Μηχανική μάθηση και ανάλυση εικόνων στον καρκίνο του εγκεφάλου
Machine learning and image analysis in brain cancer
View/
Keywords
Εγκεφαλικός όγκος ; MRI ; Transfer learning ; Convolutional neural networksAbstract
A tumour is an abnormal mass of tissue, which forms when cells grow and divide uncontrollably or in case they do not die when they should. More than 150 types of brain tumours have been identified by researchers, which are divided into benign and malignant (cancer). Both the similarity between the different types of brain tumour and the similarity between the cancerous and healthy cells surrounding them make the diagnosis of brain tumour patients particularly difficult. On the other hand, the high mortality rates of patients with malignant brain tumours make it necessary to detect and classify them at an early stage. One medical imaging technique, MRI helps doctors to detect and identify brain tumours, however, due to what has been described previously, this is a very demanding procedure that requires time and expertise. Progress in machine and deep learning allows medical specialists to provide easier and more reliable diagnosis. Traditional machine learning methods, use certain manual feature extraction techniques to perform classification of brain tumors. In contrast, the advantage of deep learning methods, such as those used in this thesis, is that they do not require any manual feature extraction. This project proposes a set of methods for detection and binary classification: tumour and no tumour brain images, from a public dataset consisting of a total of 3762 MRI images. Several approaches and modifications of InceptionV3, ResNet50 and MobileNetV2 models are proposed, using Transfer Learning. Additionally two custom made convolutional neural networks (CNNs) consisting of 8 and 10 layers respectively are used.
Postgraduate Studies Programme
Εφαρμοσμένη ΣτατιστικήDepartment
Σχολή Χρηματοοικονομικής και Στατιστικής. Τμήμα Στατιστικής και Ασφαλιστικής ΕπιστήμηςNumber of pages
87Language
GreekCollections
Except where otherwise noted, this item's license is described as
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα