Συντάχθηκε 13-10-2015 07:47
από Esthir Gelasaki
Email συντάκτη: egelasaki<στο>tuc.gr
Ενημερώθηκε:
-
Ιδιότητα: υπάλληλος.
ΠΟΛΥΤΕΧΝΕΙΟ ΚΡΗΤΗΣ
Σχολή Ηλεκτρονικών Μηχανικών και Μηχανικών Υπολογιστών
Πρόγραμμα Προπτυχιακών Σπουδών
ΠΑΡΟΥΣΙΑΣΗ ΔΙΠΛΩΜΑΤΙΚΗΣ ΕΡΓΑΣΙΑΣ
ΑΝΔΡΟΥΛΑΣ ΣΑΒΒΑ
με θέμα
Analysis and quantification of spatiotemporal features of diagnostics importance in cervical neoplasia
Τετάρτη 14 Οκτωβρίου 2015, 12πμ
Αίθουσα 137.Β73, Κτίριο Επιστημών, Πολυτεχνειούπολη
Εξεταστική Επιτροπή
Καθηγητής Κωνσταντίνος Μπάλας (επιβλέπων)
Καθηγητής Μιχάλης Ζερβάκης
Αναπληρωτής Καθηγητής Μιχαήλ Λαγουδάκης
Abstract
Cervical cancer is the fourth most common malignant disease in women worldwide. It poses a major threat, especially in developing regions, due to the high cost of screening and treatment, as well as the lack of fast, accurate diagnostic methods. Acetowhitening (AW) is a phenomenon observed after the application of an acetic acid (AA) solution on the cervix. Acetic acid interacts with pre-cancerous cells of the epithelium, altering the tissue’s scattering characteristics and causing abnormal cells to appear opaque. As part of the diagnostic chain, both Pap test and colposcopy, suffer from low sensitivity, since visual assessment of AW is subjective, qualitative and depends heavily on the visual acuity and training of the examiner. Quantitative, objective assessment of AW characteristics was first presented in DySIS, a novel instrumentation developed in the late 90’s. DySIS acquires a sequence of images over time, recording the development of the AW phenomenon. In this thesis, we present a novel algorithm that extracts features of diagnostics importance by exploiting only spatial information and is based on a physics model that encapsulates the basic biochemical procedures producing the AW. Using a single snapshot of the cervix, 144 seconds after the application of the AA solution and without prior knowledge, we identify the area and grade of the lesions utilizing bio optical properties of the tissue. The introduced method is fast, cheap and achieves high sensitivity (80%) and specificity (79%), simplifying the current diagnostic chain and showing promising results in aiding screening especially for low resource see and treat facilities.
