Machine learning (ML) and deep learning (DL) have achieved great success in different tasks. These include computer vision, image segmentation, natural language processing, predicting classification, evaluating time series, and predicting values based on a series of variables. As artificial intelligence progresses, new techniques are being applied to areas like optical spectroscopy and its uses in specific fields, such as the agrifood industry. The performance of ML and DL techniques generally improves with the amount of data available. However, it is not always possible to obtain all the necessary data for creating a robust dataset. In the particular case of agrifood applications, dataset collection is generally constrained to specific periods. Weather conditions can also reduce the possibility to cover the entire range of classifications with the consequent generation of imbalanced datasets. To address this issue, data augmentation (DA) techniques are employed to expand the dataset by adding slightly modified copies of existing data. This leads to a dataset that includes values from laboratory tests, as well as a collection of synthetic data based on the real data. This review work will present the application of DA techniques to optical spectroscopy datasets obtained from real agrifood industry applications. The reviewed methods will describe the use of simple DA techniques, such as duplicating samples with slight changes, as well as the utilization of more complex algorithms based on deep learning generative adversarial networks (GANs), and semi-supervised generative adversarial networks (SGANs).

A comprehensive review of optical biosensors for the detection of biomarkers associated with rheumatoid arthritis (RA) is presented here, including microRNAs (miRNAs), C-reactive protein (CRP), rheumatoid factor (RF), anti-citrullinated protein antibodies (ACPA), interleukin-6 (IL-6) and histidine, which are biomarkers that enable RA detection and/or monitoring. An overview of the different optical biosensors (based on fluorescence, plasmon resonances, interferometry, surface-enhanced Raman spectroscopy (SERS) among other optical techniques) used to detect these biomarkers is given, describing their performance and main characteristics (limit of detection (LOD) and dynamic range), as well as the connection between the respective biomarker and rheumatoid arthritis. It has been observed that the relationship between the corresponding biomarker and rheumatoid arthritis tends to be obviated most of the time when explaining the mechanism of the optical biosensor, which forces the researcher to look for further information about the biomarker. This review work attempts to establish a clear association between optical sensors and rheumatoid arthritis biomarkers as well as to be an easy-to-use tool for the researchers working in this field.

This work presents the implementation of a thermo-refractometer, which integrates the measurement of both refractive index and temperature in a single optical fiber structure. To this purpose, a lossy mode resonance (LMR)-based refractometer is obtained by means of the deposition of a titanium dioxide (TiO₂) thin film onto a side-polished (D-shaped) single mode fiber. Measurement and subsequent temperature compensation are achieved by means of a fiber Bragg grating (FBG) inscribed in the core of the D-shaped region. The LMR wavelength shift is monitored in transmission while the FBG (FBG peak at 1533 nm) displacement is observed in reflection. The LMR is sensitive to both the surrounding refractive index (SRI), with a sensitivity of 3725.2 nm/RIU in the 1.3324–1.3479 range, and the temperature (- 0.186 nm/°C); while the FBG is only affected by the temperature (32.6 pm/°C in the 25°C – 45°C range). With these values, it is possible to recover the SRI and temperature variations from the wavelength shifts of the LMR and the FBG, constituting a thermo-refractometer, where it is suppressed the effect of the temperature over the refractometer operation, which could cause errors in the fourth or even third decimal of the measured SRI value.

The temperature dependence of core mode resonance has been thoroughly studied in fiber Bragg gratings (FBGs), but it is not the case for cladding mode resonances in multi-resonance gratings such as tilted FBGs (TFBGs). In this work, the temperature sensitivity of ultraviolet written TFBGs in SMF-28 fibers is assessed, demonstrating in the first, to the best of our knowledge, place that a single gauge factor K T = 6.25 ⋅ 10 − 6 ± 0.02 ⋅ 10 − 6 ∘ C − 1 can be employed to characterize the response to temperature of the resonances over the full spectrum in the 10°–50°C range. Then, a simulation model is obtained, enabling to predict TFBG spectra in the 10°–50°C range with high accuracy. This requires a calibration of the core index and dispersion of the TFBG measured in air at 25°C, and determination of the glass refractive index thermo-optic coefficient ( d n / d T = 8.46 ⋅ 10 − 6 ± 0.1 ⋅ 10 − 6 ∘ C − 1 , common to both core and cladding glasses), leading to a mean error on the wavelength position of resonances between 1 and 3 pm. This mean error can be further reduced (less than 1 pm) by considering a linear dependence with temperature of d n / d T . Therefore, this model will enable to completely remove the temperature-induced shifts of all resonances in TFBG sensing applications and measure with great accuracy the variables of interest by using the scaled averages of groups of resonances instead of (less accurate) individual shifts.

This Letter presents the fabrication of dual lossy mode resonance (LMR) refractometers based on titanium dioxide ( T i O 2 ) and tin oxide ( S n O 2 ) thin films deposited on a single side-polished D-shaped optical fiber. For the first time, to the best of our knowledge, two independent LMRs are obtained in the same D-shaped optical fiber, by using a step-shaped nanostructure consisting of a first section of T i O 2 with a thickness of 120 nm and a second section with a thickness of 140 nm (120 nm of T i O 2 and 20 nm of S n O 2 ). Each section is responsible for generating a first-order LMR with TM-polarized light ( L M R T M ). T i O 2 is deposited by atomic layer deposition and S n O 2 by electron-beam deposition. The theoretical results show that the depth of each of the resonances of the dual LMR depends on the length of the corresponding section. Two experimental devices were fabricated with sections of different lengths, and their sensitivities were studied, achieving values ∼ 4000 n m / r e f r a c t i v e i n d e x u n i t (RIU) with a maximum of 4506 nm/RIU for values of the SRI between 1.3327 and 1.3485.

Double-clad fibers where the second cladding has a lower refractive index than the first cladding, prove to be ideal structures for potentiating and tuning the sensitivity in long-period fiber gratings (LPFGs) operating in mode transition. When a thin film is deposited on the optical fiber, the second cladding performs acts as a barrier that initially prevents the transition to guidance in the thin film of one of the modes guided in the first cladding. Finally, the transition to guidance occurs with a sensitivity increase, in analogy to the tunnel effect observed in semiconductors. This improvement has been demonstrated both as a function of the thin film thickness and the surrounding medium refractive index, with enhancement factors of 4 and 2, respectively. This idea reinforces the performance of LPFGs, adding a new degree of freedom to the mode transition and the dispersion turning point phenomena. Moreover, the control of the variation of the effective index of cladding modes could be applied in other structures, such as tilted-fiber gratings or evanescent wave sensors.