A self-supervised vision transformer to predict survival from histopathology in renal cell carcinoma
Purpose To develop and validate an interpretable deep learning model to predict overall and disease-specific survival (OS/DSS) in clear cell renal cell carcinoma (ccRCC). Methods Digitised haematoxylin and eosin-stained slides from The Cancer Genome Atlas were used as a training set for a vision transformer (ViT) to extract image features with a self-supervised model called DINO (self-distillation with no labels). Extracted features were used in Cox regression models to prognosticate OS and DSS. Kaplan–Meier for univariable evaluation and Cox regression analyses for multivariable evaluation of the DINO-ViT risk groups were performed for prediction of OS and DSS. For validation, a cohort from a tertiary care centre was used. Results A significant risk stratification was achieved in univariable analysis for OS and DSS in the training (n = 443, log rank test, p < 0.01) and validation set (n = 266, p < 0.01). In multivariable analysis, including age, metastatic status, tumour size and grading, the DINO-ViT risk stratification was a significant predictor for OS (hazard ratio [HR] 3.03; 95%-confidence interval [95%-CI] 2.11–4.35; p < 0.01) and DSS (HR 4.90; 95%-CI 2.78–8.64; p < 0.01) in the training set but only for DSS in the validation set (HR 2.31; 95%-CI 1.15–4.65; p = 0.02). DINO-ViT visualisation showed that features were mainly extracted from nuclei, cytoplasm, and peritumoural stroma, demonstrating good interpretability. Conclusion The DINO-ViT can identify high-risk patients using histological images of ccRCC. This model might improve individual risk-adapted renal cancer therapy in the future..
Medienart: |
E-Artikel |
---|
Erscheinungsjahr: |
2023 |
---|---|
Erschienen: |
2023 |
Enthalten in: |
Zur Gesamtaufnahme - volume:41 |
---|---|
Enthalten in: |
World journal of urology - 41(2023), 8 vom: 29. Juni, Seite 2233-2241 |
Sprache: |
Englisch |
---|
Beteiligte Personen: |
Wessels, Frederik [VerfasserIn] |
---|
Links: |
Volltext [kostenfrei] |
---|
Themen: |
Artificial intelligence |
---|
Anmerkungen: |
© The Author(s) 2023 |
---|
doi: |
10.1007/s00345-023-04489-7 |
---|
funding: |
|
---|---|
Förderinstitution / Projekttitel: |
|
PPN (Katalog-ID): |
SPR05269304X |
---|
LEADER | 01000naa a22002652 4500 | ||
---|---|---|---|
001 | SPR05269304X | ||
003 | DE-627 | ||
005 | 20230811064745.0 | ||
007 | cr uuu---uuuuu | ||
008 | 230811s2023 xx |||||o 00| ||eng c | ||
024 | 7 | |a 10.1007/s00345-023-04489-7 |2 doi | |
035 | |a (DE-627)SPR05269304X | ||
035 | |a (SPR)s00345-023-04489-7-e | ||
040 | |a DE-627 |b ger |c DE-627 |e rakwb | ||
041 | |a eng | ||
100 | 1 | |a Wessels, Frederik |e verfasserin |4 aut | |
245 | 1 | 2 | |a A self-supervised vision transformer to predict survival from histopathology in renal cell carcinoma |
264 | 1 | |c 2023 | |
336 | |a Text |b txt |2 rdacontent | ||
337 | |a Computermedien |b c |2 rdamedia | ||
338 | |a Online-Ressource |b cr |2 rdacarrier | ||
500 | |a © The Author(s) 2023 | ||
520 | |a Purpose To develop and validate an interpretable deep learning model to predict overall and disease-specific survival (OS/DSS) in clear cell renal cell carcinoma (ccRCC). Methods Digitised haematoxylin and eosin-stained slides from The Cancer Genome Atlas were used as a training set for a vision transformer (ViT) to extract image features with a self-supervised model called DINO (self-distillation with no labels). Extracted features were used in Cox regression models to prognosticate OS and DSS. Kaplan–Meier for univariable evaluation and Cox regression analyses for multivariable evaluation of the DINO-ViT risk groups were performed for prediction of OS and DSS. For validation, a cohort from a tertiary care centre was used. Results A significant risk stratification was achieved in univariable analysis for OS and DSS in the training (n = 443, log rank test, p < 0.01) and validation set (n = 266, p < 0.01). In multivariable analysis, including age, metastatic status, tumour size and grading, the DINO-ViT risk stratification was a significant predictor for OS (hazard ratio [HR] 3.03; 95%-confidence interval [95%-CI] 2.11–4.35; p < 0.01) and DSS (HR 4.90; 95%-CI 2.78–8.64; p < 0.01) in the training set but only for DSS in the validation set (HR 2.31; 95%-CI 1.15–4.65; p = 0.02). DINO-ViT visualisation showed that features were mainly extracted from nuclei, cytoplasm, and peritumoural stroma, demonstrating good interpretability. Conclusion The DINO-ViT can identify high-risk patients using histological images of ccRCC. This model might improve individual risk-adapted renal cancer therapy in the future. | ||
650 | 4 | |a Artificial intelligence |7 (dpeaa)DE-He213 | |
650 | 4 | |a Deep learning |7 (dpeaa)DE-He213 | |
650 | 4 | |a Kidney neoplasms |7 (dpeaa)DE-He213 | |
650 | 4 | |a Treatment outcome |7 (dpeaa)DE-He213 | |
650 | 4 | |a Risk assessment |7 (dpeaa)DE-He213 | |
650 | 4 | |a Oncology |7 (dpeaa)DE-He213 | |
650 | 4 | |a Survival analysis |7 (dpeaa)DE-He213 | |
700 | 1 | |a Schmitt, Max |4 aut | |
700 | 1 | |a Krieghoff-Henning, Eva |4 aut | |
700 | 1 | |a Nientiedt, Malin |4 aut | |
700 | 1 | |a Waldbillig, Frank |4 aut | |
700 | 1 | |a Neuberger, Manuel |4 aut | |
700 | 1 | |a Kriegmair, Maximilian C. |4 aut | |
700 | 1 | |a Kowalewski, Karl-Friedrich |4 aut | |
700 | 1 | |a Worst, Thomas S. |4 aut | |
700 | 1 | |a Steeg, Matthias |4 aut | |
700 | 1 | |a Popovic, Zoran V. |4 aut | |
700 | 1 | |a Gaiser, Timo |4 aut | |
700 | 1 | |a von Kalle, Christof |4 aut | |
700 | 1 | |a Utikal, Jochen S. |4 aut | |
700 | 1 | |a Fröhling, Stefan |4 aut | |
700 | 1 | |a Michel, Maurice S. |4 aut | |
700 | 1 | |a Nuhn, Philipp |4 aut | |
700 | 1 | |a Brinker, Titus J. |0 (orcid)0000-0002-3620-5919 |4 aut | |
773 | 0 | 8 | |i Enthalten in |t World journal of urology |d Berlin : Springer, 1983 |g 41(2023), 8 vom: 29. Juni, Seite 2233-2241 |w (DE-627)SPR004296664 |w (DE-600)1463303-6 |x 1433-8726 |7 nnns |
773 | 1 | 8 | |g volume:41 |g year:2023 |g number:8 |g day:29 |g month:06 |g pages:2233-2241 |
856 | 4 | 0 | |u https://dx.doi.org/10.1007/s00345-023-04489-7 |z kostenfrei |3 Volltext |
912 | |a GBV_USEFLAG_A | ||
912 | |a GBV_SPRINGER | ||
951 | |a AR | ||
952 | |d 41 |j 2023 |e 8 |b 29 |c 06 |h 2233-2241 |