Presentation
17 March 2023 Comparison of pre-processing techniques to reduce non-tissue related variations in hyperspectral reflectance imaging
Author Affiliations +
Proceedings Volume PC12363, Multiscale Imaging and Spectroscopy IV; PC123630C (2023) https://doi.org/10.1117/12.2647777
Event: SPIE BiOS, 2023, San Francisco, California, United States
Abstract
Hyperspectral reflectance imaging can be used to develop tissue classification algorithms, which is often based on machine learning. To improve the performance of classification algorithms, pre-processing is often used to remove variations in data not related to the tissue itself. In hyperspectral imaging, these variations are the result of reflections from the tissue surface (glare) and height variations within and between tissue samples. We investigated and quantified the performance of 8 commonly used pre-processing algorithms to reduce differences in spectra due to glare and height differences, while retaining contrast between tissues with different optical properties on simulated and clinical datasets.
Conference Presentation
© (2023) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Mark Witteveen, Ton A.G.J. M. Leeuwen, Maurice Aalders, Henricus Sterenborg, Theo Ruers, and Anouk L. Post "Comparison of pre-processing techniques to reduce non-tissue related variations in hyperspectral reflectance imaging", Proc. SPIE PC12363, Multiscale Imaging and Spectroscopy IV, PC123630C (17 March 2023); https://doi.org/10.1117/12.2647777
Advertisement
Advertisement
KEYWORDS
Tissues

Algorithm development

Hyperspectral imaging

Reflectivity

Machine learning

Medicine

Natural surfaces

Back to Top