine learning analysis. The synthetic minority oversampling technique (SMOTE) was used to avoid imbalanced data sets. Data normalization was performed before model develepment.

The data sets were randomly divided into training and independent testing sets. We used 10-fold cross validation of the trained models to avoid data overfitting.

Overall, 15 machine learning (ML) algorithms were used. The area under the curve (AUC), accuracy, recall, precision and F1 scores were compared with the performances of the ML algorithms. The best model for accuracy and AUC was selected and evaluated on the test set. AUC, accuracy, recall, precision, and F1 scores were given by the confusion matrix. The best model was tuned and finalized. Altogether, 667 features with ICC values above 0.7 were evaluated with ML. Wavelet filtered texture features, maximum 3D diameter, skewness kurtosis and maximum signal features showed a high correlation with ALNM. The features selected by ML algorithms and their importance are presented in Figure 2. The accuracy and AUC of ML algorithms on the training set were in the range of 57%-86% and 0.50-0.95, respectively (Table 1). Among the ML algorithms, the best model was CatBoost classifier (AUC:0.95, accuracy: 84%). The ROC curve showing the success of the CatBoost classifier in predicting ALNM is presented in Figure 3.

Recently, less invasive treatment approaches for the axilla have been accepted. SLNB is preferred instead of ALND in early stage breast cancer. Determining the condition of the axilla in the preoperative period is important in the development of surgical plans. For individualized and minimally invasive treatment options, it is important to determine the condition of the axilla before treatment. 25 Among the imaging methods, ultrasound (US) is the primary tool in the evaluation of the axilla.

However, approximately 15-20% of patients with negative US findings have metastases in the SLNB. In mammography (MG), 50% of level 1 axillary lymph nodes can be visualized and levels 2 and 3 cannot be evaluated. 26 It has been reported that MRI has similar sensitivity to US, but less specificity in detecting nodal metastases. 27 In previous studies, ALNM could be predicted by the radiomics features of the primary tumor in MRI. Yu et al., on the other hand, successfully predicted axillary lymph node status on T1+C, T2W, DW MRI with a multiomic signature including radiomics features and clinicopathologic features obtained from lymph nodes and primary tumor. 19 Qiu et al. were successful in predicting ALNM using clinicopathological features, morphological features of lymph nodes from MR images, and radiomics features of the primary mass. Radiomic features obtained from DWI, T2W and T1+C images of the primary mass were successful in predicting ALNM with an AUC value of 0.806. 20 Wang et al. were able to predict axillary metastases with an AUC value of over 0.80 with the radiomics and deep learning features obtained from dynamic contrast-enhanced (DCE) MRI. 21 Chen et al. were able to predict ALNM with the nomogram created by the radiomics and clinicopathological features of the primary tumor obtained from DW and DCE MRI (AUC value in the training set and test set, respectively 0.80, 0.71). 22 Using DCE MRI, Liu and colleagues were able to predict ALNM by tumoral and peritumoral radiomics signature (AUC values in the training and test set were 0.872, 0.863, respectively). 23 Cui et al. predicted axillary lymph node status with radiomic features obtained from second post-contrast images on MRI with an AUC of 0.86 and an accuracy of 89%. 24 However, none of these studies presented the performance of T2W images alone in predicting ALNM.

Dong and colleagues were successful in predicting sentinel lymph node metastasis with radiomic features obtained from fat-suppressed T2W and DW MRI images. With T2W alone, AUC values were 0.847 in the training set and 0.770 in the validation set. With DW MRI, they obtained AUC values of 0.847 in the training set and 0.787 in the validation set. When they combined the features obtained from T2W and DW images, the AUC values were 0.863 in the training set and 0.805 in the validation set. 18 In our study, we predicted ALNM with 0.92 AUC and 84% accuracy in the training and validation set with the radiomic features obtained from STIR sequences by modeling using machine learning analysis. Radiomics features derived from T2W sequences detected ALNM with an AUC of up to 0.85. When DWI and DCE were combined with T2W, the AUC value increased to 0.86. 18 In our study, only STIR arrays reached 0.92 AUC. Since STIR sequences are used in routine breast MRI in some centers, they may contribute to the prediction of ALNM. To our knowledge, STIR radiomics have not been used in studies investigating ALNM. STIR sequences stand out compared to other fat suppression techniques by providing more uniform fat suppression without being affected by magnetic field inhomogeneity. Although the signal-to-noise ratio is poor, it is useful because it includes both T1 and T2 contrast.

There is a correlation between tumor size and lymph node metastasis. 9 In our study, the maximum diameter of the tumor and axillary metastases showed a high correlation. Kurtosis and skewness features, which evaluate intralesional homogenetiy, were also correlated with ALNM.

Our study has some limitations. Its retrospective nature, small number of patients, and choosing the largest lesion in patients with more than one lesion are among the limitations of our study. Another limitation is that we did not divide the patients according to the number of metastatic lymph nodes.

In conclusion, radiomic features obtained from the primary tumor on STIR sequences have the potential to predict ALNM. STIR sequences are noninvasive and are currently used as a routine component of breast MRI in some centers. However, some lesions cannot be segmented since they have signal intensity close to parchyma in STIR sequences.

Ethics committee approval was obtained from our institution for this retrospective study (Approve no: 202351425036 ).

This research did not receive any funding grants from public, commercial, or nonprofit agencies.