Your browser doesn't support javascript.
loading
Deep learning generates synthetic cancer histology for explainability and education.
Dolezal, James M; Wolk, Rachelle; Hieromnimon, Hanna M; Howard, Frederick M; Srisuwananukorn, Andrew; Karpeyev, Dmitry; Ramesh, Siddhi; Kochanny, Sara; Kwon, Jung Woo; Agni, Meghana; Simon, Richard C; Desai, Chandni; Kherallah, Raghad; Nguyen, Tung D; Schulte, Jefree J; Cole, Kimberly; Khramtsova, Galina; Garassino, Marina Chiara; Husain, Aliya N; Li, Huihua; Grossman, Robert; Cipriani, Nicole A; Pearson, Alexander T.
Afiliación
  • Dolezal JM; Section of Hematology/Oncology, Department of Medicine, University of Chicago Medicine, Chicago, IL, USA.
  • Wolk R; Department of Pathology, University of Chicago Medicine, Chicago, IL, USA.
  • Hieromnimon HM; Section of Hematology/Oncology, Department of Medicine, University of Chicago Medicine, Chicago, IL, USA.
  • Howard FM; Section of Hematology/Oncology, Department of Medicine, University of Chicago Medicine, Chicago, IL, USA.
  • Srisuwananukorn A; Tisch Cancer Institute, Icahn School of Medicine at Mount Sinai, New York, NY, USA.
  • Karpeyev D; Ghost Autonomy, Inc., Mountain View, CA, USA.
  • Ramesh S; Section of Hematology/Oncology, Department of Medicine, University of Chicago Medicine, Chicago, IL, USA.
  • Kochanny S; Section of Hematology/Oncology, Department of Medicine, University of Chicago Medicine, Chicago, IL, USA.
  • Kwon JW; Department of Pathology, University of Chicago Medicine, Chicago, IL, USA.
  • Agni M; Department of Pathology, University of Chicago Medicine, Chicago, IL, USA.
  • Simon RC; Department of Pathology, University of Chicago Medicine, Chicago, IL, USA.
  • Desai C; Department of Pathology, University of Chicago Medicine, Chicago, IL, USA.
  • Kherallah R; Department of Pathology, University of Chicago Medicine, Chicago, IL, USA.
  • Nguyen TD; Department of Pathology, University of Chicago Medicine, Chicago, IL, USA.
  • Schulte JJ; Department of Pathology and Laboratory Medicine, University of Wisconsin at Madison, Madison, WN, USA.
  • Cole K; Department of Pathology, University of Chicago Medicine, Chicago, IL, USA.
  • Khramtsova G; Department of Pathology, University of Chicago Medicine, Chicago, IL, USA.
  • Garassino MC; Section of Hematology/Oncology, Department of Medicine, University of Chicago Medicine, Chicago, IL, USA.
  • Husain AN; Department of Pathology, University of Chicago Medicine, Chicago, IL, USA.
  • Li H; Department of Pathology, University of Chicago Medicine, Chicago, IL, USA.
  • Grossman R; University of Chicago, Center for Translational Data Science, Chicago, IL, USA.
  • Cipriani NA; Department of Pathology, University of Chicago Medicine, Chicago, IL, USA. nicole.cipriani@bsd.uchicago.edu.
  • Pearson AT; Section of Hematology/Oncology, Department of Medicine, University of Chicago Medicine, Chicago, IL, USA. alexander.pearson@bsd.uchicago.edu.
NPJ Precis Oncol ; 7(1): 49, 2023 May 29.
Article en En | MEDLINE | ID: mdl-37248379
Artificial intelligence methods including deep neural networks (DNN) can provide rapid molecular classification of tumors from routine histology with accuracy that matches or exceeds human pathologists. Discerning how neural networks make their predictions remains a significant challenge, but explainability tools help provide insights into what models have learned when corresponding histologic features are poorly defined. Here, we present a method for improving explainability of DNN models using synthetic histology generated by a conditional generative adversarial network (cGAN). We show that cGANs generate high-quality synthetic histology images that can be leveraged for explaining DNN models trained to classify molecularly-subtyped tumors, exposing histologic features associated with molecular state. Fine-tuning synthetic histology through class and layer blending illustrates nuanced morphologic differences between tumor subtypes. Finally, we demonstrate the use of synthetic histology for augmenting pathologist-in-training education, showing that these intuitive visualizations can reinforce and improve understanding of histologic manifestations of tumor biology.

Texto completo: 1 Colección: 01-internacional Base de datos: MEDLINE Tipo de estudio: Prognostic_studies Idioma: En Revista: NPJ Precis Oncol Año: 2023 Tipo del documento: Article País de afiliación: Estados Unidos Pais de publicación: Reino Unido

Texto completo: 1 Colección: 01-internacional Base de datos: MEDLINE Tipo de estudio: Prognostic_studies Idioma: En Revista: NPJ Precis Oncol Año: 2023 Tipo del documento: Article País de afiliación: Estados Unidos Pais de publicación: Reino Unido