Abstract
Spectral imaging is a powerful technology that uses spatially referenced spectral signatures to create informative visual maps of sample surfaces that can reveal more than what conventional RGB-visual images can show. Indeed, different spectroscopy modalities can provide different information about the same sample: for instance, Laser-Induced Breakdown Spectroscopy (LIBS) imaging can detect the presence of specific elements on the surface, while Raman imaging can identify the molecular structures and compositions of the sample, both of which have potential applications in various industrial processes, from quality control to material sorting. In the path from science to technology, the increasing accessibility to such solutions and the strong market pull have opened a window of opportunity for innovative multimodal imaging solutions, where information from distinct sources is set to be combined in order to enhance the capabilities of the single modality system. However, the practical implementation of multimodal spectral imaging is still a challenge, despite its theoretical potential, and as such, it is yet to be achieved. In this work, we will go over multimodal spectral knowledge distillation, a disruptive approach to multimodal spectral imaging techniques that tries to explore the combination of two techniques to capitalize on their individual strengths. In specific, this approach allows us to utilize one technique as an autonomous supervisor for the other, leveraging the higher degree of knowledge and interpretability of one of the techniques to increase the performance and transparency of the other. We present some example scenarios with LIBS and HSI and Raman spectroscopy and LIBS, discussing the impact of this new approach for scientific and technological applications.

Abstract
As lithium-bearing minerals become critical raw materials for the field of energy storage and advanced technologies, the development of tools to accurately identify and differentiate these minerals is becoming essential for efficient resource exploration, mining, and processing. Conventional methods for identifying ore minerals often depend on the subjective observation skills of experts, which can lead to errors, or on expensive and time-consuming techniques such as Inductively Coupled Plasma Mass Spectrometry (ICP-MS) or Optical Emission Spectroscopy (ICPOES). More recently, Raman Spectroscopy (RS) has emerged as a powerful tool for characterizing and identifying minerals due to its ability to provide detailed molecular information. This technique excels in scenarios where minerals have similar elemental content, such as petalite and spodumene, by offering distinct vibrational information that allows for clear differentiation between such minerals. Considering this case study and its particular relevance to the lithium- mining industry, this manuscript reports the development of an unsupervised methodology for lithium-mineral identification based on Raman Imaging. The deployed machine-learning solution provides accurate and interpretable results using the specific bands expected for each mineral. Furthermore, its robustness is tested with additional blind samples, providing insights into the unique spectral signatures and analytical features that enable reliable mineral identification.

Abstract
Recent advances in optical trapping have opened new opportunities for manipulating micro and nanoparticles, establishing optical tweezers (OT) as a powerful tool for single-cell analysis. Furthermore, intelligent systems have been developed to characterize these particles, as information about their size and composition can be extracted from the scattered radiation signal. In this manuscript, we aim to explore the potential of optical tweezers for the characterization of sub-micron size variations in microparticles. We devised a case study, aiming to assess the limits of the size discrimination ability of an optical tweezer system, using transparent 4.8 µm PMMA particles, functionalized with streptavidin. We focused on the heavily studied streptavidin-biotin system, with streptavidin-functionalized PMMA particles targeting biotinylated bovine serum albumin. This binding process results in an added molecular layer to the particle’s surface, increasing its radius by approximately 7 nm. An automatic OT system was used to trap the particles and acquire their forward-scattered signals. Then, the signals’ frequency components were analyzed using the power spectral density method followed by a dimensionality reduction via the Uniform Manifold Approximation and Projection algorithm. Finally, a Random Forest Classifier achieved a mean accuracy of 94% for the distinction of particles with or without the added molecular layer. Our findings demonstrate the ability of our technique to discriminate between particles that are or are not bound to the biotin protein, by detecting nanoscale changes in the size of the microparticles. This indicates the possibility of coupling shape-changing bioaffinity tools (such as APTMERS, Molecular Imprinted Polymers, or antibodies) with optical trapping systems to enable optical tweezers with analytical capability. © 2024 SPIE.

Abstract
Multimodal spectral imaging offers a unique approach to the enhancement of the analytical capabilities of standalone spectroscopy techniques by combining information gathered from distinct sources. In this manuscript, we explore such opportunities by focusing on two well-known spectral imaging techniques, namely laser-induced breakdown spectroscopy, and hyperspectral imaging, and explore the opportunities of collaborative sensing for a case study involving mineral identification. In specific, the work builds upon two distinct approaches: a traditional sensor fusion, where we strive to increase the information gathered by including information from the two modalities; and a knowledge distillation approach, where the Laser Induced Breakdown spectroscopy is used as an autonomous supervisor for hyperspectral imaging. Our results show the potential of both approaches in enhancing the performance over a single modality sensing system, highlighting, in particular, the advantages of the knowledge distillation framework in maximizing the potential benefits of using multiple techniques to build more interpretable models and paving for industrial applications.

Abstract
Laser-induced breakdown spectroscopy (LIBS) imaging has now a well-established position in the subject of spectral imaging, leveraging multi-element detection capabilities and fast acquisition rates to support applications both at academic and technological levels. In current applications, the standard processing pipeline to explore LIBS imaging data sets revolves around identifying an element that is suspected to exist within the sample and generating maps based on its characteristic emission lines. Such an approach requires some previous expert knowledge both on the technique and on the sample side, which hinders a wider and more transparent accessibility of the LIBS imaging technique by non-specialists. To address this issue, techniques based on visual analysis or peak finding algorithms are applied on the average or maximum spectrum, and may be employed for automatically identifying relevant spectral regions. Yet, maps containing relevant information may often be discarded due to low signal-to-noise ratios or interference with other elements. In this context, this work presents an agnostic processing pipeline based on a spatial information ratio metric that is computed in the Fourier space for each wavelength and that allows for the identification of relevant spectral ranges in LIBS. The results suggest a more robust and streamlined approach to feature extraction in LIBS imaging compared with traditional inspection of the spectra, which can introduce novel opportunities not only for spectral data analysis but also in the field of data compression.

Abstract
Optical tweezers are an interesting tool to enable single cell analysis, especially when coupled with optical sensing and advanced computational methods. Nevertheless, such approaches are still hindered by system operation variability, and reduced amount of data, resulting in performance degradation when addressing new data sets. In this manuscript, we describe the deployment of an automatic and intelligent optical tweezers setup, capable of trapping, manipulating, and analyzing the physical properties of individual microscopic particles in an automatic and autonomous manner, at a rate of 4 particle per min, without user intervention. Reproducibility of particle identification with the help of machine learning algorithms is tested both for manual and automatic operation. The forward scattered signal of the trapped PMMA and PS particles was acquired over two days and used to train and test models based on the random forest classifier. With manual operation the system could initially distinguish between PMMA and PS with 90% accuracy. However, when using test datasets acquired on a different day it suffered a loss of accuracy around 24%. On the other hand, the automatic system could classify four types of particles with 79% accuracy maintaining performance (around 1% variation) even when tested with different datasets. Overall, the automated system shows an increased reproducibility and stability of the acquired signals allowing for the confirmation of the proportionality relationship expected between the particle size and its friction coefficient. These results demonstrate that this approach may support the development of future systems with increased throughput and reliability, for biosciences applications.