Adapting Self-Supervised Learning for Computational Pathology
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866911864491868160 |
|---|---|
| author | Zimmermann, Eric Tenenholtz, Neil Hall, James Shaikovski, George Zelechowski, Michal Casson, Adam Milletari, Fausto Viret, Julian Vorontsov, Eugene Liu, Siqi Severson, Kristen |
| author_facet | Zimmermann, Eric Tenenholtz, Neil Hall, James Shaikovski, George Zelechowski, Michal Casson, Adam Milletari, Fausto Viret, Julian Vorontsov, Eugene Liu, Siqi Severson, Kristen |
| contents | Self-supervised learning (SSL) has emerged as a key technique for training networks that can generalize well to diverse tasks without task-specific supervision. This property makes SSL desirable for computational pathology, the study of digitized images of tissues, as there are many target applications and often limited labeled training samples. However, SSL algorithms and models have been primarily developed in the field of natural images and whether their performance can be improved by adaptation to particular domains remains an open question. In this work, we present an investigation of modifications to SSL for pathology data, specifically focusing on the DINOv2 algorithm. We propose alternative augmentations, regularization functions, and position encodings motivated by the characteristics of pathology images. We evaluate the impact of these changes on several benchmarks to demonstrate the value of tailored approaches. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2405_01688 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Adapting Self-Supervised Learning for Computational Pathology Zimmermann, Eric Tenenholtz, Neil Hall, James Shaikovski, George Zelechowski, Michal Casson, Adam Milletari, Fausto Viret, Julian Vorontsov, Eugene Liu, Siqi Severson, Kristen Computer Vision and Pattern Recognition Self-supervised learning (SSL) has emerged as a key technique for training networks that can generalize well to diverse tasks without task-specific supervision. This property makes SSL desirable for computational pathology, the study of digitized images of tissues, as there are many target applications and often limited labeled training samples. However, SSL algorithms and models have been primarily developed in the field of natural images and whether their performance can be improved by adaptation to particular domains remains an open question. In this work, we present an investigation of modifications to SSL for pathology data, specifically focusing on the DINOv2 algorithm. We propose alternative augmentations, regularization functions, and position encodings motivated by the characteristics of pathology images. We evaluate the impact of these changes on several benchmarks to demonstrate the value of tailored approaches. |
| title | Adapting Self-Supervised Learning for Computational Pathology |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2405.01688 |