To pretrain or not to pretrain? A case study of domain-specific pretraining for semantic segmentation in histopathology

by   Tushar Kataria, et al.

Annotating medical imaging datasets is costly, so fine-tuning (or transfer learning) is the most effective method for digital pathology vision applications such as disease classification and semantic segmentation. However, due to texture bias in models trained on real-world images, transfer learning for histopathology applications might result in underperforming models, which necessitates the need for using unlabeled histopathology data and self-supervised methods to discover domain-specific characteristics. Here, we tested the premise that histopathology-specific pretrained models provide better initializations for pathology vision tasks, i.e., gland and cell segmentation. In this study, we compare the performance of gland and cell segmentation tasks with domain-specific and non-domain-specific pretrained weights. Moreover, we investigate the data size at which domain-specific pretraining produces a statistically significant difference in performance. In addition, we investigated whether domain-specific initialization improves the effectiveness of out-of-domain testing on distinct datasets but the same task. The results indicate that performance gain using domain-specific pretraining depends on both the task and the size of the training dataset. In instances with limited dataset sizes, a significant improvement in gland segmentation performance was also observed, whereas models trained on cell segmentation datasets exhibit no improvement.


Efficient Domain Adaptation of Language Models via Adaptive Tokenization

Contextual embedding-based language models trained on large data sets, s...

Deep Transfer-Learning for patient specific model re-calibration: Application to sEMG-Classification

Accurate decoding of surface electromyography (sEMG) is pivotal for musc...

Effective Transfer of Pretrained Large Visual Model for Fabric Defect Segmentation via Specifc Knowledge Injection

Fabric defect segmentation is integral to textile quality control. Despi...

Unbiasing Semantic Segmentation For Robot Perception using Synthetic Data Feature Transfer

Robot perception systems need to perform reliable image segmentation in ...

Pengembangan Domain Specific Language Untuk Pengelolaan Data Warehouse

Efforts to improve the performance of services on the transaction at a b...

Learning High-Resolution Domain-Specific Representations with a GAN Generator

In recent years generative models of visual data have made a great progr...

Prior Knowledge-Guided Attention in Self-Supervised Vision Transformers

Recent trends in self-supervised representation learning have focused on...

Please sign up or login with your details

Forgot password? Click here to reset