MR-radiomic-based pathological response prediction to neoadjuvant chemotherapy in breast cancer

In the evolving landscape of artificial intelligence (AI) and personalized medicine, the significance of employing supervised and unsupervised learning techniques has surged, particularly in the comprehensive analysis of medical images across diverse domains. This thesis is dedicated to leveraging AI to predict the pathological response to neoadjuvant chemotherapy (NAC) in breast cancer patients, emphasizing the attainment of a pathological complete response. Multiparametric Magnetic Resonance Imaging (MRI), routinely used for dense breast imaging and high-risk patient screening, holds promise in enhancing the clinical diagnosis of breast cancer. The initial phase of the research focuses on developing a robust deep learning algorithm for the automatic segmentation of breast lesions on DCE-MR images, with comparative analysis against manual segmentation by experienced radiologists. Subsequently, it delves into radiomic feature extraction from the segmented lesions and the development of a machine learning predictive model for treatment response. This process incorporates a meticulous selection method integrating multiple pieces of information. The study explores the application of radiomics, an advanced analysis method extracting numerous medical imaging features that might unveil key components of tumor phenotype. Acknowledging the multidimensional nature of radiomics, precise machine-learning methods play a crucial role in constructing predictive models.