In this paper we present a longitudinal study of bacteria metabolism performed with a novel Raman spectrometer system. Longitudinal study is possible with our Raman setup since the overall procedure to localize a single bacterium and collect a Raman spectrum lasts only 1 minute. Localization and detection of single bacteria are performed by means of lensfree imaging, whereas Raman signal (from 600 to 3200 cm -1 ) is collected into a prototype spectrometer that allows high light throughput (HTVS technology, Tornado Spectral System). Accomplishing time-lapse Raman spectrometry during growth of bacteria, we observed variation in the net intensities for some band groups, e.g. amides and proteins. The obtained results on two different bacteria species, i.e. Escherichia coli and Bacillus subtilis clearly indicate that growth affects the Raman chemical signature. We performed a first analysis to check spectral differences and similarities. It allows distinguishing between lag, exponential and stationary growth phases. And the assignment of interest bands to vibration modes of covalent bonds enables the monitoring of metabolic changes in bacteria caused by growth and aging. Following the spectra analysis, a SVM (support vector machine) classification of the different growth phases is presented. In sum this longitudinal study by means of a compact and low-cost Raman setup is a proof of principle for routine analysis of bacteria, in a real-time and non-destructive way. Real-time Raman studies on metabolism and viability of bacteria pave the way for future antibiotic susceptibility testing.

In this paper we present results on single bacteria rapid identification obtained with a low-cost and compact Raman spectrometer. At present, we demonstrate that a 1 minute procedure, including the localization of single bacterium, is sufficient to acquire comprehensive Raman spectrum in the range of 600 to 3300 cm -1 . Localization and detection of single bacteria is performed by means of lensfree imaging over a large field of view of 24 mm 2 . An excitation source of 532 nm and 30 mW illuminates single bacteria to collect Raman signal into a Tornado Spectral Systems prototype spectrometer (HTVS technology). The acquisition time to record a single bacterium spectrum is as low as 10 s owing to the high light throughput of this spectrometer. The spectra processing features different steps for cosmic spikes removal, background subtraction, and gain normalization to correct the residual inducted fluorescence and substrate fluctuations. This allows obtaining a fine chemical fingerprint analysis. We have recorded a total of 1200 spectra over 7 bacterial species (E. coli, Bacillus species, S. epidermis, M. luteus, S. marcescens). The analysis of this database results in a high classification score of almost 90 %. Hence we can conclude that our setup enables automatic recognition of bacteria species among 7 different species. The speed and the sensitivity (<30 minutes for localization and spectra collection of 30 single bacteria) of our Raman spectrometer pave the way for high-throughput and non-destructive real-time bacteria identification assays. This compact and low-cost technology can benefit biomedical, clinical diagnostic and environmental applications.