SparseConvMIL: Sparse Convolutional Context-Aware Multiple Instance Learning for Whole Slide Image Classification

Marvin Lerousseau, Maria Vakalopoulou, Eric Deutsch, Nikos Paragios

Résultats de recherche: Contribution à un journal???researchoutput.researchoutputtypes.contributiontojournal.conferencearticle???Revue par des pairs

9 Citations (Scopus)

Résumé

Multiple instance learning (MIL) is the preferred approach for whole slide image classification. However, most MIL approaches do not exploit the interdependencies of tiles extracted from a whole slide image, which could provide valuable cues for classification. This paper presents a novel MIL approach that exploits the spatial relationship of tiles for classifying whole slide images. To do so, a sparse map is built from tiles embeddings, and is then classified by a sparse-input CNN. It obtained state-of-the-art performance over popular MIL approaches on the classification of cancer subtype involving 10, 000 whole slide images. Our results suggest that the proposed approach might (i) improve the representation learning of instances and (ii) exploit the context of instance embeddings to enhance the classification performance. The code of this work is open-source at https://github.com/MarvinLer/SparseConvMIL.

langue originaleAnglais
Pages (de - à)129-139
Nombre de pages11
journalProceedings of Machine Learning Research
Volume156
étatPublié - 1 janv. 2021
Modification externeOui
Evénement2021 MICCAI Workshop on Computational Pathology, COMPAY 2021 - Virtual, Online
Durée: 27 sept. 2021 → …

Contient cette citation