External site

By following this link, you are leaving Medically and entering a website that is not owned or controlled by Roche. Roche does not take any responsibility for access to or use of this website, nor for any content therein.

Leave site
Home Oncology MICCAI 2023
Roche at

MICCAI 2023

October 08 - October 12
Vancouver, CANADA
miccai.org
Add to calendar
Share feedback
Share page
-
Medical Materials
Coming soon

Sign up or login to unlock the full suite of MEDICALLY features
Personalised Healthcare
Oct 9 / Roche
Deep cellular embeddings: an explainable plug and play improvement for feature representation in histopathology
Deep-learning approaches can classify whole-slide images (WSIs) of H&E-stained pathology slides by analyzing data at different scales: whole-slide level, region level and cell level. Cell-level analysis may produce more detailed and explainable data than region- or whole-slide level analyses but is limited by a lack of sufficiently annotated training data. To overcome this problem, weakly supervised and multiple instance learning (MIL) approaches have been applied to WSI classification tasks. However, many existing models use only tile-level embeddings and these often fail to capture useful information from individual cells.
Oncology Diffuse large B cell lymphoma Breast Cancer Deep learning Artificial Intelligence Digital Pathology Imaging Machine Learning MICCAI-2023

Sign up or login to unlock the full suite of MEDICALLY features
Personalised Healthcare
Oct 8 / Roche
Cellular Features Based Interpretable Network for Classifying Cell-of-Origin from Whole Slide Images for Diffuse Large B-cell Lymphoma Patients
Cell-of-origin (COO) is a published prognostic method. Up to now, this classification requires either complex gene expression analysis or multiple immunohistochemistry (IHC) stains requiring expert scoring and assessment. In this paper, we aim to develop an effective and tissue-saving COO classification method based on H&E stained whole slide images (WSIs). Specifically, we develop a new approach named Cellular Features Based Interpretable Network (CellFiNet), by leveraging both interpretable cellular features derived from image tiles and attention based multi-instance learning (AMIL) framework to train a WSI classification model.
Oncology Lymphoma Diffuse large B cell lymphoma Deep learning Artificial Intelligence Digital Pathology Imaging Machine Learning MICCAI-2023

Sign up or login to unlock the full suite of MEDICALLY features
Personalised Healthcare
Oct 8 / Roche
Cellular Features Based Interpretable Network for Classifying Cell-of-Origin from Whole Slide Images for Diffuse Large B-cell Lymphoma Patients
Cell-of-origin (COO) is a published prognostic method. Up to now, this classification requires either complex gene expression analysis or multiple immunohistochemistry (IHC) stains requiring expert scoring and assessment. In this paper, we aim to develop an effective and tissue-saving COO classification method based on H&E stained whole slide images (WSIs). Specifically, we develop a new approach named Cellular Features Based Interpretable Network (CellFiNet), by leveraging both interpretable cellular features derived from image tiles and attention based multi-instance learning (AMIL) framework to train a WSI classification model.
Oncology Deep learning Artificial Intelligence Digital Pathology Imaging Machine Learning MICCAI-2023

This website is intended for healthcare professionals outside the United Kingdom (UK) and Australia. Registration status and prescribing information of medicinal products may differ between countries. Please refer to local product information for any medicinal products mentioned on this website.