Multi-Class Cell Detection Using Spatial Context Representation
Encadrants : Vannary Meas-Yedid
Email : vmeasyed@pasteur.fr
Disponible : OUI
Spécialité : IMA
Nombre d'étudiants : 1
Description : The development of automated methods for nuclear segmentation and classification is a critical prerequisite task for the computational pathology. It allows quantitative analysis of tens of thousands of nuclei within whole-slide-images (WSIs).
However, automated segmentation and classification of nuclei faces a major challenge: there are many different types of nuclei, some of which, such as tumor cell nuclei, show great intra-class variability.
Pré-requis : Python
Travail demandé : This work addresses the multi-class cell detection by using spatial context representation [1].
After a review on existing cell detection, experiments will be performed on whole slide images and compared with Hovernet model.
Liens complémentaires : [1] Shahira Abousamra, Rajarsi Gupta, Tahsin Kurc, Dimitris Samaras, Joel Saltz, Chao Chen: "Topology-Guided Multi-Class Cell Context Generation for Digital Pathology", in IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2023
[2] Simon Graham, Quoc Dang Vu, Shan E Ahmed Raza, Ayesha Azam, Yee Wah Tsang, Jin Tae Kwak, and Nasir Rajpoot. Hover-net: Simultaneous segmentation and classificationof nuclei in multi-tissue histology images. Medical Image Analysis, 58:101563, 2019.
