Cancer cell metastasis classification using image processing and convolutional neural networks
Ταξινόμηση μετάστασης καρκινικών κυττάρων με χρήση επεξεργασίας εικόνας και συνελεκτικών νευρωνικών δικτύων
Master Thesis
Author
Kefalas, Dimitrios - Christos
Κεφαλάς, Δημήτριος - Χρήστος
Date
2024-07Advisor
Prentza, AndrianaΠρέντζα, Ανδριάνα
View/ Open
Keywords
AFM ; CNN ; Metastasis ; Image processing ; Deep learningAbstract
Cancer metastasis occurs when a certain form of cancer, known as proteogenic
cancer, is spread from an original source to a different part of the body. Once spread,
cancer is much harder to treat; therefore, having a reliable method to predict cancer
metastasis can greatly enhance treatment strategies and help produce positive
outcomes.
Deep Learning (DL) algorithms and, more specifically, the Convolutional Neural
Network (CNN) architecture was employed since, at its core, this is an image classification
problem. Images depicting cancer cell tissue captured by an Atomic Force Microscope
(AFM) at the nanoscale are processed by the model, the result being the generation of an
accurate prediction as to whether said cancer is metastatic or non-metastatic. The
results mentioned above are compared to existing detection and prediction methods.
The convergence of AFM and DL represents a promising multidisciplinary
approach that leverages both technologies' strengths. As research progresses, adopting
these advanced methodologies in clinical practice could revolutionize early cancer
detection and treatment, ultimately contributing to better patient outcomes.