Separating Surface Reflectance from Volume Reflectance in Medical Hyperspectral Imaging

Area of Science:

• Biomedical optics
• Medical imaging
• Spectroscopy

Background:

• Hyperspectral imaging (HSI) shows potential for cancer surgery diagnostics.
• Non-bulk tissue spectral variations complicate HSI data analysis.
• Standard normalization techniques can cause loss of amplitude and scattering information.

Purpose of the Study:

• To develop a novel method for addressing spectral variations in hyperspectral images.
• To separate surface and volume reflectance in hyperspectral data.
• To improve the analysis of hyperspectral images of biological tissues.

Main Methods:

• Developed a novel approach to separate surface and volume reflectance in hyperspectral images.
• Utilized a specialized illumination setup with a hyperspectral camera at varying axial positions and constant zenith angles.
• Implemented a data analysis technique to estimate and separate directional surface reflectance and omnidirectional volume reflectance.

Main Results:

• Validated the method with optical phantoms, achieving an 83% reduction in spectral variability.
• Demonstrated the method's effectiveness on excised human breast tissue.
• Successfully addressed variations from surface reflectance and glare while preserving crucial surface reflectance information.

Conclusions:

• The novel method effectively reduces spectral variability in hyperspectral imaging of tissues.
• This technique enhances sample analysis and evaluation by preserving surface reflectance information.
• The method is adaptable for various fields utilizing hyperspectral imaging, especially for samples with unknown refractive index spectra.