Automatic Detection of Healthy or Cancerous cells [AD HOC²]

This project aims at analizing cytological images in order to identify and characterize malignant cells. The goal is to reduce subjectivity due to omission or misinterpretation of important information carried by the cytological image.

Current work

The current work has two main branches

• the creation of an Olena.Olena program to compute the number of cells in an image and describe their attributes: size, area, compactness, and so on,
• the creation of a decision program to dissociate cancerous cells from non cancerous cells or unknown cells.

This work is still in progress, however you can have a look at some intermediate reports :

• Image Segmentation Applied to Cytology.
• Evidenz

For the scheduling, see our calendar

Related articles and web pages

Some images

We have two kinds of images of cells based on two different methods :

• images from Papanicolaous' smears,
• images from the spot method. In this case, the enlargement is 10 times the number in the name of the file, spot40.tif means enlarged 400 times.

People working on the project

• LRDE :
  Olivier Ricou
  Reda Dehak
  Niels van Vliet
  Astrid Wang
  Youssef El Fassy Firhy
  David Lesage

• Institut Curie :
  Philippe Vielh

---

Ad Hoc² Zone

• AdHoc:
• AdHocAIVoc:
• AdHocAlgo:
• AdHocArticles:
• AdHocDiagClue:
• AdHocImage:
• AdHocMedicalVoc:
• AdHocVocabulary:
• AdHocWeb:

to top