Radiomics and Artificial Intelligence in Predicting Axillary Lymph Node Metastasis in Breast Cancer: A Systematic Review

Abstract

Background: Breast cancer is the most common malignancy and the second most common cause of death in women worldwide. Axillary lymph node metastasis (ALNM) is the most significant prognostic factor in breast cancer. Under the current guidelines, sentinel lymph node biopsy (SLNB) is the standard of axillary staging in patients with clinically-node negative breast cancer. Despite the minimally invasive nature of SLNB, it can cause short and long-term morbidities including pain, sensory impairment, and upper limb motor dysfunction. However, lymphedema remains the most feared adverse event, and it affects 7% of patients within 36 months of follow-up. Recently, we have witnessed the implication of radiomics and artificial intelligence domains in the diagnosis and follow-up of many malignancies with promising results. Therefore, we have conducted a systematic search to investigate the potentials of radiomics and artificial intelligence in predicting ALNM. Methods: Four electronic databases were searched: PubMed, Scopus, CINAHL, and Web of Science. We used the Preferred Reporting Items for Systematic Reviews and Meta-Analysis as our basis of organization. Results: For radiomics, area under the curve (AUC) for the included studies ranged from 0.715 to 0.93. Accuracy ranged from 67.7% to 98%. Sensitivity and specificity ranged from 70.3% to 97.8% and 58.4% to 98.2%, respectively. For other artificial intelligence methods, AUC ranged from 0.68 to 0.98, while accuracy ranged from 55% to 89%. Conclusion: The results of radiomics and artificial intelligence in predicting ALNM are promising. However, validation as a substitute to SLNB requires more substantial evidence from large randomized trials.

Radiomics in cone-beam breast CT for the prediction of axillary lymph node metastasis in breast cancer: a multi-center multi-device study

“…In this study, radiomics models were developed and validated for preoperative prediction of ALN status and metastatic burden based on CE-CBBCT images. For both tasks, the AUCs of the radiomics model were significantly Previously, several mammography, ultrasound, and MRI studies have explored the application of radiomics analysis of primary tumors in predicting ALN status and metastatic burden, and achieved good prediction performance, with AUC ranging from 0.64 to 0.89 and from 0.74 to 0.79 [36][37][38], respectively. Some studies incorporated ultrasound or MRI report of ALN or clinicopathologic characteristics to construct nomogram for further Fig.…”

Future Directions in the Assessment of Axillary Lymph Nodes in Patients with Breast Cancer

“…Artificial Intelligence (AI), particularly deep learning (DL) algorithms using convolutional neural networks (CNNs), has gained popularity in the medical community for revolutionizing the diagnosis of diseases based on image analysis [ 45 , 46 ]. CNNs are a specific type of DL algorithm commonly used to analyze BC images, as they excel in determining image features [ 47 ]. These emerging innovations have shown to enhance the accuracy of US and MRI in assessing ALNs by providing automated image segmentation and radiomics feature extraction.…”