PhD Defense: 'Algorithms and software for Image-Based Analysis and Prediction in Breast Cancer and Other Applications'

Breast cancer diagnosis and treatment benefit from computer vision and machine learning (ML). This thesis presents automated methods for immunohistochemical (IHC) image quantification, prediction of pathological complete response (pCR) in HER2-positive breast cancer, and image-based bread quality assessment. BreastAnalyser software enables recognition of diaminobenzidine (DAB) regions, nuclei detection, staining classification, and expert-assisted quantification in IHC images, reducing manual effort and improving consistency. Machine learning models predict pCR using clinicopathological and MRI data. A computer vision framework estimates bread sensory characteristics from bread images. The proposed methods demonstrated reliable performance, supporting decision-making in biomedical research and food quality evaluation.

Supervisors: Eva Cernadas and Manuel Fernández Delgado.