Presentation
7 March 2023 Deep learning for hyperspectral imaging of human skin and automated extraction of its functional and optical properties
Author Affiliations +
Abstract
Current report considers development of several Machine Learning (ML) classifiers for the quantitative functional imaging of human skin using polarized hyperspectral imaging. A validated optical model of has been combined with Monte Carlo-based computational approach and subsequently used in the training of Deep Learning methods. While demonstrating >98% accuracy for detecting important tissue features such as blood/oxygenation, pigment, etc. content we present detailed results of numerous data tests, training processes and the avenues for improving the performance of the developed techniques for newly captured data. The proposed techniques have a great potential to be implemented in low-coast clinical setting/wearable devices for e.g. monitoring and diagnosis of chronic skin ulcers and other relevant diseases.
Conference Presentation
© (2023) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Alexander Doronin, Igor V. Meglinski, and Alexander V. Bykov "Deep learning for hyperspectral imaging of human skin and automated extraction of its functional and optical properties", Proc. SPIE PC12378, Dynamics and Fluctuations in Biomedical Photonics XX, PC1237807 (7 March 2023); https://doi.org/10.1117/12.2649398
Advertisement
Advertisement
KEYWORDS
Skin

Hyperspectral imaging

Optical properties

Tissue optics

Data processing

Tissues

Light scattering

Back to Top