Visit our website: http://bit.ly/2GtXaiw
Hakima Ibaroudene, Group Leader – Research & Development, Southwest Research Institute
Abstract:
We demonstrate an approach to predicting cancer cellularity scores from patches extracted from Hematoxylin and Eosin (H&E) stained whole slide images. The dataset contains 2,579 training and validation patches with cellularity labels assigned by pathologists. Our method won the BreastPathQ challenge prize, achieving a prediction probability, Pk, of 0.941 on the BreastPathQ test set. Qualitatively, the pre-pooling activations demonstrate segmentation of the malignant cells in the patch images. Our method is also extremely fast, processing each patch in approximately 19ms, and a whole slide in a matter of minutes.
Add comment