BruNet improves bruise outlines using self-supervised vision transformer
BruNet: A Cross-Domain Transfer Framework for Bruise Segmentation
Computer Vision and Pattern Recognition
Summary
Bruises are hard to pinpoint in medical images because they sometimes look blurry and change a lot in appearance. The authors created BruNet, which uses a special image understanding method called a vision transformer combined with a technique that helps find shapes in images. BruNet was trained on a skin lesion dataset but then used to identify bruises without extra training. It works better than older methods and even approaches that use large language models to help. This is the first computer method that tries to map bruises precisely at the pixel level.
What this means in practice
- •For medical imaging developers: Create bruise segmentation tools that work well despite limited bruise-specific training data by using BruNet's cross-domain approach.
- •For forensic analysts: Use BruNet to automatically delineate bruises in images for improved injury documentation and assessment without manual outlining.
Authors
Qiming Wang, Richard J. Motley, Ebube E. Obi, Xianfang Sun, Paul L. Rosin
Abstract
Segmenting bruises is a challenging task in medical imaging due to limited data and annotations, diffuse boundaries, and highly variable appearance. In this work, we propose BruNet, a segmentation framework that combines a ViT-based visual encoder (a self-supervised DINOv3 or a pretrained LingBot-Vision backbone) with a SAM-based mask decoder. BruNet is trained on the HAM10000 skin lesion dataset and evaluated on a separate bruise dataset without additional fine-tuning. Although a small number of prior studies have explored machine learning and computer vision for bruise analysis, existing work has primarily focused on detection, classification, or colour analysis rather than pixel-level localisation. To the best of our knowledge, this is the first study to address automatic bruise segmentation. Our results show that BruNet outperforms CNN-based models, state-of-the-art segmentation models, ChatGPT-4o/5-assisted SAM2 zero-shot baselines, and the medical-oriented MedSAM model, demonstrating strong cross-domain generalisation to bruise segmentation.