once set - one click and you are done
Process your complex Raman data reliably, robustly and most accurately. Our multivariant and intelligent software will analyze your specific Raman spectroscopy data with only a few clicks. Biophotonics Diagnostics is your partner to build up your own spectral library and develop analytical methods that meet your specific needs, no matter if in development, production or quality control of foods, beverages or medicine, environmental analysis, agriculture, chemical processes or spotting counterfeits. Contact us for consultation!
An intuitive and easy to use interface enables you to navigate with comfort through all necessary data processing steps. Every step on the way of data elaboration overview plots are generated to make checks and balances by an experienced expert possible. Those plots can be saved as PNG- and CSV format or summarized as report.
Manage data processing parameters via a user-friendly graphical interface and adapt it according to the respectable user knowledge. Upload data and metadata, standardize Raman spectral data and different machine-learning models including deep learning as you need.
After preprocessing your data classification and regression models are implemented on the test data – even if completely unknown to the system – to determine a material or microbe identity or its concentration.
our easy-to-use data analyst pipeline
tailored specifically for your needs
1. Task or question – all the answers are already there, just ask
Using Raman spectroscopy provides you with an extensive data collection. You own all possible answers concerning your processes to the very detail. Challenge yourself to the questions that are key to improve and intensify your chemical production and quality checks. With RamanMetrix you might even unlock unexpected innovations crucial to future success.
2. Design of experiments – enlighten your path through the maze
Define objectives, determine factors and responses, select design, develop strategy and run the experiment supported by our services. RamanMetrix was designed and reviewed in close cooperation with spectroscopy experts, analytical chemists and microbiologists. We are a very reliable and experienced partner when it comes to designing the experiments to gather your spectral library data.
3. Spike correction – not every point matters
There is always something messing with your data. Be it cosmic rays, disturbances by experimental parameters or some effect of your very measuring device - we eliminate such contribution and keep your measured data safe and sound.
4. Spectrometer calibration – different instruments same tune
Ensure inter device comparability of your data by using our internal references and our calibration strategy for your device. Eliminate variations by instruments, temperature as well as physical and chemical states of the sample insignificant to your data set.
5. Pre-processing – the difference between jungle and garden
Treating your data by adjustment of baselines, background noise and spectral normalization is useful for a start, but there are many additional expert data analysis routines available.
Optimized, robust and repeatable - RamanMetrix has all you need for the most accurate analysis of Raman Spectra on board. All irrelevant and corrupted data is eliminated.
6. Quality rating – you name it
As quality rating is all about definition, we offer you various tools to set exactly those quality filters that matter for your processes.
7. Analysis Model – all shapes and sizes
Once set, you can run your data by RamanMetrix to get your answers by one click or in your own API. You adjust or change model sets with a few mouse clicks, making it more valuable and accurate within minutes.for controlling your own processes.
Contact us for consultation!
RamanMetrix is employable for various industries.
machine-learning for chemometrics
> developed specifically for Raman application in close collaboration with analytic chemists and microbiologists
> intuitive graphical user interface
> easy import of your spectral data supporting various formats
> removal of noise or irrelevant variation
> reprocessing like baseline offset and scatter, preview function, assess transformation effect and tune parameters
> diagnostics to evaluate the model fit – direct and automatical testing
> techniques for multivariate analysis and standard machine-learning techniques
> find a target substance, protein or microbe
qualitative and quantitative analysis
RamanMetrix is suited best for processing Raman spectra and learns to detect the presence and quantity of a target analyte in complex mixtures. You receive much more accurate results than conventional chemometric methods. Find target substances, proteins or microbes and know their concentration within your samples.
RamanMetrix supports Raman spectroscopical data in various formats such as TXT, CSV, LPE and SPC. After data import spectra are standardized by the following preprocessing and sorting steps:
adjustment of cosmic spectral peaks
adjustment of base lines
Advanced users might choose amongst many their very own preprocessing parameters manually. Averaged spectra are saved for every data process step for better control of data process routines. Advanced users may want to use data quality filters by setting thresholds for:
selected quality peaks (minimal and maximal)
correlation with centre spectrum
signal to noise ratio (minimal)
background intensity (maximum)
Integrated intensity of base-line adjusted spectra (minimum and maximum)
Models integrated in RamanMetrix
After preprocessing (optional quality check possible) surveilled or unsurveilled machine-learning models are constructed. To avoid over-adjustment of the model, most mentioned below methods implement dimension reduction by principal component analysis (PCA) or convolutional neural network (CNN). Depending if the goal is identification or determination of concentration regression or classification models are needed. The following models are available within RamanMetrix:
Linear discriminant analysis (LDA)
Random Forest (RF)
K-nearest neighbors (kNN)
Partial least squares discriminant analysis
Unsurveilled hierarchical Ward-Cluster-Analysis (HCR)
Partial least squares regression (PLSR)