Spectroscopic techniques are increasingly utilized in various disciplines such as biology, medicine, and diagnostics. This increase in application scenarios is linked to an improvement of the experimental setups, but it is also fostered by the development of computational data science methods. With the help of these data science methods the detection and extraction of bio-medical information from subtle differences in biomedical Raman spectra is feasible. The high-level information depends on the task and the sample, for example, the prediction of tissue types, disease states or properties of the samples like concentrations of constituents. A spectroscopic technique with many advantages is Raman spectroscopy, which yields a non-destructive fingerprint of the sample [1]. To utilize the full potential of bio-medical Raman spectra, the whole data life cycle of the spectroscopic data from its generation, via the data modelling and to the archiving is important and must be studied in a holistic way. Relevant points within the data life cycle are the experiment design, the sample size planning, the data pre-treatment, the data pre-processing, chemometric and machine learning based data modelling, model transfer methods and transfer learning. All procedures are sequentially combined in a data pipeline, which standardizes the data and extracts reliable high-level information from the Raman spectral data. Within the presentation, our recent studies aiming to construct a standardized data analysis pipeline for bio-medical Raman spectra [2] will be presented. A major focus will be on the comparability of Raman spectra between instruments and labs, which was studied in a European ring trial [3].

In this talk I’ll discuss different options based on linear and non linear spectroscopies aimed to obtain a label free fingerprint of molecular compounds. Single point or imaging methods will be discussed to extend this analysis to morphofunctional charaterization of tissues. This will allow to introduce the concept of liquid biopsy when analising organ pathologies in a non invasive modality or online optical biopsies for direct tissue characterizations. Multimodal spectroscopic approach will be described to show its enhancement in the specificity and the sensitivity of the pathology assessment. Different methodological approaches will be discussed, from non linear scanning microscopy for cell level analysis to high density hyperspectral imaging for large area characterization, to fiber optics detection for high modality spectral detection. Same examples ranging from skin to brain, up to liquid biopsy analisis, will be also discussed.