Artificial Intelligence and Machine Learning in Breast Cancer: From Imaging Diagnosis to Treatment Response Prediction

Breast cancer diagnosis and treatment are shaped by molecular heterogeneity – hormone receptor status, HER2 expression, proliferation markers, and genomic subtype – that determines both prognosis and therapeutic strategy. This complexity, combined with the large volume of imaging, pathology, and longitudinal treatment data generated for each patient, makes breast cancer one of the most active areas for artificial intelligence (AI) and machine learning (ML) research in oncology.In diagnostic imaging and pathology, machine learning models integrating multimodal ultrasound have been developed to distinguish luminal from non-luminal breast cancer subtypes, while MRI radiomic models predict Ki-67 expression and triple-negative subtype based on features extracted from the tumor and even the contralateral fibroglandular tissue. Digital pathology has advanced through ensemble machine learning models that predict relapse from histopathological images and deep learning classifiers applied directly to breast histopathology slides, alongside germline DNA methylation profiling that has revealed immune-related signatures using machine learning approaches. Fine-needle aspiration-based diagnostic models further extend AI’s reach into minimally invasive diagnostic workflows, offering the potential to reduce unnecessary biopsies and accelerate time to diagnosis.On the treatment side, machine learning is increasingly used to predict response and resistance to targeted and systemic therapies. Models have been developed to predict progression-free survival in hormone receptor-positive, HER2-negative metastatic breast cancer treated with CDK4/6 inhibitors, while circulating tumor DNA combined with machine learning has been used in large retrospective cohorts to predict treatment outcome and progression-free survival in metastatic disease. Radiomic and systemic inflammation-marker-based models have been applied to assess neoadjuvant chemotherapy efficacy, and generative AI is entering the therapeutic decision space directly, with studies evaluating the quality of ChatGPT-generated therapy recommendations for breast cancer treatment.This Research Topic invites contributions that advance the application of AI and ML to the diagnosis, molecular characterization, and treatment of breast cancer. To gather further insights into this important and clinically relevant field, we welcome original research, review, and perspective articles addressing, but not limited to, the following themes:• Machine learning and radiomics for breast cancer molecular subtyping (luminal, HER2, triple-negative)• Digital pathology and deep learning for histopathological classification and relapse prediction• AI-based prediction of response and resistance to endocrine therapy and CDK4/6 inhibitors• Circulating tumor DNA and liquid biopsy approaches combined with machine learning• Radiomic and multi-omic biomarkers for neoadjuvant and systemic treatment response prediction• Ultrasound, MRI, and mammography-based AI models for diagnosis and screening• Large language models and generative AI in treatment recommendation and patient communication• Predictive modeling of treatment-related toxicity and quality of life in breast cancer patients• Validation, generalizability, and clinical implementation of AI models in breast oncology Topic editors: Grete Francesca Privitera, Daniela Sambataro, Fiorenza Gianì Source: https://www.frontiersin.org/research-topics/87039/artificial-intelligence-and-machine-learning-in-breast-cancer-from-imaging-diagnosis-to-treatment-response-prediction