Machine learning methods have found applications in areas such as security, molecular biology, medicine, computational physics and mechanics, etc. In this paper a review on using machine learning technologies for LIDAR measurements is present based on Google Scholar, Scopus, and Web of Science citing databases. Search includes keywords “lidar”, “atmospheric sensing”, “machine learning” through past 5 years. Most relevant and significant papers were selected.
The main greenhouse gases are ozone and the gas components of ozone cycles. Operational determination of ozone concentration profiles is carried out by lidar methods, which limits the number of measurements obtained. Machine learning methods can be used to build predictive models of the data as well as to approximate them. This paper investigates the possibility of generating data to build robust predictive models of ozone concentration profiles based on generative adversarial neural networks (GAN). Several GAN architectures were proposed and the benefits of each one is discussed.
This article describes the methods and approaches used by us to solve the problem of a high error in the determination of a component with a low concentration in a gas mixture. The approaches based on the modification of the machine learning model were considered, the approach to the generation of the training sample was changed, an iterative method for increasing the accuracy of the model results was proposed.
The paper presents an algorithm based on low order statistics for the informative feature extraction for Raman spectroscopy data. The proposed method was tested on mouse preimplantation embryos Raman spectra. Both supervised and unsupervised machine learning methods were applied to selected the most informative features to test the separability of the processed data.
Raman spectra of blood plasma were studied in the dynamics of the experimental glioma. We used a DXR Raman Microscope (Thermo Scientific), excitation wavelengths of 532 nm, range 80–3200 cm–1. Each sample of blood plasma was a droplet with a volume of 10 μL placed on a special aluminum plate. Machine learning methods were used to identify the most informative frequencies associated with cancer molecular markers. The most significant changes in the Raman spectra are observed in the 900–1700 cm–1 range.
Two-photon microscopy methods have been actively developing in the last two decades. In particular, various approaches are being developed to analyze metabolic activity obtained by laser microscopy with high temporal resolution. One such approach is ploting the autofluorescence lifetime signal data to the phasor plot and observe deviations from the normal state. This tool has proven itself well for analyzing the metabolic activity of biological objects, including bacteria. This study analyzes the antimicrobial activity for MRSA bacteria by phasor plot approach. It was possible to show that, the lifetime of autofluorescence in the phasor plane changes depending on the type of nanoparticles used, with which bacteria are incubated. This result can be used in the future for rapid assessment of the antimicrobial activity of nanoparticles.
This book focuses on machine-learning medical applications based on molecular spectroscopy and molecular imaging data. Written with specialists in biomedical optics, laser spectroscopy, bioengineering, and medical engineering in mind, the chapters cover topics such as biomarker conception, molecular laser imaging, and artificial intelligence; laser-based molecular-data-acquisition technologies; feature selection and extraction methods; unsupervised and supervised approaches; and in vivo non-invasive diagnostics using laser molecular spectroscopy and imaging combined with machine learning. Sample datasets and Python modules are provided as supplemental materials for the most useful algorithms.
Significance: The creation of fundamentally new approaches to storing various biomaterial and estimation parameters, without irreversible loss of any biomaterial, is a pressing challenge in clinical practice. We present a technology for studying samples of diabetic and non-diabetic human blood plasma in the terahertz (THz) frequency range.
Aim: The main idea of our study is to propose a method for diagnosis and storing the samples of diabetic and non-diabetic human blood plasma and to study these samples in the THz frequency range.
Approach: Venous blood from patients with type 2 diabetes mellitus and conditionally healthy participants was collected. To limit the impact of water in the THz spectra, lyophilization of liquid samples and their pressing into a pellet were performed. These pellets were analyzed using THz time-domain spectroscopy. The differentiation between the THz spectral data was conducted using multivariate statistics to classify non-diabetic and diabetic groups’ spectra.
Results: We present the density-normalized absorption and refractive index for diabetic and non-diabetic pellets in the range 0.2 to 1.4 THz. Over the entire THz frequency range, the normalized index of refraction of diabetes pellets exceeds this indicator of non-diabetic pellet on average by 9% to 12%. The non-diabetic and diabetic groups of the THz spectra are spatially separated in the principal component space.
Conclusion: We illustrate the potential ability in clinical medicine to construct a predictive rule by supervised learning algorithms after collecting enough experimental data.
The paper presents an analysis of the Raman spectra of mouse preimplantation embryos using machine learning for visualization, assessing the separability of classes, and highlighting informative areas of the spectrum. Separation of lipid reach areas and nucleus spectra was shown by principal component analysis coupled with a linear support vector machine.
An important role in component analysis with spectral methods has a spectral resolution of used tools. The most useful and perspective methods to improve spectral resolution is decreasing of impulse response function (IRF) and improving resolution using superresolution (SR) reconstruction methods. We have analyzed different types of neural networks (convolution neural network, multilayered perceptron) for improving the spectral resolution of initial absorption spectra. The used approach is based on an association of a high-resolution and a low-resolution spectrum. The latter was constructed from high-resolution spectra to which IRF and some random noise were added. Highresolution spectra were generated using the HITRAN database. Most optimal architectures of neural networks to improve spectral resolution were defined.
The regression model was applied to solve the problem of restoration of the concentration of components in a gas mixture using infrared absorption spectra. A solution to the problem of determining the concentrations of individual components of a multicomponent gas mixture is suggested and tested.
Based on time-independent Helmholtz equation and its solution in frame of inhomogeneous approximation a hybrid computational method for imitation of propagation of bounded laser beam focused into biological tissue is introduced. The biological tissue is simulated as a semi-infinite randomly inhomogeneous medium. The developed approach is intended to model laser beams in the super-sharp focusing mode. The results of modeling of laser light focusing into the turbid tissue-like scattering medium with lenses of various shapes are presented.
We present a computational modeling approach for imitation of the time-domain optical coherence tomography (OCT) images of biotissues. The developed modeling technique is based on the implementation of the Leontovich–Fock equation into the wave Monte Carlo (MC) method. We discuss the benefits of the developed computational model in comparison to the conventional MC method based on the modeling of OCT images of a nevus. The developed model takes into account diffraction on bulk-absorbing microstructures and allows consideration of the influence of the amplitude–phase profile of the wave beam on the quality of the OCT images. The selection of optical parameters of modeling medium, used for simulation of optical radiation propagation in biotissues, is based on the results obtained experimentally by OCT. The developed computational model can be used for imitation of the light waves propagation both in time-domain and spectral-domain OCT approaches.
We applied the method of statistical trials to the parabolic equation of laser radiation propagation in biotissue to perform a new method of optical coherence tomography modeling. Results of modeling tests show the efficiency of the developed approach.
The method of laser IR radiation propagation simulation in a case of randomly inhomogeneous media based on Leontovich – Fock equation in the application of optical coherent tomography modeling in biotissues is proposed. We describe the proposed methodology and demonstrate its implementation on a test case.
Methods of two-photon microscopy are widely used in the study of biological objects, in particular, skin, due to the possibility to study objects both on the surface and at depth without attracting additional fluorophores due to endogenous autofluorescence. In this paper, the methods of image analysis of the AF signal and SHG signal are applied to assess the condition of the skin during the development of lymphedema. It is shown that for groups of healthy tissue and lymphedematous using SAAID distribution histograms, changes in tissues can be detected.
Last years the development of computer-aided diagnostic systems for medical image analysis has become a hot topic. A key step is connected with informative features extraction. Here, we discussed multiphoton microscopy and optical coherent tomography lymphedema tissue images analysis using gradient processing methods.
The problem of extracting useful information for medical diagnosis from 2D and 3D optical imaging experimental data is of great importance. We are discussing challenges and perspectives of medical diagnosis using machine learning analysis of NIR and THz tissue imaging. The peculiarities of tissue optical clearing for tissue imaging in NIR and THz spectral ranges aiming the improvement of content data analysis, methods of extracting of informative features from experimental data and creating of prognostic models for medical diagnosis using machine learning methods are discussed.
The kernel construction for the biomedical data classification using Support Vector Machine based on Green function for Ornstein-Uhlenbeck equation is presented. Quantitative estimates of classification quality of exhaled air samples absorption spectra for patients with chronic obstructive pulmonary disease and healthy volunteers were carried out.
We examined possibilities of the Kalman filter for reducing the noise effects in the analysis of absorption spectra of gas samples, in particular, for samples of the exhaled air. It has been shown that when comparing groups of patients with broncho-pulmonary diseases on the basis of the absorption spectra analysis of exhaled air samples the data preprocessing with the Kalman filtering can improve the classification sensitivity using a support vector kernel with mpl.
In this work results of classification of patients with broncho-pulmonary diseases based on analysis of exhaled air samples are presented. These results obtained by application of laser photoacoustic spectroscopy method and intellectual data analysis ones (Principal Component Analysis, Support vector machines, neural networks). Absorption spectra of exhaled air of gathered volunteers were registered; data preparation for classification procedure of absorption spectra of exhaled air of healthy and sick people was made. Also error matrices for neural networks and sensitivity/specificity values in case of classification with SVM method were obtained. This work was partially supposed by the Federal Target Program for Research and Development, Contract No. 14.578.21.0082 (unique identifier of applied scientific research and experimental development RFMEFI57814X0082).
The results of numerical simulation of application principal component analysis to absorption spectra of breath air of patients with pulmonary diseases are presented. Various methods of experimental data preprocessing are analyzed.
The results of comparison of quality of two classificators – SVM (support vector machine) and SIMCA (soft independent modelling of class analogies) on model data contained profiles of absorbtion specra of exhalted air are presented. It is shown, that SVM classification results can be improved by preprocessing if input data with principal component analysis method.
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.