Image analysis in digital pathology based on machine learning techniques & deep neural networks
Ανάλυση ιατρικών εικόνων μικροσκοπίας με χρήση τεχνικών μηχανικών μάθησης και βαθιών νευρωνικών δικτύων
View/
Keywords
Image classification ; Breast cancer ; Convolutional neural networks ; Deep learning ; Digital pathology ; Epithelium segmentation ; Instance Segmentation ; Instance segmentation ; Machine learning ; Nuclei segmentation ; Object detection ; Tubule segmentation ; Tumor proliferation ; Computer visionAbstract
Detection of regions of interest (e.g. mitosis or histologic primitives) in Whole Slide Images in a clinical setting is a highly subjective and labor-intensive task. In this thesis we explore recent developments in Machine Learning and Computer Vision algorithms to assess their possible usage and performance in tasks such as the above, in order to enhance and accelerate healthcare procedures. A state-of-the-art Deep Learning framework (Detectron2) is trained on the TUPAC16 dataset for object detection, and on the JPATHOL dataset for instance segmentation. We evaluate its predictions against competing models and discuss further possible improvements.
Postgraduate Studies Programme
Ψηφιακά Συστήματα και ΥπηρεσίεςDepartment
Σχολή Τεχνολογιών Πληροφορικής και Επικοινωνιών. Τμήμα Ψηφιακών ΣυστημάτωνNumber of pages
85Language
EnglishCollections
Except where otherwise noted, this item's license is described as
Attribution-NonCommercial-NoDerivatives 4.0 Διεθνές
Attribution-NonCommercial-NoDerivatives 4.0 Διεθνές