Hemoglobinopathies are among the most common inherited diseases worldwide, affecting approximately 7% of the global population. Despite advances in the standardization and harmonization of methods for HbA1c determination, an increasing number of hemoglobinopathies cause false HbA1c results. One of the common techniques for screening hemoglobinopathies is through high-performance liquid chromatography (HPLC) separation, followed by UV–VIS detection. Although UV–VIS can quantify the hemoglobin fractions, it is unable to identify them. In this study, we use Raman spectroscopy to study the fingerprint spectra of hemoglobin fractions based on which the fractions can be identified. To evaluate the potential of Raman spectroscopy in identifying these fractions, we utilize a range of commercially available hemoglobin fractions, including fetal hemoglobin. We automate the classification process with machine learning approaches such as support vector machines (SVM), fully connected neural networks (NN), k-Nearest Neighbors (KNN), Decision Trees (DT), and Bernoulli Naive Bayes (BNB). These models are fine-tuned and optimized to classify the hemoglobin fractions and achieve test accuracies of 98.2% and 98.5%, respectively. Our research highlights the potential of Raman spectroscopy as an identification tool when combined with HPLC.
Machine learning techniques have been proposed in the literature for the modeling of photonic devices. These techniques can be used to speed up the design process. The data samples needed to build machine learning models are collected from electromagnetic simulations. Electromagnetic solvers can result computationally expensive and therefore minimizing the computational effort needed to collect these data samples is an important aspect. Using frequency-domain electromagnetic solvers to collect data samples requires a suitable sampling of the wavelength variable to avoid undersampling and oversampling phenomena. An adaptive frequency-domain sampling approach for nanophotonic applications is illustrated in this work.
We present the development and evaluation of metalenses fabricated with the two-photon polymerization-based 3D nanoprinting technology. In our design, we investigated a periodic lattice of multilevel nanopillars, based on the natural ellipsoidal shape of the 3D voxel in the fabrication process. By creating nanopillars with various heights, we can tune the effective refractive index of the metasurface in order to modulate the phase profile of an incoming light beam. We therefore push the fast and flexible two-photon polymerization technique to its limits in terms of dimensions in view of creating high performance metalenses. To demonstrate the optical performance of these metalenses, we also created their refractive and diffractive counterparts with the same fabrication technology to allow for a direct performance comparison. Moreover, we show that these metalenses can be fabricated on the tip of standard telecom single-mode optical fibers for the effective collimation of their output light beam.
Fiber-optic microendoscopy imaging systems are actively being researched. They are interesting as they can be designed with a single fiber in conjunction with gradient refractive index (GRIN) micro-lenses. However, these systems face a significant limitation in the form of optical aberrations caused by beam scanning, resulting in reduced resolution at the edges of the imaging field. The current solutions involve bulky refractive optics, combining micro-optics with an aspherical lens that present challenges in fabrication and alignment. To overcome these limitations, we propose the design of a compact metasurface correction element (diameter = 1 mm) that can be seamlessly integrated into the existing optical system alongside the 1 mm diameter GRIN lens. Accurately modeling such a complex system involving nanoscale metasurface and macroscale optics is challenging. We present the interconnection of ray tracing and electromagnetic simulations to simultaneously achieve the desired optical system performance and the required phase profile. In our imaging probe design, the target phase profile of the correction element is optimized using Zemax Optic Studio for multiple beam scan angles to achieve a minimal and uniform spot of 1 μm across the imaging field, which spans approximately 100 μm. The target phase mask obtained from ray tracing simulation and the phase-look-up table obtained from electromagnetic simulation of the unit cell are used to create the metasurface. The simulation of the metasurface of 1mm diameter is performed in Lumerical and the solved near field is propagated in Zemax to assess the imaging system through physical optics propagation and diffraction analysis.
Hemoglobinopathies are the most common genetic disorders caused by a mutation in the genes encoding for one of the globin chains and leading to structural (hemoglobin [Hb] variants) or quantitative defects (thalassemias) in hemoglobin. Early diagnosis and characterization of hemoglobinopathies are essential to avoid severe hematological consequences in the offspring of healthy carriers of a mutation. Despite being extensively studied, hemoglobinopathies continue to provide a diagnostic challenge. Sickle-cell hemoglobin (HbS) is the most common and clinically significant hemoglobin variant among all Hb variants. To overcome the challenge of diagnosing Hb variants, we propose the use of Surface-Enhanced Raman Spectroscopy (SERS). SERS is a powerful label-free tool for providing fingerprint structural information of analyses. It can rapidly generate the spectral signature of samples. This study investigates the structural differences between HbS and normal Hb using gold nanopillar SERS substrates with a leaning effect. The SERS spectra of Hb variants showed subtle spectral differences between HbS and normal Hb located in the valine (975 cm-1) and glutamic acid (1547 cm-1) band, reflecting the amino acid substitution in the HbS β-globin chain. We also automated the identification of HbS and normal Hb with principal component analysis (PCA) combined with support vector machine (SVM) and linear discriminant analysis (LDA) classifiers, leading to an accuracy of 98% and 96%, respectively. This study demonstrated that SERS can provide a fast, highly sensitive, noninvasive, and accurate detection module for the diagnosis of Sickle-cell disease and potentially other hemoglobinopathies.
Machine learning techniques have been proposed in the literature for the modeling of photonic devices. In this paper, a modeling technique based on system identification, feature extraction, and machine learning methods is proposed for the design of photonic devices. Design features of interest are extracted based on a system identification step that uses a few samples of the electromagnetic device response. This system identification step allows saving computational resources significantly while collecting the data needed for the further machine learning step. Modeling design features instead of the wavelength-dependent device response as a function of the design parameters allows compacting the output space of interest in neural networks and reducing related model complexity issues. These features can be modeled as a function of design parameters by means of neural networks. The generated neural networks are of very limited complexity. Design features represent very valuable and meaningful information for designers. Numerical results successfully validate the proposed technique.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.