National and Kapodistrian University of Athens
Ο ρόλος της τεχνητής νοημοσύνης στη διερεύνηση παθήσεων του ενδομητρίου σε υλικό κυτταρολογίας υγρής φάσης, Thin Prep
Abstract
dc:descriptionObjective: To investigate the potential value of classification and regression trees (CART) for the discrimination of benign from malignant endometrial nuclei and lesions. Material and method: The study has been performed on ThinPrep cytological material obtained by Endogyn cell sampler. In total 222 histologically confirmed cases have been studied, 117 benign cases, 62 malignant and 43 hyperplasias with or without atypia. From each case about 100 nuclei have been measured with an image analysis system, in total 22783 nuclei have been collected. Nuclei from 50% of the cases have been used to construct CART algorithms and knowledge extraction, the so called training set. The nuclei from the remaining cases has been used to evaluate the knowlde system (test set). Results: The application of CARTs gave an overall correct classification of endometrial cell nuclei equal to 85% with specificity 90.68% and sensitivity 72.05%. Based on the results of CART there are proposed two case classification methods having similar performance: overall accuracy in the range of 94-95% specificity 95% and sensitivity 91-94%. These results outperform the standard cytological diagnosis of the endometrium and explain the difficulty of the morphologists in the safe classification of endometrial hyperplasia. Conclusion: The proposed study highlights interesting features of endometrial nuclear morphology and provides a classification approach that after verification with additional cases, can be a usefull tool for the everyday practice of the cytological laboratory.
Degree
thesis:*- Grantor dc:publisher
- National and Kapodistrian University of Athens
- Year dc:date
- 2014
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Margari, Charalampia
- Μάργαρη, Χαραλαμπία
Subjects
dc:subject × 16- Τεχνητή νοημοσύνη
- Παθήσεις ενδομητρίου
- Κυτταρολογία υγρής φάσης
- Καρκίνος ενδομητρίου
- Δέντρα ταξινόμησης
- Artificial intelligence
- Endometrial lesions
- Liquid based cytology
- Endometrial cancer
- Thin prep
- Ιατρική και Επιστήμες Υγείας
- Κλινική Ιατρική
- Επιστήμες Υγείας
- Medical and Health Sciences
- Clinical Medicine
- Health Sciences
Rights
dc:rights- Statement dc:rights
-
- BY_NC_SA
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/42055