ML4CervicalAdequacy – Machine learning approaches for automated image quality assessment and specimen adequacy of cervical cytology smears

Description:

This work emerges in the ambit of the CLARE project, which aims to achieve results that exceed the current state-of-the-art in cervical cancer screening, by creating a Computer Aided-Diagnosis (CADx) system that can be easily integrated in the conventional clinical workflow. Particularly, this project will explore automated approaches based on computer vision and machine learning for image quality assessment and specimen adequacy of liquid-based cytology (LBC) smears.

 

Outcome:

Expand FhP background knowledge in:
- Unsupervised and semi-supervised image annotation
- Object detection and segmentation strategies based on deep learning
- Classification strategies based on deep learning
- Expected direct impact in task execution of current and future MICRON projects, such as CLARE and TAMI.

 

Author: Vladyslav Mosiichuk

Type: MSc thesis

Partner: ISEP - Instituto Superior de Engenharia do Porto

Year: 2021