HLSS/HSR_Burns

HSR_Burn Dataset

Dataset Description

HSR_Burn is a small-scale validation dataset for burn total body surface area (TBSA) estimation, constructed from four high-resolution human scan models obtained from the HUMAN SCAN Repository. Burn patterns were manually annotated on repaired high-resolution meshes, and frontal and dorsal rendered images with corresponding ground-truth TBSA values are provided.

Dataset Structure

The dataset is organized into four folders, each corresponding to one human model:

Folder	Model ID	Gender	Height (cm)	BMI
HSR0018-Body-070	HSR0018-Body-070	Male	165	20.2
HSR0036-Body-071	HSR0036-Body-071	Male	171	28.0
HSR0046-Body-059	HSR0046-Body-059	Female	160	39.1
HSR0189-Body-092	HSR0189-Body-092	Female	168	19.5

Within each model folder, images are organized as follows:

• 01–09 (HSR_Burn_36): Burn patterns derived from real clinical burn configurations, applied in sequential accumulation (each configuration retains all burns from the preceding configuration while adding a new pattern). These patterns cover diverse anatomical regions including the face, trunk, upper and lower extremities.
• 10–15 (HSR_Burn_24): Burn patterns reproduced from Desbois et al. (2018), originally designed for physical mannequin experimentation. These patterns are relatively simple and regular in shape.

For each configuration, two images are provided:

• *_A.jpg: Frontal view
• *_B.jpg: Dorsal view

Each image shows the body surface with burn regions highlighted in red.

Ground-Truth TBSA

Ground-truth TBSA values were computed via automated mesh-area measurement on the repaired high-resolution 3D models. These values are provided in the accompanying metadata file. See project.blend file.

Intended Use

This dataset is intended for preliminary validation of TBSA estimation methods on high-resolution scan data. Due to the small sample size (four models, 60 total configurations), it should not be used for training and is not representative of the full diversity of real patient populations.