Detection of Out Of Focus Regions in Histology and Cytology images
Encadrants : A. BOUYSSOUX
Email : alexandre.bouyssoux@pasteur.fr
Disponible : OUI
Nombre d'étudiants : 1
Description : In Whole Slide Images (histology or cytology), acquired images often contains blurred / out of focus (OOF) regions. To prevent applying image analysis algorithms on those regions, they must be detected beforehand. Some recent studies showed Deep Learning methods can leverage such issue and accurately detect OOF digital histology patches.
https://arxiv.org/pdf/1901.04619.pdf
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6201886/
Pré-requis : Python, PyTorch (preferable)
Travail demandé : The student will have to:
- Review bibliography on OOF region detection in histology and cytology, including classical image analysis methods and ML / DL approaches
- Choose one ML / DL approach for extended study
- Implement select method (network and toy dataset creation if needed)
- Apply on histology patches classification
- Apply on (cytology) single cell classification
- Apply on (cytology) cell clumps segmentation : when some cells are in focus and some other cells are out of focus in a single group of cells or when on cell is partially out of focus.
