-
Anatomy-aware Self-supervised Learning for Anomaly Detection in Chest Radiographs
Authors:
Junya Sato,
Yuki Suzuki,
Tomohiro Wataya,
Daiki Nishigaki,
Kosuke Kita,
Kazuki Yamagata,
Noriyuki Tomiyama,
Shoji Kido
Abstract:
Large numbers of labeled medical images are essential for the accurate detection of anomalies, but manual annotation is labor-intensive and time-consuming. Self-supervised learning (SSL) is a training method to learn data-specific features without manual annotation. Several SSL-based models have been employed in medical image anomaly detection. These SSL methods effectively learn representations i…
▽ More
Large numbers of labeled medical images are essential for the accurate detection of anomalies, but manual annotation is labor-intensive and time-consuming. Self-supervised learning (SSL) is a training method to learn data-specific features without manual annotation. Several SSL-based models have been employed in medical image anomaly detection. These SSL methods effectively learn representations in several field-specific images, such as natural and industrial product images. However, owing to the requirement of medical expertise, typical SSL-based models are inefficient in medical image anomaly detection. We present an SSL-based model that enables anatomical structure-based unsupervised anomaly detection (UAD). The model employs the anatomy-aware pasting (AnatPaste) augmentation tool. AnatPaste employs a threshold-based lung segmentation pretext task to create anomalies in normal chest radiographs, which are used for model pretraining. These anomalies are similar to real anomalies and help the model recognize them. We evaluate our model on three opensource chest radiograph datasets. Our model exhibit area under curves (AUC) of 92.1%, 78.7%, and 81.9%, which are the highest among existing UAD models. This is the first SSL model to employ anatomical information as a pretext task. AnatPaste can be applied in various deep learning models and downstream tasks. It can be employed for other modalities by fixing appropriate segmentation. Our code is publicly available at: https://github.com/jun-sato/AnatPaste.
△ Less
Submitted 11 June, 2022; v1 submitted 9 May, 2022;
originally announced May 2022.
-
On a comparison theorem for parabolic equations with nonlinear boundary conditions
Authors:
Kosuke Kita,
Mitsuharu Ôtani
Abstract:
In this paper, a new type of comparison theorem is presented for some initial-boundary value problems of second order nonlinear parabolic systems with nonlinear boundary conditions. This comparison theorem has an advantage over the classical ones, since this makes it possible to compare two solutions satisfying different types of boundary conditions. Some applications are given in the last section…
▽ More
In this paper, a new type of comparison theorem is presented for some initial-boundary value problems of second order nonlinear parabolic systems with nonlinear boundary conditions. This comparison theorem has an advantage over the classical ones, since this makes it possible to compare two solutions satisfying different types of boundary conditions. Some applications are given in the last section, where the existence of blow-up solutions is shown for some nonlinear parabolic equations and systems with nonlinear boundary conditions.
△ Less
Submitted 4 September, 2021;
originally announced September 2021.
-
Hexagonal Nanopits with the Zigzag Edge State on Graphite Surfaces Synthesized by Hydrogen-Plasma Etching
Authors:
Tomohiro Matsui,
Hideki Sato,
Kazuma Kita,
André E. B. Amend,
Hiroshi Fukuyama
Abstract:
We studied, by scanning tunneling microscopy, the morphology of nanopits of monolayer depth created at graphite surfaces by hydrogen plasma etching under various conditions such as H$_2$ pressure, temperature, etching time, and RF power of the plasma generation. In addition to the known pressure-induced transition of the nanopit morphology, we found a sharp temperature-induced transition from many…
▽ More
We studied, by scanning tunneling microscopy, the morphology of nanopits of monolayer depth created at graphite surfaces by hydrogen plasma etching under various conditions such as H$_2$ pressure, temperature, etching time, and RF power of the plasma generation. In addition to the known pressure-induced transition of the nanopit morphology, we found a sharp temperature-induced transition from many small rather round nanopits of ~150 nm size to few large hexagonal ones of 300-600 nm within a narrow temperature range. The remote and direct plasma modes switching mechanism, which was proposed to explain the pressure-induced transition, is not directly applicable to this newly found transition. Scanning tunneling spectroscopy (STS) measurements of edges of the hexagonal nanopits fabricated at graphite surfaces by this method show clear signatures of the peculiar electronic state localized at the zigzag edge (edge state), i.e., a prominent peak near the Fermi energy accompanied by suppressions on either side in the local density of states. These observations indicate that the hexagonal nanopits consist of a high density of zigzag edges. The STS data also revealed a domain structure of the edge state in which the electronic state varies over a length scale of ~3 nm along the edge. The present study will pave the way for microscopic understanding of the anisotropic etching mechanism and of spin polarization in zigzag nanoribbons which are promising key elements for future graphene nanoelectronics.
△ Less
Submitted 31 July, 2019;
originally announced August 2019.
-
Accurate decoding of materials using a finger mounted accelerometer
Authors:
Kuniharu Sakurada,
Gowrishankar Ganesh,
Wenwei Yu,
Kahori Kita
Abstract:
Sensory feedback is the fundamental driving force behind motor control and learning. However, the technology for low-cost and efficient sensory feedback remains a big challenge during stroke rehabilitation, and for prosthetic designs. Here we show that a low-cost accelerometer mounted on the finger can provide accurate decoding of many daily life materials during touch. We first designed a customi…
▽ More
Sensory feedback is the fundamental driving force behind motor control and learning. However, the technology for low-cost and efficient sensory feedback remains a big challenge during stroke rehabilitation, and for prosthetic designs. Here we show that a low-cost accelerometer mounted on the finger can provide accurate decoding of many daily life materials during touch. We first designed a customized touch analysis system that allowed us to present different materials for touch by human participants, while controlling for the contact force and touch speed. Then, we collected data from six participants, who touched seven daily life materials-plastic, cork, wool, aluminum, paper, denim, cotton. We use linear sparse logistic regression and show that the materials can be classified from accelerometer recordings with an accuracy of 88% across materials and participants within 7 seconds of touch.
△ Less
Submitted 19 June, 2019;
originally announced June 2019.
-
Google Street View image of a house predicts car accident risk of its resident
Authors:
Kinga Kita,
Łukasz Kidziński
Abstract:
Road traffic injuries are a leading cause of death worldwide. Proper estimation of car accident risk is critical for appropriate allocation of resources in healthcare, insurance, civil engineering, and other industries. We show how images of houses are predictive of car accidents. We analyze 20,000 addresses of insurance company clients, collect a corresponding house image using Google Street View…
▽ More
Road traffic injuries are a leading cause of death worldwide. Proper estimation of car accident risk is critical for appropriate allocation of resources in healthcare, insurance, civil engineering, and other industries. We show how images of houses are predictive of car accidents. We analyze 20,000 addresses of insurance company clients, collect a corresponding house image using Google Street View, and annotate house features such as age, type, and condition. We find that this information substantially improves car accident risk prediction compared to the state-of-the-art risk model of the insurance company and could be used for price discrimination. From this perspective, public availability of house images raises legal and social concerns, as they can be a proxy of ethnicity, religion and other sensitive data.
△ Less
Submitted 10 April, 2019;
originally announced April 2019.
-
Bounds for global solutions of a reaction diffusion system with the Robin boundary conditions
Authors:
Kosuke Kita,
Mitsuharu Ôtani
Abstract:
In this paper, we consider the large-time behavior of solutions of a reaction diffusion system arising from a nuclear reactor model with the Robin boundary conditions, which consists of two real-valued unknown functions. In particular, we show that global solutions of this system are uniformly bounded in a suitable norm with respect to time. Since this system has no variational structure, we canno…
▽ More
In this paper, we consider the large-time behavior of solutions of a reaction diffusion system arising from a nuclear reactor model with the Robin boundary conditions, which consists of two real-valued unknown functions. In particular, we show that global solutions of this system are uniformly bounded in a suitable norm with respect to time. Since this system has no variational structure, we cannot apply the standard methods relying on the Lyapunov functional in order to obtain a priori estimates of global solutions. To cope with this difficulty, we make use of the weighted $L^1$ norm characterized by the first eigenfunction of Laplacian with the Robin boundary condition.
△ Less
Submitted 9 November, 2018;
originally announced November 2018.
-
On some parabolic systems arising from a nuclear reactor model with nonlinear boundary conditions
Authors:
Kosuke Kita,
Mitsuharu Ôtani,
Hiroki Sakamoto
Abstract:
In this paper, we are concerned with a reaction diffusion system arising from a nuclear reactor model in bounded domains with nonlinear boundary conditions. We show the existence of a stationary solution and its ordered uniqueness. It is also shown that every positive stationary solution possesses threshold property to determine blow-up or globally existence for solutions of nonstationary problem.
In this paper, we are concerned with a reaction diffusion system arising from a nuclear reactor model in bounded domains with nonlinear boundary conditions. We show the existence of a stationary solution and its ordered uniqueness. It is also shown that every positive stationary solution possesses threshold property to determine blow-up or globally existence for solutions of nonstationary problem.
△ Less
Submitted 11 May, 2018;
originally announced May 2018.
-
3D Channel Model in 3GPP
Authors:
Bishwarup Mondal,
Timothy A. Thomas,
Eugene Visotsky,
Frederick W. Vook,
Amitava Ghosh,
Young-Han Nam,
Yang Li,
Charlie Zhang,
Min Zhang,
Qinglin Luo,
Yuichi Kakishima,
Koshiro Kitao
Abstract:
Multi-antenna techniques capable of exploiting the elevation dimension are anticipated to be an important air-interface enhancement targeted to handle the expected growth in mobile traffic. In order to enable the development and evaluation of such multi-antenna techniques, the 3rd generation partnership project (3GPP) has recently developed a 3-dimensional (3D) channel model. The existing 2-dimens…
▽ More
Multi-antenna techniques capable of exploiting the elevation dimension are anticipated to be an important air-interface enhancement targeted to handle the expected growth in mobile traffic. In order to enable the development and evaluation of such multi-antenna techniques, the 3rd generation partnership project (3GPP) has recently developed a 3-dimensional (3D) channel model. The existing 2-dimensional (2D) channel models do not capture the elevation channel characteristics lending them insufficient for such studies. This article describes the main components of the newly developed 3D channel model and the motivations behind introducing them. One key aspect is the ability to model channels for users located on different floors of a building (at different heights). This is achieved by capturing a user height dependency in modelling some channel characteristics including pathloss, line-of-sight (LOS) probability, etc. In general this 3D channel model follows the framework of WINNERII/WINNER+ while also extending the applicability and the accuracy of the model by introducing some height and distance dependent elevation related parameters.
△ Less
Submitted 5 February, 2015;
originally announced February 2015.
-
Melting Transition of Vortex Lattice in Point Vortex Systems
Authors:
Hidetsugu Sakaguchi,
Kenji Kita
Abstract:
Point vortices take a triangular lattice structure in a rotating system as a minimum energy state. We perform a numerical simulation of point vortex systems using initial conditions indicating that the triangular lattice is randomly perturbed. The total energy increases with the magnitude of the perturbation. When the energy is increased, the vortex lattice becomes irregular and a layered structur…
▽ More
Point vortices take a triangular lattice structure in a rotating system as a minimum energy state. We perform a numerical simulation of point vortex systems using initial conditions indicating that the triangular lattice is randomly perturbed. The total energy increases with the magnitude of the perturbation. When the energy is increased, the vortex lattice becomes irregular and a layered structure appears. When the energy is further increased, the layered structure disappears and a liquidlike state appears. We interpret the melting transition with a mean-field approximation for layered structures.
△ Less
Submitted 25 September, 2012;
originally announced September 2012.
-
Electrical transport properties of graphene on SiO2 with specific surface structures
Authors:
K. Nagashio,
T. Yamashita,
T. Nishimura,
K. Kita,
A. Toriumi
Abstract:
The mobility of graphene transferred on a SiO2/Si substrate is limited to ~10,000 cm2/Vs. Without understanding the graphene/SiO2 interaction, it is difficult to improve the electrical transport properties. Although surface structures on SiO2 such as silanol and siloxane groups are recognized, the relation between the surface treatment of SiO2 and graphene characteristics has not yet been elucidat…
▽ More
The mobility of graphene transferred on a SiO2/Si substrate is limited to ~10,000 cm2/Vs. Without understanding the graphene/SiO2 interaction, it is difficult to improve the electrical transport properties. Although surface structures on SiO2 such as silanol and siloxane groups are recognized, the relation between the surface treatment of SiO2 and graphene characteristics has not yet been elucidated. This paper discusses the electrical transport properties of graphene on specific surface structures of SiO2 prepared by O2-plasma treatments and reoxidization.
△ Less
Submitted 28 June, 2011;
originally announced June 2011.
-
Contact resistivity and current flow path at metal/graphene contact
Authors:
K. Nagashio,
T. Nishimura,
K. Kita,
A. Toriumi
Abstract:
The contact properties between metal and graphene were examined. The electrical measurement on a multiprobe device with different contact areas revealed that the current flow preferentially entered graphene at the edge of the contact metal. The analysis using the cross-bridge Kelvin structure (CBK) suggested that a transition from the edge conduction to area conduction occurred for a contact lengt…
▽ More
The contact properties between metal and graphene were examined. The electrical measurement on a multiprobe device with different contact areas revealed that the current flow preferentially entered graphene at the edge of the contact metal. The analysis using the cross-bridge Kelvin structure (CBK) suggested that a transition from the edge conduction to area conduction occurred for a contact length shorter than the transfer length of ~1 micron. The contact resistivity for Ni was measured as ~5*10-6 Ohmcm2 using the CBK. A simple calculation suggests that a contact resistivity less than 10-9 Ohmcm2 is required for miniaturized graphene field effect transistors.
△ Less
Submitted 10 August, 2010;
originally announced August 2010.
-
Systematic Investigation of the Intrinsic Channel Properties and Contact Resistance of Monolayer and Multilayer Graphene FET
Authors:
Kosuke Nagashio,
Tomonori Nishimura,
Koji Kita,
Akira Toriumi
Abstract:
The intrinsic channel properties of monolayer and multilayer graphene were systematically investigated as a function of layer number by the exclusion of contact resistance using four-probe measurements. We show that the continuous change in normalized sheet resistivity from graphite to a bilayer graphene is governed by one unique property, i.e., the band overlap, which markedly increases from 1 me…
▽ More
The intrinsic channel properties of monolayer and multilayer graphene were systematically investigated as a function of layer number by the exclusion of contact resistance using four-probe measurements. We show that the continuous change in normalized sheet resistivity from graphite to a bilayer graphene is governed by one unique property, i.e., the band overlap, which markedly increases from 1 meV for a bilayer graphene to 11 meV for eight layers and eventually reaches 40 meV for graphite. The monolayer graphene, however, showed a deviation in temperature dependence due to a peculiar linear dispersion. Additionally, contact resistivity was extracted for the case of typical Cr/Au electrodes. The observed high contact resistivity, which varies by three orders of magnitude (from ~103 to 106 Ohm micron), might significantly mask the outstanding performance of the monolayer graphene channel, suggesting its importance in future research.
△ Less
Submitted 12 April, 2010;
originally announced April 2010.
-
Mobility variations in mono- and multi-layer graphene films
Authors:
Kosuke Nagashio,
Tomonori Nishimura,
Koji Kita,
Akira Toriumi
Abstract:
The electric properties of mono- and multi-layer graphene films were systematically studied with the layer number determined by their optical contrast. The current modulation increased monotonically with a decrease in the layer number due to the reduction of the interlayer scattering. Carrier mobility in the monolayer was significantly greater than that in the multilayer due to linear dispersion…
▽ More
The electric properties of mono- and multi-layer graphene films were systematically studied with the layer number determined by their optical contrast. The current modulation increased monotonically with a decrease in the layer number due to the reduction of the interlayer scattering. Carrier mobility in the monolayer was significantly greater than that in the multilayer due to linear dispersion relation. On the other hand, in the monolayer, carrier transport was extremely sensitive to charged impurity density due to the reduction in screening effect, which causes larger mobility variation. Reduction of the charged impurity density is thus key for high mobility.
△ Less
Submitted 11 December, 2008;
originally announced December 2008.
