Field of view extension in Mueller matrix microscopy for whole-slide imaging of biological samples

Quantitative measures of the interaction of polarized light with tissue have been established as a powerful tool for biomedical diagnosis in recent years. In this regard, we implemented a microscopy setup that incorporates a polarized sensor in the imaging plane to obtain Stokes parameters correspon...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autor principal: Demczylo, Roman (author)
Otros Autores: Silva Piedra, Diego (author), Lecumberry, Federico (author), Fernández, Ariel (author)
Formato: article
Lenguaje:inglés
Publicado: 2025
Materias:
Acceso en línea:https://www.sciencedirect.com/science/article/pii/S2666950125001440
https://hdl.handle.net/20.500.12008/52117
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
Descripción
Sumario:Quantitative measures of the interaction of polarized light with tissue have been established as a powerful tool for biomedical diagnosis in recent years. In this regard, we implemented a microscopy setup that incorporates a polarized sensor in the imaging plane to obtain Stokes parameters corresponding to a given Field of View (FoV) of a tissue sample. By illuminating with linearly independent States of Polarization in the input, Mueller matrix elements can also be retrieved from the same FoV of the sample. In order to achieve whole-slide imaging the FoV can be extended by stitching multiple images taken after XY displacement. We propose introducing polarimetric features, specifically the Mueller matrix norm for each pixel, into the stitching algorithm. This allows for FoV extension with minimal overlap between neighboring images, substantially reducing the total number of images required for the entire sample. This approach can significantly reduce acquisition time and data storage requirements for whole-slide MM imaging. Validation results for the retrieval of whole-slide MM of tissue samples show SSIM = 0.93±0.04 and 100% stitching success from images with overlapping as low as 35%.