Search

Bone tissue features a complex microarchitecture and biomolecular composition, which determine biomechanical properties. In addition to state-of-the-art technologies, innovative optical approaches allowing the characterization of the bone in native, label-free conditions can provide new, multi-level insight into this inherently challenging tissue. Here, we exploited multimodal nonlinear optical (NLO) microscopy, including co-registered stimulated Raman scattering, two-photon excited fluorescence, and second-harmonic generation, to image entire vertebrae of murine spine sections. The quantitative nature of these nonlinear interactions allowed us to extract accurate biochemical, morphological, and topological information on the bone tissue and to highlight differences between normal and pathologic samples. Indeed, in a murine model showing bone loss, we observed increased collagen and lipid content as compared to the wild type, along with a decreased craniocaudal alignment of bone collagen fibres. We propose that NLO microscopy can be implemented in standard histopathological analysis of bone in preclinical studies, with the ambitious future perspective to introduce this technique in the clinical practice for the analysis of larger tissue sections.

We introduce label-free multimodal nonlinear optical microscopy, based on different coherent Raman and multi-photon ultrafast light-matter interactions, as a noninvasive, quantitative and chemically selective tool to assess chemotherapy-induced senescence in human cancer cells.

We introduce a broadband coherent anti-Stokes Raman scattering (CARS) microscope based on a 2-MHz repetition rate ytterbium laser generating 1035-nm high-energy (≈µJ level) femtosecond pulses. These features of the driving laser allow producing broadband red-shifted Stokes pulses, covering the whole fingerprint region (400–1800 cm−1), employing supercontinuum generation in a bulk crystal. Our system reaches state-of-the-art acquisition speed (

Deep Reinforcement Learning (deep RL) is a branch of Machine Learning dealing with the solution of optimization problems, usually formalized as Markov Decision Processes [1] . At discrete temporal steps, an agent takes an action on the system receiving back a reward that depends on the state reached by the system. The goal of the agent is to determine an optimal policy, i.e. a map between states and actions, to maximise the future rewards, by directly experiencing and sampling the environment without any a-priori knowledge of the system. Among the various optimization algorithms, deep RL is particularly versatile thanks to the underlying neural networks [2] , a powerful universal approximator.

Stimulated Raman scattering (SRS) microscopy is a powerful technique for label-free identification of molecules based on their intrinsic vibrational spectrum. We present a novel approach to broadband SRS microscopy based on a recently developed balanced low-noise integrated-circuit multichannel lock-in amplifier with 10-μs integration time and 32 channels working in parallel, enabling multiplex SRS detection at speed faster than one hyperspectral frame per second. The system is powered by a narrowband Stokes pulse at 1040 nm and a broadband (approx. 500 cm^-1) pump pulse generated by a home-built low-noise optical parametric oscillator with up to 100-mW average power, covering the whole C-H stretching band. We measure stimulated Raman loss on the broadband pump pulse employing an in-line balanced detection approach to suppress the laser fluctuations and achieve close to shot-noise-limited sensitivity. We will show the system performances in high-resolution cell imaging.

We present innovative approaches to broadband coherent Raman scattering microscopy, both in the stimulated Raman scattering (SRS) and broadband coherent anti-Stokes Raman scattering (CARS) modalities. A convolutional neural network removes the unwanted non-resonant background from broadband CARS spectra. The deep-learning model processes experimental data in 100microseconds and correctly retrieves all the relevant vibrational peaks without any user intervention or independent background measurement. A multi-channel lock-in amplifier, in combination with in-line balanced detection and a broadband optical parametric oscillator, allows sensitive measurement of the SRS spectrum at 32 frequencies in parallel, with pixel dwell times as short as 40 microseconds.

As a consequence of the fourth industrial revolution, the data produced by companies’ day-by-day activities and the rate at which the transactions occur are growing exponentially. In order to extract business value from those data, they need to be organised under a reference conceptual model facilitating data analysis and decision making. Since no sound reference model for organising digital factory production data has been proposed in the literature, this paper aims at developing and testing a conceptual multidimensional model to support a broad range of data analytics activities for the management and optimisation of production in a smart factory. The testing of the model in a case study company of the printing sector provides insights into the applicability of the model and the connected benefits.

We introduce a novel approach based on deep learning to remove non-resonant background from coherent anti-Stokes Raman scattering spectra in real time. The model is trained on synthetic spectra and successfully applied to experimental data.

Broadband coherent Raman is a non-invasive and non-destructive spectroscopic technique with growing applications in the analysis of molecules and biological compounds. This paper presents the first 32-ch modular platform for highspeed Broadband Raman Imaging, able to simultaneously acquire and process 32 wavelengths of the spectrum with a multichannel pseudo-differential lock-in structure for the compensation of the excess noise given by the laser source. The system is based on a custom integrated CMOS front-end, specifically designed for Broadband Raman applications, and a Xilinx Artix-7 FPGA for the parallel acquisition and real-time data elaboration. Experimental results show that the system is able to reach the shot noise limit and acquire a Raman image with a pixel dwell time of 100μs.

Coherent anti-Stokes Raman scattering is an extremely powerful non-linear optical (NLO) microscopy technique for label-free vibrational imaging allowing for a detailed study of biological samples in their native state. To overcome the long acquisition times associated with raster sample scanning required in NLO microscopy, which impair real-time investigation of fast biological dynamics, we employ here wide-field signal generation over a large field of view, covering tens of micrometers. To this aim, we exploit an innovative approach based on the use of an amplified femtosecond ytterbium laser source delivering high energy (≈μJ) pulses in the near infrared. This enables the generation of stable broadband Stokes pulses to measure the entire fingerprint region of the molecular vibrational spectrum, the richest in chemical information. Our results pave the way for future translational applications and clinical diagnostics with video-rate imaging capabilities. ©2022 The Authors

Broadband Coherent Anti-Stokes Raman Scattering (B-CARS) is a powerful label-free nonlinear spectroscopy technique allowing one to measure the full vibrational spectrum of molecules and solids. B-CARS spectra, however, suffer from the presence of a spurious signal, called non-resonant background (NRB), which interferes with the resonant vibrational one, distorting the line shapes and degrading the chemical information. While several numerical techniques are available to remove this unwanted contribution and extract the resonant vibrational signal of interest, they all require the user’s intervention and sensitively depend on the spectral shape of the NRB, which needs to be measured independently. In this work, we present a novel approach to remove NRB from B-CARS spectra based on deep learning. Thanks to the high generalization capability offered by the deep architecture of the designed neural network, trained through realistic simulated spectra, our fully automated model is able to process B-CARS spectra in real time and independently of the detailed shape of the NRB spectrum. This results in fast extraction of vibrational spectra without requiring user intervention or the measurement of reference spectra. We expect that this model will significantly simplify and speed-up the analysis of B-CARS spectra in spectroscopy and microscopy.