SALVE 3D Silicon Wounds Dataset, WACV25

Brief description

Chronic wounds represent a significant health and economic burden worldwide. Effective treatments require wound clinical measurements, typically performed manually by specialized healthcare professionals.
Wound surface area is typically measured by performing planimetry of the wound bed. These procedures are not only invasive and cause patient discomfort but are also prone to errors due to ambiguous definitions of metrics and variations in professionals’ skill levels.

Most existing automatic approaches that perform wound analysis rely solely on 2D images or only consider previous-generation 3D reconstruction pipelines. We are interested in benchmarking recent photogrammetric toolboxes and neural rendering alternatives, e.g. NeRF and Gaussian splatting.

3D analysis of wounds allows for the computation of richer wound biomarkers. However, studies in this direction only considered previous-generation 3D reconstruction frameworks, which have recently been surpassed by highly optimized photogrammetric toolboxes and recent neural rendering alternatives.
To this end, we propose SALVE, a collection of 3 common wound types including pressure wounds and a surgical dehiscence.
Lineage: We decided to source silicon realistic wounds from an Australian company TraumaSim which specialises in providing realistic simulations for health professionals' training.
We collected the data with 2 different consumer devices: iPhone 14 Pro Max; and Logitech 4k webcam. The ground truth in point cloud format was acquired with a Revopoint POP 3D scanner.
We acquired short videos of the wounds in a controlled environment at the CSIRO Herston site.

In order to make the data digestible by 3D reconstruction pipelines, we selected frames from the video sequences following criteria based on the sharpness of the frames.
For more detailed information, our website https://remichierchia.github.io/SALVE/ provides a link to the arXiv paper.

The ethical clearance for this project is 2022_015_LR.