Weakly-supervised tumor purity prediction from frozen H&E stained slides
Summary: Background: Estimating tumor purity is especially important in the age of precision medicine. Purity estimates have been shown to be critical for correction of tumor sequencing results, and higher purity samples allow for more accurate interpretations from next-generation sequencing result...
Main Authors: | , , , , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Elsevier
2022-06-01
|
Series: | EBioMedicine |
Subjects: | |
Online Access: | http://www.sciencedirect.com/science/article/pii/S2352396422002481 |
_version_ | 1811258268469690368 |
---|---|
author | Matthew Brendel Vanesa Getseva Majd Al Assaad Michael Sigouros Alexandros Sigaras Troy Kane Pegah Khosravi Juan Miguel Mosquera Olivier Elemento Iman Hajirasouliha |
author_facet | Matthew Brendel Vanesa Getseva Majd Al Assaad Michael Sigouros Alexandros Sigaras Troy Kane Pegah Khosravi Juan Miguel Mosquera Olivier Elemento Iman Hajirasouliha |
author_sort | Matthew Brendel |
collection | DOAJ |
description | Summary: Background: Estimating tumor purity is especially important in the age of precision medicine. Purity estimates have been shown to be critical for correction of tumor sequencing results, and higher purity samples allow for more accurate interpretations from next-generation sequencing results. Molecular-based purity estimates using computational approaches require sequencing of tumors, which is both time-consuming and expensive. Methods: Here we propose an approach, weakly-supervised purity (wsPurity), which can accurately quantify tumor purity within a digitally captured hematoxylin and eosin (H&E) stained histological slide, using several types of cancer from The Cancer Genome Atlas (TCGA) as a proof-of-concept. Findings: Our model predicts cancer type with high accuracy on unseen cancer slides from TCGA and shows promising generalizability to unseen data from an external cohort (F1-score of 0.83 for prostate adenocarcinoma). In addition we compare performance of our model on tumor purity prediction with a comparable fully-supervised approach on our TCGA held-out cohort and show our model has improved performance, as well as generalizability to unseen frozen slides (0.1543 MAE on an independent test cohort). In addition to tumor purity prediction, our approach identified high resolution tumor regions within a slide, and can also be used to stratify tumors into high and low tumor purity, using different cancer-dependent thresholds. Interpretation: Overall, we demonstrate our deep learning model's different capabilities to analyze tumor H&E sections. We show our model is generalizable to unseen H&E stained slides from data from TCGA as well as data processed at Weill Cornell Medicine. Funding: Starr Cancer Consortium Grant (SCC I15-0027) to Iman Hajirasouliha. |
first_indexed | 2024-04-12T18:10:42Z |
format | Article |
id | doaj.art-00e402cf736a4c4ba9c5ef3c9f127b32 |
institution | Directory Open Access Journal |
issn | 2352-3964 |
language | English |
last_indexed | 2024-04-12T18:10:42Z |
publishDate | 2022-06-01 |
publisher | Elsevier |
record_format | Article |
series | EBioMedicine |
spelling | doaj.art-00e402cf736a4c4ba9c5ef3c9f127b322022-12-22T03:21:50ZengElsevierEBioMedicine2352-39642022-06-0180104067Weakly-supervised tumor purity prediction from frozen H&E stained slidesMatthew Brendel0Vanesa Getseva1Majd Al Assaad2Michael Sigouros3Alexandros Sigaras4Troy Kane5Pegah Khosravi6Juan Miguel Mosquera7Olivier Elemento8Iman Hajirasouliha9Department of Physiology and Biophysics, Institute for Computational Biomedicine, Weill Cornell Medicine, New York, NY, USADepartment of Physiology and Biophysics, Institute for Computational Biomedicine, Weill Cornell Medicine, New York, NY, USA; Department of Theoretical and Applied Science, Ramapo College of New Jersey, Mahwah, NJ, USAEnglander Institute for Precision Medicine, Weill Cornell Medicine, New York, NY, USA; Department of Pathology and Laboratory Medicine, Weill Cornell Medicine, New York, NY, USAEnglander Institute for Precision Medicine, Weill Cornell Medicine, New York, NY, USADepartment of Physiology and Biophysics, Institute for Computational Biomedicine, Weill Cornell Medicine, New York, NY, USA; Englander Institute for Precision Medicine, Weill Cornell Medicine, New York, NY, USAEnglander Institute for Precision Medicine, Weill Cornell Medicine, New York, NY, USADepartment of Physiology and Biophysics, Institute for Computational Biomedicine, Weill Cornell Medicine, New York, NY, USA; Englander Institute for Precision Medicine, Weill Cornell Medicine, New York, NY, USA; Computational Oncology, Department of Epidemiology and Biostatistics, Memorial Sloan Kettering Cancer Center, New York 10021, USAEnglander Institute for Precision Medicine, Weill Cornell Medicine, New York, NY, USA; Department of Pathology and Laboratory Medicine, Weill Cornell Medicine, New York, NY, USADepartment of Physiology and Biophysics, Institute for Computational Biomedicine, Weill Cornell Medicine, New York, NY, USA; Englander Institute for Precision Medicine, Weill Cornell Medicine, New York, NY, USA; The Meyer Cancer Center, Weill Cornell Medicine, New York, NY, USADepartment of Physiology and Biophysics, Institute for Computational Biomedicine, Weill Cornell Medicine, New York, NY, USA; Englander Institute for Precision Medicine, Weill Cornell Medicine, New York, NY, USA; The Meyer Cancer Center, Weill Cornell Medicine, New York, NY, USA; Corresponding author at: Department of Physiology and Biophysics, Institute for Computational Biomedicine, Weill Cornell Medicine, New York, NY, USA.Summary: Background: Estimating tumor purity is especially important in the age of precision medicine. Purity estimates have been shown to be critical for correction of tumor sequencing results, and higher purity samples allow for more accurate interpretations from next-generation sequencing results. Molecular-based purity estimates using computational approaches require sequencing of tumors, which is both time-consuming and expensive. Methods: Here we propose an approach, weakly-supervised purity (wsPurity), which can accurately quantify tumor purity within a digitally captured hematoxylin and eosin (H&E) stained histological slide, using several types of cancer from The Cancer Genome Atlas (TCGA) as a proof-of-concept. Findings: Our model predicts cancer type with high accuracy on unseen cancer slides from TCGA and shows promising generalizability to unseen data from an external cohort (F1-score of 0.83 for prostate adenocarcinoma). In addition we compare performance of our model on tumor purity prediction with a comparable fully-supervised approach on our TCGA held-out cohort and show our model has improved performance, as well as generalizability to unseen frozen slides (0.1543 MAE on an independent test cohort). In addition to tumor purity prediction, our approach identified high resolution tumor regions within a slide, and can also be used to stratify tumors into high and low tumor purity, using different cancer-dependent thresholds. Interpretation: Overall, we demonstrate our deep learning model's different capabilities to analyze tumor H&E sections. We show our model is generalizable to unseen H&E stained slides from data from TCGA as well as data processed at Weill Cornell Medicine. Funding: Starr Cancer Consortium Grant (SCC I15-0027) to Iman Hajirasouliha.http://www.sciencedirect.com/science/article/pii/S2352396422002481Deep LearningComputational pathologyTumor purity estimationPrecision medicine |
spellingShingle | Matthew Brendel Vanesa Getseva Majd Al Assaad Michael Sigouros Alexandros Sigaras Troy Kane Pegah Khosravi Juan Miguel Mosquera Olivier Elemento Iman Hajirasouliha Weakly-supervised tumor purity prediction from frozen H&E stained slides EBioMedicine Deep Learning Computational pathology Tumor purity estimation Precision medicine |
title | Weakly-supervised tumor purity prediction from frozen H&E stained slides |
title_full | Weakly-supervised tumor purity prediction from frozen H&E stained slides |
title_fullStr | Weakly-supervised tumor purity prediction from frozen H&E stained slides |
title_full_unstemmed | Weakly-supervised tumor purity prediction from frozen H&E stained slides |
title_short | Weakly-supervised tumor purity prediction from frozen H&E stained slides |
title_sort | weakly supervised tumor purity prediction from frozen h e stained slides |
topic | Deep Learning Computational pathology Tumor purity estimation Precision medicine |
url | http://www.sciencedirect.com/science/article/pii/S2352396422002481 |
work_keys_str_mv | AT matthewbrendel weaklysupervisedtumorpuritypredictionfromfrozenhestainedslides AT vanesagetseva weaklysupervisedtumorpuritypredictionfromfrozenhestainedslides AT majdalassaad weaklysupervisedtumorpuritypredictionfromfrozenhestainedslides AT michaelsigouros weaklysupervisedtumorpuritypredictionfromfrozenhestainedslides AT alexandrossigaras weaklysupervisedtumorpuritypredictionfromfrozenhestainedslides AT troykane weaklysupervisedtumorpuritypredictionfromfrozenhestainedslides AT pegahkhosravi weaklysupervisedtumorpuritypredictionfromfrozenhestainedslides AT juanmiguelmosquera weaklysupervisedtumorpuritypredictionfromfrozenhestainedslides AT olivierelemento weaklysupervisedtumorpuritypredictionfromfrozenhestainedslides AT imanhajirasouliha weaklysupervisedtumorpuritypredictionfromfrozenhestainedslides |