FPI Based Hyperspectral Imager for the Complex Surfaces-Calibration, Illumination and Applications.

Area of Science:

• Biomedical imaging
• Dermatological applications
• Hyperspectral imaging (HSI)

Background:

• Medical HSI offers non-invasive methods with high spatial and spectral resolution for analyzing skin.
• Challenges include complex skin surfaces and limited accessibility to certain areas.
• HSI can delineate tumors, detect invasions, and classify lesion types.

Purpose of the Study:

• Introduce a novel spectral imaging concept, SICSURFIS.
• Conduct a clinical pre-test to evaluate the concept's potential for clinical application.
• Address challenges of complex skin surfaces in dermatological HSI.

Main Methods:

• Developed the SICSURFIS spectral imager, combining a Fabry-Pérot interferometer (FPI) with a specialized LED module and stray light protection cones.
• Designed for photometric stereo imaging to generate 3D skin surface models at each wavelength.
• Captured hyperspectral images at 33 wavelengths (477-891 nm) using selected LEDs and three light angles.

Main Results:

• The SICSURFIS imager successfully accessed and imaged complex skin surfaces.
• Collected data enabled analysis in spectral and spatial domains, incorporating surface model information.
• Pixel-wise classification and delineation of healthy skin, basal cell carcinomas, and intradermal nevi were achieved with promising results using a convolutional neural network.

Conclusions:

• The SICSURFIS spectral imager concept shows potential for clinical dermatological applications.
• The pre-test demonstrated the imager's capability to handle complex skin surfaces and classify lesions.
• Further studies are necessary to fully develop and validate the SICSURFIS system for clinical use.