May 05, 2020
History
Fourier Transform Infrared microspectroscopy is often used to analyse complex systems such as cells or animal and vegetal tissues that produce complex signals that are hard to interpret.
The use of Focal Plane Array imagers gives very large datasets in short times and may overwhelm even the capacity of modern microcomputers.
Information extracted from these spectral data is no more limited to the position or intensity of a few absorption bands but may be scattered a large spectral domain.
The analysis is thus made more complex by the sheer quantities of data and the subtle nature of the sought after information, but also by the existence of confusing phenomena such as scattering, interference fringes, noise…
These make necessary to use specific signal processing and data analysis techniques that are not taught in classical university courses.
The SMIS beamline decided to organize a dedicated training on microspectroscopy data with a particular focus on biospectroscopy data which is at the heart of its expertise.
These techniques are nevertheless useful to analyse complex datasets from other spectrometry methods such as Raman microscopy, UV fluorescence spectroscopy, MS, XRF, NMR…
The SMIS beamline initiated a partnership with other synchrotrons µFTIR beamlines and the University of Ljubljana to develop and adapt QUASAR a powerful machine learning to the analysis of spectroscopic data. QUASAR makes powerful and sophisticated analysis and preprocessing tools available in a simple and intuitive manner.
The formation will have a short theoretical introduction on conventional analysis methods but will concentrate mainly on preprocessing methods and on multivariate statistical analysis (also known as machine learning or pattern recognition methods) for large infrared microspectroscopy datasets and hyperspectral images. The methods taught will allow using infrared spectra to classify samples based on their chemical composition and establish predictive models for classification and quantification.
The 2020 edition is the sixth edition to be held at SOLEIL and the training was also held at various other institutions (SLRI in Thailand in 2011, INRA in 2012 and 2018, SESAME in Jordan in 2018 and 2018 as part of HERCULES).
Training objectives
The training is aimed at scientists who are past, present, or future users of the SMIS beamline.
The course will focus on the analysis of infrared microspectroscopy data for by multivariate statistical analysis.
Hand-on training will be carried out on the Quasar (AKA Orange spectroscopy) software.
- Understanding the requirement for multivariate data analysis of infrared spectra for biomedical applications
- Introduction to multivariate data analysis principles and methods
- Preparing the data for analysis
- Inspecting data (plots, descriptive statistics)
- Understanding the different types of pre-processing
- Advanced pretreatments (EMSC, ATR)
- Comparison and classification methods (execution and interpretation)
- Principal Component Analysis (PCA)
- Hierarchical Cluster Analysis (HCA)
- Nearest Neighbour methods
- Identification and prediction methods (execution and interpretation)
- Partial Least Square Discriminant Analysis (PLS-DA)
- Soft Modelling By Class Analogy (SIMCA)
- Discriminant Analysis (DA)
- Random Forest Classification (RFC)
- Experimental and data analysis strategy
- Planning of experiments and analysis
- Data selection
- Validation and interpretation of results
This training is intended to give users (mostly biologists) that are not familiar with multivariate statistical analysis and machine learning, tools to analyze their data independently.
This will increase the output of the beamline and optimize the involvement of beamline scientists in helping users exploit their data.
Preliminary program
Wednesday May 27, 2020 |
Thursday May 28, 2020 |
Friday May 29, 2020 |
||
---|---|---|---|---|
Session 1 |
08:30 - 10:30 | Introduction to multivariate statistical analysis for IR: the need for MVA |
Principal Component Analysis: theory |
Imaging: |
10:30 - 10:45 | Coffe break / Pause | Coffe break / Pause | Coffe break / Pause | |
Session 2 |
10:45 - 12:15 | Introduction to multivariate statistical analysis: basic methods |
Principal Component Analysis: practical |
Imaging: |
12:15 - 13:30 | Lunch / Déjeuner | Lunch / Déjeuner | Lunch / Déjeuner | |
Session 3 |
13:30 - 15:30 | Preparing data: basic preprocessing methods, advanced preprocessing methods |
Regression methods:
|
Hand on practical user data 1 |
15:30 - 15:45 | Coffe break / Pause | Coffe break / Pause | Coffe break / Pause | |
Session 4 |
15:45 - 17:15 |
Basic clustering methods
|
Advanced clustering methods
|
Hand on practical user data 1 |
We will devote some time to the analysis of spectral images with the open software Quasar (https://www.synchrotron-soleil.fr/en/news/new-tools-data-analysis-orange-infrared-and-spectral-orange). A hand-on session with participant data will be held.
Organizing Committee:
Teachers, SMIS beamline:
Ferenc BORONDICS
Christophe SANDT
Marko TOPLAK
Local organizing:
Sylvie BONNARDEL
Frédérique FRAISSARD
- Registration open from March 31, 2020
- Registration deadline May 5, 2020
- The formation can accommodate only 20 participants .
- Registrations will be made on selection of the application file.
- The training will be in English.
- The registration fees cost 100 € and include coffee breaks, 3 lunches and 2 dinners.
- Accommodation (available at SOLEIL guesthouse) and travel will be at participant expense.
Accommodation
SOLEIL Guest House
L'Orme des Merisiers - 91190 GIF / YVETTE
Tel: +33(0)1 69 35 82 00
Email: hebergement@synchrotron-soleil.fr
- Reference: STAT IR6
- Night: 30 €
- Breakfast: 5 €
Access to SOLEIL
If you come by car:
Geographical address:
Synchrotron SOLEIL
L’Orme des Merisiers
Rond point du Golf de Saint Aubin
91190 Saint Aubin
Location coordinates:
Latitude : 48.711922
Longitude : 2.146156
Intersection between RD306 and D128
If you come by Public Transports:
From PARIS and CHARLES-DE-GAULLE Airport: RER B direction to SAINT-RÉMY-LÈS-CHEVREUSE and stop at:
"MASSY-PALAISEAU" station - BUS n°91-06 B ou C direction to SAINT-QUENTIN GARE, stop at “L’ORME DES MERISIERS”;
"LE GUICHET" station - walk to the bus station, take bus n°9, direction to SACLAY, stop at "L’ORME DES MERISIERS";
"GIF SUR YVETTE" station, take bus n° 10, stop at "SAINT AUBIN";
From ORLY Airport : ORLYVAL train, stop at ANTONY, take the RER B (then same as above) or BUS n°91-10 direction "Christ / N306, Saclay", stop at "L'ORME DES MERISIERS" (around 1h - end of service at 9 p.m).
Christophe SANDT,scientist on the SMIS beamline
Sylvie BONNARDEL, event adminsitrative