Hibou: A Family of Foundational Vision Transformers for Pathology
Dmitry Nechaev, Alexey Pchelnikov, Ekaterina Ivanova
TL;DR
This work introduces Hibou, a family of foundational vision transformers for pathology pretrained with DINOv2 on a massive, diverse WSIs dataset. Hibou-B and Hibou-L achieve state-of-the-art results across patch-level, slide-level, and segmentation benchmarks, with Hibou-L attaining the highest average accuracy and Hibou-B offering strong, resource-efficient performance; Hibou-B is open-sourced to enable reproducibility and broader adoption. The study demonstrates robust generalization across diverse tissues, stains, and tasks, promising impactful applications in digital pathology and potential integration into larger vision-language systems. The authors highlight data scale as a key driver of performance and outline plans to extend pretraining, evaluation, and LVLM-based usage of Hibou models.
Abstract
Pathology, the microscopic examination of diseased tissue, is critical for diagnosing various medical conditions, particularly cancers. Traditional methods are labor-intensive and prone to human error. Digital pathology, which converts glass slides into high-resolution digital images for analysis by computer algorithms, revolutionizes the field by enhancing diagnostic accuracy, consistency, and efficiency through automated image analysis and large-scale data processing. Foundational transformer pretraining is crucial for developing robust, generalizable models as it enables learning from vast amounts of unannotated data. This paper introduces the Hibou family of foundational vision transformers for pathology, leveraging the DINOv2 framework to pretrain two model variants, Hibou-B and Hibou-L, on a proprietary dataset of over 1 million whole slide images (WSIs) representing diverse tissue types and staining techniques. Our pretrained models demonstrate superior performance on both patch-level and slide-level benchmarks, surpassing existing state-of-the-art methods. Notably, Hibou-L achieves the highest average accuracy across multiple benchmark datasets. To support further research and application in the field, we have open-sourced the Hibou models, which can be accessed at https://github.com/HistAI/hibou.