The Genomics and AI in Personalized Medicine (GAIPM) aims to advance cutting-edge research at the intersection of genomics, artificial intelligence, and individualized healthcare. The journal promotes scientific innovation in precision diagnostics, genome-based therapies, predictive analytics, and machine-learning–driven medical solutions. It seeks high-quality research that enhances understanding of genetic variation, disease prediction models, and AI-enabled clinical decision systems. GAIPM serves as a global platform for researchers, clinicians, and biomedical data scientists to contribute knowledge that transforms patient-centered medicine and improves healthcare outcomes.
Article Views: 557
Background: Radiomics combined with machine-learning has turned out to be a promising initiative in the field of breastcancer imaging that provides quantitative and non-invasive biomarkers to improve diagnostic and prognostic accuracy. Even though the technological advancement is very fast, issues about reproducibility, interpretability, and clinical implementation readiness still exist. Objective: The objective of the current review was to discuss the applications of ML-based radiomics in breast-cancer imaging and to compare the diagnostic accuracy, predictive performance, reproducibility, and translational potential of the different imaging modalities. Methods: Twenty peer-reviewed publications, published in the period between 2018 and 2025, were synthesized in the form of a narrative and included MRI, mammography, ultrasound, and PET/CT-based radiomics. Data that were extracted included imaging protocols, type of features, ML algorithms, performance measures and validation measures. Due to the high heterogeneity in the methods, the results were summarized narratively and thematically. Results: The performance of multipara metric MRI and hybrid radiomics deep-learning models indicated a high-diagnostic and-prognostic performance, with area-under-the-curve (AUC) of over 0.90 of lesion classification, nodal staging, recurrenceprognostication and treatment-response. Deep-learning radiomics using ultrasound also had strong non-invasive predictive power. The key limitations among studies involved inconsistency in the imaging acquisition, lack of external validation, and the lack of interpretation of intricate models. Conclusions: ML enhanced radiomics is a significant step towards precision diagnostics in breast cancer. However, standardized processes, explainable model structures, and multi-institutional validation are necessary in order to ensure clinical reliability, and to support translation into everyday practice.
breast cancer, radiomics, machine learning algorithms, deep learning, artificial intelligence, multi-parametric MRI, axillary lymph node, tumor heterogeneity, federated learning.