Vibrational spectroscopic image analysis of biological material using multivariate curve resolution - alternating least squares
Multivariate data analysis techniques are ideal to decrypt chemical differences between anatomical features or tissue areas in hyperspectral images of biological samples. This protocol provides a user-friendly pipeline and graphical user interface (GUI) for data pre-processing and un-mixing of pixel...
| Autores: | , , , , , |
|---|---|
| Tipo de recurso: | artículo |
| Estado: | Versión aceptada para publicación |
| Fecha de publicación: | 2015 |
| País: | España |
| Institución: | Universidad de Barcelona |
| Repositorio: | Dipòsit Digital de la UB |
| OAI Identifier: | oai:diposit.ub.edu:2445/127937 |
| Acceso en línea: | https://hdl.handle.net/2445/127937 |
| Access Level: | acceso abierto |
| Palabra clave: | Anàlisi multivariable Espectroscòpia d'infraroigs per transformada de Fourier Espectroscòpia Raman Processament d'imatges Multivariate analysis Fourier transform infrared spectroscopy Raman spectroscopy Image processing |
| Sumario: | Multivariate data analysis techniques are ideal to decrypt chemical differences between anatomical features or tissue areas in hyperspectral images of biological samples. This protocol provides a user-friendly pipeline and graphical user interface (GUI) for data pre-processing and un-mixing of pixel spectra into their contributing pure components by multivariate curve resolution-alternating least squares (MCR-ALS) analysis. The analysis considers the full spectral profile to identify the chemical compounds and to visualize their distribution across the sample to categorize chemically distinct areas. Results are rapidly achieved (usually less than 30 - 60 min/image) and are easy to interpret and evaluate both in terms of chemistry and biology, making the method generally more powerful than principal component analysis (PCA) or single band intensity heap maps. In addition, chemical and biological evaluation of the results by means of reference matching and segmentation maps (based on k-means clustering) are possible. |
|---|