Cancer cell segmentation for cellularity prediction via a weakly labeled/strongly labeled hybrid convolutional neural network

David R. Chambers, Bradley B. Brimhall, Donald R. Poole, Edward A. Medina

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

In our work, we present an approach to regressing breast cancer cellularity in patches extracted from Whole Slide Imagery (WSI) on Hematoxylin and Eosin (H&E) stains using a fully-convolutional neural network which is trained with two heads: one which computes a global average pool for weakly-labeled data (data with a cellularity score of 0- 1.0) and another which enforces pixel-wise activations for strongly-labeled (segmentation) data. Our method was the top-performing algorithm of all submissions to the BreastPathQ challenge, achieving a prediction probability of 0.941.

Original languageEnglish (US)
Title of host publicationMedical Imaging 2022
Subtitle of host publicationDigital and Computational Pathology
EditorsJohn E. Tomaszewski, Aaron D. Ward, Richard M. Levenson
PublisherSPIE
ISBN (Electronic)9781510649538
DOIs
StatePublished - 2022
EventMedical Imaging 2022: Digital and Computational Pathology - Virtual, Online
Duration: Mar 21 2022Mar 27 2022

Publication series

NameProgress in Biomedical Optics and Imaging - Proceedings of SPIE
Volume12039
ISSN (Print)1605-7422

Conference

ConferenceMedical Imaging 2022: Digital and Computational Pathology
CityVirtual, Online
Period3/21/223/27/22

Keywords

  • cancer cellularity
  • convolutional neural networks
  • Digital pathology
  • whole slide imagery

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Atomic and Molecular Physics, and Optics
  • Biomaterials
  • Radiology Nuclear Medicine and imaging

Fingerprint

Dive into the research topics of 'Cancer cell segmentation for cellularity prediction via a weakly labeled/strongly labeled hybrid convolutional neural network'. Together they form a unique fingerprint.

Cite this