An Approach for Clustering Subjects According to Similarities in Cell Distributions within Biopsies
Authors:
Yassine El Ouahidi,
Matis Feller,
Matthieu Talagas,
Bastien Pasdeloup
Abstract:
In this paper, we introduce a novel and interpretable methodology to cluster subjects suffering from cancer, based on features extracted from their biopsies. Contrary to existing approaches, we propose here to capture complex patterns in the repartitions of their cells using histograms, and compare subjects on the basis of these repartitions. We describe here our complete workflow, including creat…
▽ More
In this paper, we introduce a novel and interpretable methodology to cluster subjects suffering from cancer, based on features extracted from their biopsies. Contrary to existing approaches, we propose here to capture complex patterns in the repartitions of their cells using histograms, and compare subjects on the basis of these repartitions. We describe here our complete workflow, including creation of the database, cells segmentation and phenoty**, computation of complex features, choice of a distance function between features, clustering between subjects using that distance, and survival analysis of obtained clusters. We illustrate our approach on a database of hematoxylin and eosin (H&E)-stained tissues of subjects suffering from Stage I lung adenocarcinoma, where our results match existing knowledge in prognosis estimation with high confidence.
△ Less
Submitted 6 July, 2020; v1 submitted 30 June, 2020;
originally announced July 2020.
