3D Deep-Learning-Based Segmentation of Human Skin Sweat Glands and Their 3D Morphological Response to Temperature...

Area of Science:

• Dermatology
• Biomedical Imaging
• Computational Biology

Background:

• Sweat glands are vital for thermoregulation, and their morphology is key in diagnosing skin conditions.
• Existing methods for studying sweat glands are limited, being 2D, in vitro, and destructive.

Purpose of the Study:

• To develop a novel 3D segmentation framework for precise sweat gland analysis.
• To quantify in vivo, real-time, 3D sweat gland morphological changes in response to temperature variations.

Main Methods:

• Utilized a transformer-based segmentation framework for 3D sweat gland segmentation.
• Employed optical coherence tomography (OCT) for capturing skin volume data.
• Quantified morphological parameters including volume, surface area, length, and S/V ratio.

Main Results:

• Achieved precise 3D sweat gland segmentation from OCT data.
• Demonstrated significant increases in sweat gland volume (42.0%), surface area (26.4%), and length (12.8%) at 43°C compared to 10°C (p < 0.001).
• Observed a significant decrease in the S/V ratio with increasing temperature (p = 0.01).

Conclusions:

• The developed framework provides a non-invasive, real-time tool for quantifying 3D sweat gland structure.
• Establishes a benchmark for normal sweat gland morphology and its temperature-dependent variations.
• Facilitates research into individual variability and pathological changes in sweat glands for dermatology.