Faculty Profiles - SAKAI Osamu

Papers - SAKAI Osamu

Division display1 - 20 of about 104／ All the affair displays >>

Skin Diagnostic Method Using Fontana-Masson Stained Images of Stratum Corneum Cells

Shuto Hasegawa, Koichiro Enomoto, Taeko Mizutani, Yuri Okano, Takenori Tanaka, Osamu Sakai

IEICE Transactions on Information and Systems E107.D ( 8 ) 1070 - 1078 2024.8

Publishing type：Research paper (scientific journal)

Melanin, which is responsible for the appearance of spots and freckles, is an important indicator in evaluating skin condition. To assess the efficacy of cosmetics, skin condition scoring is performed by analyzing the distribution and amount of melanin from microscopic images of the stratum corneum cells. However, the current practice of diagnosing skin condition using stratum corneum cells images relies heavily on visual evaluation by experts. The goal of this study is to develop a quantitative eval- uation system for skin condition based on melanin within unstained stratum corneum cells images. The proposed system utilizes principal component regression to perform five-level scoring, which is then compared with vi- sual evaluation scores to assess the system's usefulness. Additionally, we evaluated the impact of indicators related to melanin obtained from images on the scores, and verified which indicators are effective for evaluation. In conclusion, we confirmed that scoring is possible with an accuracy of more than 60% on a combination of several indicators, which is comparable to the accuracy of visual assessment.

DOI： 10.1587/transinf.2023EDP7256

Maze-solving in a plasma system based on functional analogies to reinforcementlearning model

Osamu Sakai, Toshifusa Karasaki, Tsuyohito Ito, Tomoyuki Murakami, Manabu Tanaka, Makoto Kambara, Satoshi Hirayama

PLoS ONE 19 ( 4 April ) 2024.4

Publishing type：Research paper (scientific journal)

Maze-solving is a classical mathematical task, and is recently analogously achieved using various eccentric media and devices, such as living tissues, chemotaxis, and memristors. Plasma generated in a labyrinth of narrow channels can also play a role as a route finder to the exit. In this study, we experimentally observe the function of maze-route findings in a plasma system based on a mixed discharge scheme of direct-current (DC) volume mode and alternative-current (AC) surface dielectric-barrier discharge, and computationally generalize this function in a reinforcement-learning model. In our plasma system, we install two electrodes at the entry and the exit in a square lattice configuration of narrow channels whose cross section is 1×1 mm2 with the total length around ten centimeters. Visible emissions in low-pressure Ar gas are observed after plasma ignition, and the plasma starting from a given entry location reaches the exit as the discharge voltage increases, whose route converging level is quantified by Shannon entropy. A similar short-path route is reproduced in a reinforcement- learning model in which electric potentials through the discharge voltage is replaced by rewards with positive and negative sign or polarity. The model is not rigorous numerical representation of plasma simulation, but it shares common points with the experiments along with a rough sketch of underlying processes (charges in experiments and rewards in modelling). This finding indicates that a plasma-channel network works in an analog computing function similar to a reinforcement-learning algorithm slightly modified in this study.

DOI： 10.1371/journal.pone.0300842

MACHINE LEARNING EVALUATION OF FRUIT RIPENESS WITH MULTIDIMENSIONAL SPARSE-DATASET CALIBRATION

Shinya Ueno, Osamu Sakai, Takuya Morimoto

Applied Engineering in Agriculture 40 ( 5 ) 501 - 514 2024

Publishing type：Research paper (scientific journal)

. In agriculture, harvests of fruits and vegetables depend on their ripeness, in which farmers should rely not only on the eye-sensed outlook but rigorously on constituents inside the targets, like sugar content. So far, detail measurements on the outlook color of fruits have been widely reported to detect suitable harvest times, where non-destructive instruments, which are costly in general agriculture fields, work well for color estimation whose resolution level is much higher than human eyes. In this study, we propose a scheme with a calibration procedure for sugar content prediction from outlook-color quantities, which is applicable even if a sensor includes systematic errors in output signals or the base dataset includes some datapoint sparsity. Most multidimensional data, which are both in agriculture and in other general cases, are imperfect in terms of several aspects, but they can be useful when we take appropriate sparsity detection and setting of rebalancing weights into account for machine learning procedures. We newly introduce a complex-network method for this purpose, and after the color calibration, we successfully obtain accurate hue and chroma, which are quantities representing a position on calibrated color coordinate, to predict °Brix sugar content in Japanese pears.

DOI： 10.13031/aea.15899

Formation of a Dirac Cone and Dynamic Control of its Half-Metallic Properties Using Double-Layer Ring Dipole Arrays Reviewed

Itami G., Sakai O.

Physical Review Applied 20 ( 3 ) 2023.9

Detection of Sparsity in Multidimensional Data Using Network Degree Distribution and Improved Supervised Learning with Correction of Data Weighting Reviewed

Ueno S., Sakai O.

Studies in Computational Intelligence 1077 SCI 390 - 401 2023.1

Language：English Publishing type：Research paper (scientific journal) Publisher：Studies in Computational Intelligence

Multidimensional data are representatives in a wide range of applications, from those in the latest state-of-the-art science and technology to specific social issues. And they have been subject to analysis using methods such as regression analysis and machine learning. However, they are rarely obtained as complete data and contain more or less biases and deficiencies. In this study, we form a network from a multidimensional dataset and use its degree distribution to detect data sparsity. Although model analysis based on the degree distribution has been conducted for many years, sparsity detection has not been a target of the degree distribution analysis. Furthermore, we attempt to increase the accuracy and precision of supervised learning by applying regressive weighting according to node grouping in the degree distribution spectrum. By making use of this algorithm, we can expand the range of utilization of incomplete data together with other promising progresses in complex networks.

DOI： 10.1007/978-3-031-21127-0_32

Other Link： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85148704019&origin=inward

Independent control method for plasmonic skin depth based on transformation from spoof surface plasmon polaritons to bound states in the continuum Reviewed

Itami G., Sakai O.

Physical Review B 106 ( 24 ) 2022.12

Language：English Publishing type：Research paper (scientific journal) Publisher：Physical Review B

A metal plate array (MPA) is the complementary structure of a metal hole array and forms spoof surface plasmon polaritons (SSPPs) supported by parallel plate modes in a unit cell. A ring dipole array, which is a periodic structure designed by providing an aperture in the conductive part of the unit cell of an MPA, has the potential to produce bound states in the continuum (BICs). In this paper, the transformation from SSPPs to BICs is confirmed using these two structures by broadening the aperture in the unit cell. Making use of this phenomenon, control of the skin depth of surface modes is numerically demonstrated while retaining the resonant frequency, and the fitting values of the plasmonic skin depth are in good agreement with theoretical values.

DOI： 10.1103/PhysRevB.106.245406

Other Link： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85143690192&origin=inward

Foundations of plasma photonics: lamps, lasers, and electromagnetic devices Reviewed International coauthorship

Nijdam S., Desai K.V., Park S.J., Sun P.P., Sakai O., Lister G., Eden J.G.

Plasma Sources Science and Technology 31 ( 12 ) 2022.12

Language：English Publishing type：Research paper (scientific journal) Publisher：Plasma Sources Science and Technology

The enduring contributions of low temperature plasmas to both technology and science are largely a result of the atomic, molecular, and electromagnetic (EM) products they generate efficiently such as electrons, ions, excited species, and photons. Among these, the production of light has arguably had the greatest commercial impact for more than a century, and plasma sources emitting photons over the portion of the EM spectrum extending from the microwave to soft x-ray regions are currently the workhorses of general lighting (outdoor and indoor), photolithography for micro- and nano-fabrication of electronic devices, disinfection, frequency standards (atomic clocks), lasers, and a host of other photonic applications. In several regions of the EM spectrum, plasma sources have no peer, and this article is devoted to an overview of the physics of several selected plasma light sources, with emphasis on thermal arc and fluorescent lamps and the more recently-developed microcavity plasma lamps in the visible and ultraviolet/vacuum ultraviolet regions. We also briefly review the physics of plasma-based metamaterials and plasma photonic crystals in which low temperature plasma tunes the EM properties of filters, resonators, mirrors, and other components in the microwave, mm, and sub-mm wavelength regions.

DOI： 10.1088/1361-6595/ac8448

Other Link： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85146427239&origin=inward

Complexity visualization, dataset acquisition, and machine-learning perspectives for low-temperature plasma: A review Reviewed

Sakai O., Kawaguchi S., Murakami T.

Japanese Journal of Applied Physics 61 ( 7 ) 2022.7

Language：English Publishing type：Research paper (scientific journal) Publisher：Japanese Journal of Applied Physics

Low-temperature plasma plays various roles in industrial material processing as well as provides a number of scientific targets, both from theoretical and experimental points of view. Such rich features in variety are based on its complexities, arising from diverse parameters in constituent gas-phase species, working gas pressure, input energy density, and spatial boundaries. When we consider causalities in these complexities, direct application of machine-learning methods is not always possible since levels of complexities are so high in comparison with other scientific research targets. To overcome this difficulty, progresses in plasma diagnostics and data acquisition systems are inevitable, and the handling of a large number of data elements is one of the key issues for this purpose. In this topical review, we summarize previous and current achievements of visualization, acquisition, and analysis methods for complex plasma datasets which may open a scientific and technological category mixed with rapid machine-learning advancements and their relevant outcomes. Although these research trends are ongoing, many reports published so far have already convinced us of various expanding aspects of low-temperature plasma leading to the potential for scientific progress as well as developments of intellectual design in industrial plasma processes.

DOI： 10.35848/1347-4065/ac76fa

Other Link： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85134431891&origin=inward

Data Driven Calibration of Color-Sensitive Optical Sensor by Supervised Learning for Botanical Application

Shinya Ueno, Osamu Sakai

IEEE Sensors Journal 22 ( 12 ) 11915 - 11927 2022.6

Publishing type：Research paper (scientific journal)

Color is a fundamental and informative element in botanical applications, as it is in our various daily activities. The conventional ways of evaluating colors based on human visibility are person-specific and thus likely to be biased, whereas color sensing equipment with high sensitivity and accuracy is often a very expensive product. Recently, nonlinear data conversions using machine learning, such as supervised learning, have become available in various fields, and such techniques may be useful for calibrations of low-cost, poor-accuracy sensors. In this study, we propose a method for color-value calibration applicable to low-cost sensors by supervised learning. Specifically, after a supervised learning procedure using training data based on a large number of color samples positioned on a two-dimensional color plane, we improve calibration accuracy over mixed-color areas containing red, green and blue values. We apply this calibrated sensor to estimating the ripeness level of Japanese pears, and the results indicate that the chroma shown in a color coordinate is a good measure for this purpose. In another botanical case, temporal changes in color values of autumn tree leaves in chroma and hue could be successfully detected. This technology is aimed at use with low-cost consumer electronic devices, and it is applicable to color sensing in other areas beyond in the botanical field, thus showing high potential for other kinds of low-cost sensors.

DOI： 10.1109/JSEN.2022.3171221

Synthesis of Electromagnetic Wave-Absorbing Co-Ni Alloys and Co-Ni Core-Shell Structured Nanoparticles Reviewed

Ishijima M., Takada T., Cuya Huaman J.L., Mizutomi T., Sakai O., Shinoda K., Uchikoshi M., Mamiya H., Suzuki K., Miyamura H., Balachandran J.

Inorganic Chemistry 2022.1

Language：English Publishing type：Research paper (scientific journal) Publisher：Inorganic Chemistry

Co-Ni alloy nanoparticles, a potential candidate for microwave absorption material, were successfully synthesized by tuning the reduction timing of Co and Ni ions by introducing oleylamine as a complexing agent and 1-heptanol as a reducing solvent. The formation mechanism elucidated using time-resolved sampling and in situ X-ray absorption spectroscopy (XAS) and ultraviolet-visible (UV-vis) spectrophotometry measurements suggested that the delay in the reduction of Co ions via complexation with oleylamine facilitated the co-reduction of Co with Ni ions and led to the formation of Co-Ni alloys. The successful synthesis of Co-Ni alloys experimentally confirmed the differences in magnetic properties between alloy and core-shell structured Co50Ni50 particles. Further, the syntheses of Co-Ni alloys with different compositions were also possible using the above technique. In addition, the microwave absorption properties were measured using the free-space method utilizing a vector network analyzer of Co50Ni50-polyethylene composite with different sheet thicknesses. A reflection loss (RL) value of -25.7 dB at 13.6 GHz for the alloy structure was more significant than the core-shell counterpart. The above values are high compared to results reported in the past. The validity of the measurements was confirmed by utilizing the parameter retrieval method to extract permittivity and permeability from the scattering parameter (S) and recalculation of the RL as a function of frequency.

DOI： 10.1021/acs.inorgchem.2c02633

Other Link： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85140324092&origin=inward

Logistics Route Planning in Agent-Based Simulation and Its Optimization Represented in Higher-Order Markov-Chain Networks Reviewed

Ikai R., Miyagi S., Sakai O.

Springer Proceedings in Complexity 38 - 50 2021.7

Language：English Publishing type：Research paper (scientific journal) Publisher：Springer Proceedings in Complexity

Route planning in logistics, in which multiple pickup and delivery positions exist in a road network, is a complicated task with many choices in a path selection and their influences on the following procedures. Solving this task by multi-agent simulations, we examine the route optimization process by monitoring motions in networks based on simple or higher-order Markov chains (MCs). Agent footprints in the networks, which spread over the entire network at the initial phase, converge on small number of edges as the transportation path gets shortened. When we increase the order of MCs in agent mobilities, the MC networks are enlarged and possess a large number of nodes and edges with structural regularity so that one node contains partial trace history, while the optimized route that frequently overlaps edge groups with high transition probabilities is equivalent to a smaller and more noticeable subgraph around a local optimal solution. In other words, this localization of the traces indicates a convergence level in optimization, which can be a measure for route planning in logistics.

DOI： 10.1007/978-3-030-81854-8_4

Other Link： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85113302052&origin=inward

In-vacuum active colour sensor and wireless communication across a vacuum-air interface Reviewed

Sakai O., Kitagawa T., Sakurai K., Itami G., Miyagi S., Noborio K., Taguchi K.

Scientific Reports 11 ( 1 ) 2021.1

Language：English Publishing type：Research paper (scientific journal) Publisher：Scientific Reports

© 2021, The Author(s). In situ sensing with wireless digital-data transfer is a potential processing scheme that works very closely to the location of an event monitored by a sensor and converts the sensor’s raw output into digitized and informative small-volume bits, as suggested by recent proposals for edge computing and the Internet of Things (IoT). Colour perception may be a target of in situ sensor data acquisition; however, in contrast to from other sensing devices, colour sensors that detect visible light signals are usually located away from light-emitting sources, collecting light transmitting through the space and attenuating it in some manner. For example, in a vacuum chamber whose gas pressure is much less than the ambient atmosphere in which the sensors usually work, there are many veiled light sources, such as discharge plasma, for various industrial purposes including nanoscale manufacturing. In this study, we designed an in-vacuum colour sensor that can work with analogue-to-digital conversion and transfer data by wireless communication; this sensor is active in a low-pressure plasma chamber, detecting light signals and transferring them to a personal computer located outside the vacuum chamber. In addition to detecting lights with controlled spectra from outside successfully, we achieved complete operation of our in-vacuum active sensor for plasma emissions generated at 100 Pa. Comparing the signals with data from simultaneous monitoring by a monochromator, we established that the recorded signals arose from the plasma, confirming successful direct detection of low-pressure plasma emissions without any filtering effects between the sensor and the target object.

DOI： 10.1038/s41598-020-80501-z

Other Link： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85099396381&origin=inward

Order Estimation of Markov-Chain Processes in Complex Mobility Network Embedded in Vehicle Traces Reviewed

Yamamoto K., Miyagi S., Sakai O.

Studies in Computational Intelligence 944 231 - 242 2021.1

Language：English Publishing type：Research paper (scientific journal) Publisher：Studies in Computational Intelligence

Vehicle mobility in urban traffic systems is complex, partly because it reflects mobility of a human who drives a vehicle, and partly because it depends on many roles which the vehicle plays. Previous studies on human mobility revealed that it includes Lévy-flights-like motions and memoryless deterministic walks as well as random walks, but the mobility of vehicles may be more biased due to their functions. Focusing our research target on a sightseeing vehicle with sufficiently limited functions, we show a method to measure regularity of visitation patterns, quantified by order(s) of Markov chains in their mobility. Graphs of higher-order Markov chains, which are representatives of mobility in a network style, possess statistical properties; in our observation dataset, they include degree distributions similar to scale-free networks. The detection of mobility in real social experiments, which is also assumed on these graphs, yields the order of Markov chains inside it with its comparison with the results of agent-based simulations. Centrality indices of the mobility networks well coincide with prediction of these analytical and numerical results.

DOI： 10.1007/978-3-030-65351-4_19

Other Link： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85101814485&origin=inward

Statistics of Growing Chemical Network Originating from One Molecule Species and Activated by Low-Temperature Plasma Reviewed

Mizui Y., Miyagi S., Sakai O.

Studies in Computational Intelligence 944 398 - 409 2021.1

Language：English Publishing type：Research paper (scientific journal) Publisher：Studies in Computational Intelligence

Chemistry in plasma is complicated because it has so many reactions in parallel and in series. A complex network is suitable for the visualization and the analysis of its complexity. A numerical calculation based on hundreds of rate equations is a typical tool for plasma chemistry, but such a computational process does not clarify the undergoing physical and chemical properties that stabilize many industrial plasma processes for a number of applications. In this study, we focus on low-temperature plasma in which high-energy electrons are activators for chemical reactions, and investigate the origin of the stability by examining the statistical properties of networks for silane (SiH ) plasma. There is only one seed species in the initial space, SiH , which is surrounded by high-energy electrons. SiH is decomposed into several fragments composed of Si and/or H atoms with possible charges, and such radical and ion species are decomposed or synthesized into other species, leading to the formation of temporal reaction networks in chemistry. With the effects of rate constants that determine chemical reaction rates, we create temporal networks and observe preferential attachments that induce a new reaction in a transient state. The centrality indices for participant species and degree distributions reveal what is occurring in this complex system, and during the sequential process we observe an exponential-tail degree distribution, which is a significant source of reaction stability. 4 4 4

DOI： 10.1007/978-3-030-65351-4_32

Other Link： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85101815755&origin=inward

Low-Cost Color-Sensitive Optical Sensor Calibrated by Sparse Training Data

Shinya Ueno, Osamu Sakai

2021 IEEE 10th Global Conference on Consumer Electronics, GCCE 2021 371 - 372 2021

Publishing type：Research paper (international conference proceedings)

Color is one of the important practical indices as an aid in a typical daily life. To detect colors qualitatively, inexpensive sensors are effective for promoting applicable fields, which include tests of products in factories and leveling of agriculture harvest. In our previous study, we have achieved low-cost but accurate sensors by calibrated sensors using a simple supervised learning, although, its training data has relatively sparse parameter spaces as compared with other areas, which is a factor of lowering the accuracy after calibration. In this study, when there are sparse areas in the two-dimensional training data, we show that the sparse areas can be reinforced accurately by weighting the data according to its density profile. We confirmed that this method is effective for supervised learning in calibration of color sensors.

DOI： 10.1109/GCCE53005.2021.9621774

Analysis of Visualized Complex Reaction Network in Low-Temperature Molecular Plasma

Osamu Sakai, Yasutaka Mizui, Kyosuke Nobuto, Shigeyuki Miyagi

IEEE International Symposium on Semiconductor Manufacturing Conference Proceedings 2020-December 2020.12

Publishing type：Research paper (international conference proceedings)

It is quite frequent that a factorial fabrication process includes very complex systems, leading to difficulties in its regulation, estimation and prediction. To overcome these difficulties, causality in the complex systems is a key issue, which has not been frequently stressed but is of importance for effective performance of machine learning. One of the examples with such complexity is plasma and its chemistry, where they are in processes of dry etching and plasma chemical vapor deposition. So far, we successfully visualized the complexities using graphs or networks, where nodes represent elements and edges imply interactions between them. In this study, focusing on silane (SiH4) and methane (CH4) low-Temperature molecular plasma chemistry, we clarify roles of species in the chemical reaction network, like reactants, intermediates and products, where a species is a node in this species network and a reactant-product pair is an edge. This distinction is straightforward for selection of reactants as input and products as output variables. We also show and discuss another network, reaction network, in which a reactant-product pair is an edge and its size is so huge that its network statistics is categorized by complex network science. By visualizing and analyzing a complex chemical reaction network in molecular plasma, we obtain useful information for parameter regulation in real processes and also identification of input/output variables for machine learning of a given process.

DOI： 10.1109/ISSM51728.2020.9377518

Analysis and observation of the breakdown of Babinet’s principle in complementary spoof surface plasmon polariton structures Reviewed

Itami G., Sakai O.

Scientific Reports 10 ( 1 ) 2020.12

Language：English Publishing type：Research paper (scientific journal) Publisher：Scientific Reports

© 2020, The Author(s). A metal plate array (MPA) which is a structure complimentary to a metal hole array (MHA), supports spoof surface plasmon polaritons (SSPP) as well as an MHA does. Babinet’s principle attributes the phenomenon of duality to transmission characteristics of the complimentary impedance surfaces because of the symmetry of the behaviors of electric and magnetic fields. However, it is also a fact that the complimentary structures do not follow this principle if they have wavelength-size thickness, because electromagnetic waves do not treat such thick structures as a boundary surface but as propagation spaces with the specific boundaries such as a waveguide which shows SSPP modes. If the thickness is so small that it is negligible, Babinet’s principle is still valid, while it has been uncertain how the layer thickness works to break the principle as it is increased. The unconfirmed transformation is revealed analytically and experimentally with the use of MPAs and MHAs of varying thicknesses.

DOI： 10.1038/s41598-020-67923-5

Other Link： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85087404276&origin=inward

Analysis and observation of the breakdown of Babinet’s principle in complementary spoof surface plasmon polariton structures

Go Itami, Osamu Sakai

Scientific Reports 10 ( 1 ) 2020.12

Publishing type：Research paper (scientific journal)

A metal plate array (MPA) which is a structure complimentary to a metal hole array (MHA), supports spoof surface plasmon polaritons (SSPP) as well as an MHA does. Babinet’s principle attributes the phenomenon of duality to transmission characteristics of the complimentary impedance surfaces because of the symmetry of the behaviors of electric and magnetic fields. However, it is also a fact that the complimentary structures do not follow this principle if they have wavelength-size thickness, because electromagnetic waves do not treat such thick structures as a boundary surface but as propagation spaces with the specific boundaries such as a waveguide which shows SSPP modes. If the thickness is so small that it is negligible, Babinet’s principle is still valid, while it has been uncertain how the layer thickness works to break the principle as it is increased. The unconfirmed transformation is revealed analytically and experimentally with the use of MPAs and MHAs of varying thicknesses.

DOI： 10.1038/s41598-020-67923-5

Rescaling the complex network of low-temperature plasma chemistry through graph-theoretical analysis Reviewed

Murakami T., Sakai O.

Plasma Sources Science and Technology 29 ( 11 ) 2020.11

Language：English Publishing type：Research paper (scientific journal) Publisher：Plasma Sources Science and Technology

© 2020 IOP Publishing Ltd. We propose graph-theoretical analysis for extracting inherent information from complex plasma chemistry and devise a systematic way to rescale the network under the following key criteria: (1) maintain the scale-freeness and self-similarity in the network topology and (2) select the primary species considering its topological centrality. Network analysis of reaction sets clarifies that the scale-freeness emerging from a weak preferential mechanism reflects the uniqueness of plasma-induced chemistry. The effect of chemistry rescaling on the dynamics and chemistry of the He + O2 plasma is quantified through numerical simulations. The present chemical compression dramatically reduces the computational load, whereas the concentration profiles of reactive oxygen species (ROS) remain largely unchanged across a broad range of time, space and oxygen admixture fraction. The proposed analytical approach enables us to exploit the full potential of expansive chemical reaction data and would serve as a guideline for creating chemical reaction models.

DOI： 10.1088/1361-6595/abbdca

Other Link： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85097937713&origin=inward

Classifying dysphagic swallowing sounds with support vector machines Reviewed

Miyagi S., Sugiyama S., Kozawa K., Moritani S., Sakamoto S.I., Sakai O.

Healthcare (Switzerland) 8 ( 2 ) 2020.6

Language：English Publishing type：Research paper (scientific journal) Publisher：Healthcare (Switzerland)

Swallowing sounds from cervical auscultation include information related to the swallowing function. Several studies have been conducted on the screening tests of dysphagia. The literature shows a significant difference between the characteristics of swallowing sounds obtained from different subjects (e.g., healthy and dysphagic subjects; young and old adults). These studies demonstrate the usefulness of swallowing sounds during dysphagic screening. However, the degree of classification for dysphagia based on swallowing sounds has not been thoroughly studied. In this study, we investigate the use of machine learning for classifying swallowing sounds into various types, such as normal swallowing or mild, moderate, and severe dysphagia. In particular, swallowing sounds were recorded from patients with dysphagia. Support vector machines (SVMs) were trained using some features extracted from the obtained swallowing sounds. Moreover, the accuracy of the classification of swallowing sounds using the trained SVMs was evaluated via cross-validation techniques. In the two-class scenario, wherein the swallowing sounds were divided into two categories (viz. normal and dysphagic subjects), the maximum F-measure was 78.9%. In the four-class scenario, where the swallowing sounds were divided into four categories (viz. normal subject, and mild, moderate, and severe dysphagic subjects), the F-measure values for the classes were 65.6%, 53.1%, 51.1%, and 37.1%, respectively.

DOI： 10.3390/healthcare8020103

Other Link： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85088588517&origin=inward

　PREV- NEXT 　

1
2
3
4
5
6