Researchers demonstrate improved analysis of complex spectral signatures
About Optica
05 November 2024
Researchers demonstrate improved analysis of complex spectral signatures
Hyperspectral unmixing method lays groundwork for better disease detection in tissue samples
Researchers report a new approach for hyperspectral unmixing that can be used to analyze data from dyed histological tissue samples containing a mixture of known and unknown information. By enhancing the ability to separate complex spectral signatures, the new approach could lead to more efficient and precise detection of diseases in tissue samples.
Juan N. Mendoza Chavarría from Universidad Autónoma de San Luis Potosí in Mexico will present this research at the Latin America Optics and Photonics Congress, held 10 – 14 November 2024 in Puerto Vallarta, Mexico.
“This research is important because it helps us look at complex samples, like tissue from the body, in a way that reveals details the doctors couldn’t see before,” said Chavarría. “By improving the way we analyze these samples, they can more accurately detect diseases, which means earlier and more precise diagnosis and treatment of health problems.”
Hyperspectral images capture a wide spectrum of light across many wavelengths, which can be used to identify and analyze different materials based on their spectral signatures. Analyzing hyperspectral images typically requires hyperspectral unmixing to separate mixed pixels into their pure spectral components (end-members) to gain information about the spatial contributions (abundances) of those components.
It can be challenging to apply spectral unmixing when some information about the components of a sample is known while other components are entirely unknown. This often occurs with dyed histological samples, which are used for diagnosing and studying cancer and other diseases.
To address this unmixing problem, the researchers extended the mathematical formulation of the extended blind end-member and abundance extraction method to create a new approach, which they call extended semi-supervised end-member and abundance extraction (ESEAE).
They tested ESEAE using a visible and near-infrared dataset of breast histological samples dyed with the commonly used histology stains eosin and hematoxylin. The spectral range of the images ranged from 400 to 1000 nm with a spectral resolution of 2.8 nm, which generated hyperspectral images with 400×1004 pixels and 128 spectral bands after pre-processing. The researchers applied a mask to remove areas with no tissue.
After manually tuning the ESEAE hyperparameters to improve the unmixing performance, they applied ESEAE to separate the known spectral information of the dyes from the unknown biological information of the tissue. The results showed that the new method was able to successfully unmix hyperspectral information for this type of scenario, where the spectral signatures of some elements are known but the rest need to be estimated.
Next, the researchers plan to use the method as a feature extraction step in hyperspectral image classification.
“Beyond histology, this spectral unmixing approach has potential applications in fields like food safety and environmental monitoring,” said Chavarría. “For example, it could help detect adulterants in food, or identify pollutants in environmental samples. In histology specifically, this method could improve the precision of identifying and separating different tissue components, leading to better disease diagnosis and monitoring. By effectively separating known and unknown spectral signatures, this approach enables more detailed analysis, and helps handle data variability caused by changing conditions, different sampling scenarios, and the inherent complexity of biological samples This could ultimately enhance our understanding of complex samples and improve patient outcomes.”
About the Optica Latin America Optics and Photonics Conference (LAOP)
LAOP is the major international conference sponsored by Optica in Latin America with the explicit objective to promote Latin American excellence in optics and photonics research and support the regional community. LAOP is a peer-reviewed, international meeting with content presented in English, which enables maximum international participation. Featuring a comprehensive technical program with recognized experts in fields critical to Latin America, the conference covers all major areas of optics and photonics and features the latest research results that are making an impact in fundamental research and applications.
About Optica
Optica, Advancing Optics and Photonics Worldwide, is the society dedicated to promoting the generation, application, archiving and dissemination of knowledge in the field. Founded in 1916, it is the leading organization for scientists, engineers, business professionals, students and others interested in the science of light. Optica's renowned publications, meetings, online resources and in-person activities fuel discoveries, shape real-life applications and accelerate scientific, technical and educational achievement. Discover more at: Optica.org
Media Contact
