「医工連携」は長い間強調されてきた概念であるが，医療における新規性や有用性と工学技術における新規性や有用性が必ずしも一致しないということをまず押さえておくことが肝要である。これは，“医”と“工”が等しく主になることが難しく，医工の横連携プロセスがそもそも困難であ

The medical field is evolving day by day, and the role of training clinical engineers is also undergoing changes. In this explanatory article, we will focus on new training methods for clinical engineers based on recent advances in AI and ICT technology and their impact on medical care. In modern medicine, diagnostic support using AI and remote medical care technology using ICT are considered essential skills. It is necessary for clinical engineers to acquire new perspectives and knowledge to utilize these technologies effectively and provide optimal care to patients. This explanatory article explains new educational models and research policies. Specifically, we will propose a review of the educational curriculum and training content for clinical engineers based on examples of image analysis using AI and remote monitoring systems using ICT. Furthermore, he touched on innovations in medical education that incorporate AR and VR technology, emphasized the importance of practical training, and discussed the challenges that the next generation of clinical engineers should tackle, how to overcome those challenges, and the future.

MicroRNA (miRNA), a type of non-coding RNA, is known to suppress the expression of target genes and there is an increasing number of studies analyzing various miRNAs in cancer. In an integrated analysis of miRNA and gene expressions, it is often difficult to obtain a significant result with a small amount of data. In this study, we proposed a method that allows integrated analysis even with a small amount of miRNA and gene expression data, and examined its validity. The method is based on a group comparison between target genes having specific functions controlled by a specific miRNA and a background group of randomly selected genes. We conducted a simulation to verify the validity of the proposed method. The simulation results indicated that the method can detect significant differences in expression of target genes controlled by a miRNA showing significant expression change. We then applied the proposed method to real data derived from multiple myeloma cells. As a result, some functions were detected as significantly changed due to a significant change of miRNA.

Due to the increasing number of patients with chronic obstructive pulmonary disease (COPD) and lung cancer, the use of low-field MRI (magnetic resonance imaging) is attracting attention. Low-field MRI is less expensive and safer than high-field MRI. It is also possible to obtain high-sensitivity images using "hyperpolarization" of certain gases (eg ¹²⁹Xe gas). In this study, we aimed to develop lung field imaging technology by low-field MRI of 0.3 T using hyperpolarized ¹²⁹Xe gas, and observed chemical shifts in saline and porcine blood. As a result, it is expected to expand the possibility of high-sensitivity imaging diagnosis in a low magnetic field of 0.3 T and contribute to a new understanding of hyperpolarized ¹²⁹Xe gas, and the practical application of these technologies is expected.

Many aspects of how magnetic particles, particularly magnetite particles, are distributed in living organisms and how they affect brain function and neurodegenerative diseases remain unclear. There is an urgent need to develop new methods and techniques to non-invasively and highly accurately detect the presence of these magnetic particles and estimate their location. In this study, we adopted Nearest Neighbors as a machine learning algorithm and analyzed the inverse problem of the machine learning model to estimate the position of magnetic particles that are the source of biomagnetic signals. By arranging the magnetic sensors in three dimensions, the percentage of position estimation errors of 1 cm or less increased, even though there were only 8 magnetic sensors, and the average position estimation error per horizontal plane was approximately 8 mm. Since the resolution of conventional magnetoencephalography equipment is 5 to 7 mm, and measurements are performed using approximately 150 SQUID sensors, it is possible to improve position estimation accuracy by adjusting the placement conditions of the magnetic sensors.

Spatial data from a portable GPS logger have the potential to allow objective monitoring of locations where physical activity is taking place in a community. However, few studies have identified common physical activity locations visited by many people. This study aimed to propose a method for extracting such locations using GPS and accelerometer data using the k means method. The validity of the proposed method was examined using simulation data and then the method was applied to actual data obtained in a previous study. The participants in the previous study were healthy older people who wore GPS loggers and accelerometers for seven consecutive days. The k means method with various k was applied to the combined data of GPS and accelerometer to extract physical activity locations common to some participants. Simulation results indicated that the proposed method was able to extract locations where many participants performed physical activity of 3.0 METs or more. The results using the actual data from 50 participants also showed the validity of the proposed method.

Many healthcare-oriented recipe recommendation methods are proposed in previous research. In some previous research, the additional purchase cost and the food loss are minimized in addition to the maximization of the healthcare score under the limits on the cost of purchasing additional ingredients and food loss when there are multiple optimal solutions. But in the previous research, only the consecutive additional ingredient purchase dates and the consecutive recipe recommendation dates are considered. In this research, new healthcare-oriented recipe recommendation methods with arbitrarily setting of additional ingredient purchase dates and recipe recommendation dates. Healthcare loss per 1 recommendation and healthcare loss per T recommendations are defined. Each healthcare loss is minimized by each proposed method under the limits on the cost of purchasing additional ingredients and food loss. The additional purchase cost and the food loss are also minimized in addition to the minimization of the healthcare loss when there are multiple optimal solutions. The effectiveness of the proposed methods is shown by some computational examples.

In this study, we designed, fabricated and evaluated a dark hole generation optical system that combines an annular phase plate and an annular wave plate for super-resolution microscopy based on stimulated emission or upconversion fluorescence depletion. The phase plate is a compound phase plate that combines two types of spiral phase plates in a ring shape, and the wave plate is a ring shape combination of two types of wave plates with crystal axes perpendicular to each other. Despite its complex structure, the composite phase plate could be precisely fabricated using the exposure and etching processes used in semiconductor manufacturing. A super-resolution microscope equipped with these devices confirmed the generation of a three-dimensional dark hole of fluorescence-suppressed light.

When developing a blink input interface, conscious (voluntary) and natural (involuntary) blink types must be automatically classified. We previously proposed a method for blink type classification using a 3D convolutional neural network (3D CNN). This CNN model outputs a predicted probability that determines three classes: “voluntary blinking,” “involuntary blinking,” and “not blinking” from a periocular image sequence. Previously, we found that the bias of the eye position in the input image is a factor that reduces the classification accuracy. To address this problem, we employ data augmentation with a shifting 5 or 10 pixels in the horizontal and/or vertical directions.

The purpose of this research is to propose a small-scale adaptive HDR (High Dynamic Range) system that can be implemented in cameras that can be applied to simple monitoring systems widely used in small cameras such as webcams, and to develop a gamma correction circuit to be installed in the HDR system to correct image degradation in the camera. Another purpose of this research is to show that image degradation can be largely corrected by controlling the optimal gamma correction value according to the average luminance value of the captured image, and proposed a conversion formula. We also proposed a gamma correction circuit that can electronically control its exponent value, and confirmed the correcting operation as designed by making a prototype circuit. Actually, the gamma correction circuit was used to display a grayscale pattern on an LCD, and the effect of image correction was verified. Future task is correcting image damages of black down and blown out at the same time. And another task is to increase higher cut off frequency of the gamma correction circuit.

A lateral overflow integration capacitor (LOFIC) complementary metal oxide semiconductor (CMOS) image sensor can realize high-dynamic-range (HDR) imaging with combination of low-noise signals and high-full-well-capacity (high-FWC) signals. However, signal-to-noise ratio (SNR) drop at the switching point between the low-noise signals and the high-FWC signals is not negligible because pixel reset kTC noise to reset small in-pixel capacitor is not canceled in the high-FWC signals. In this paper, a reset circuit for the LOFIC CMOS image sensor is presented, which reduces the pixel reset noise with additional switch and large capacitor outside the pixel array. In order to further reduce the reset noise, an asymmetric structured in-pixel reset transistor is also proposed. Since the charge flow through the channel of in-pixel reset transistor is suppressed, the pixel reset kTC noise is reduced. A test chip of the reset circuit is fabricated with a 0.18 µm CMOS process. As an evaluation result, the pixel reset noise is reduced to 556 µV_rms, which is 43.0 % lower than that of the conventional LOFIC pixel. The estimated SNR drop at the switching point is 6.5 dB, which is improved by 4.2 dB.

Peltier device is used for both thermoelectric cooling and thermoelectric power generation, where the Peltier and Seebeck effect occur simultaneously when current is applied. Conventional thermal equivalent circuits for Peltier device consider only the effects that are used out of the effects that occur simultaneously. Therefore, in the past, our laboratory proposed a thermal equivalent circuit considers the Seebeck effect in thermoelectric cooling. In this paper, we propose a thermal equivalent circuit that considers the Peltier effect in thermoelectric generation. Furthermore, from the analysis results of the thermal and electrical equivalent circuits during thermoelectric cooling and thermoelectric power generation, the amount of thermal and electrical energy converted is quantitatively clarified.

Also, General-purpose Peltier devices have a hybrid integrated circuit structure in which heat is transferred through a semiconductor and a welded metal, and the difference in thermal expansion coefficient between the two causes a short life. Therefore, the authors devised a structure in which the Peltier device is completely monolithic and fabricated a prototype Peltier integrated circuit using a general-purpose CMOS process to extend the life of Peltier device. we report on the structure of the prototype monolithic Peltier IC and the experimental confirmation of the thermoelectric effect.

In this paper, we present a visual feedback control method via an input-to-state safe control barrier function (ISSf-CBF). The control objective is that a camera-mounted mobile robot achieves the desired relative pose to a target object with avoiding any elliptic cylinder obstacles under input disturbances. First, we derive a visual feedback error system with visual motion observers to discuss stability via the Lyapunov method. Through the kinematic model for a relative rigid body, we design an ISSf-CBF for elliptic cylinder obstacle avoidance under input disturbances. Next, we present a control Lyapunov function (CLF) for the visual feedback error system. By using the ISSf-CBF and the CLF, we discuss a unified quadratic program formulation that connects the safety and the stability for pose control of a camera-mounted mobile robot under input disturbances. Finally, simulation and experimental results confirm the validity of the proposed ISSf-CBF control method compared to the other one which only achieves the desired pose.

We focus on household garbage collection services to find an optimization method that simultaneously satisfies the convenience of residents and the efficiency of vehicle collection. The cost of garbage collection depends on the location of the garbage stations and the routes taken by a fleet of vehicles. We try to reduce collecting costs by optimizing the placement of the garbage station. In this problem, it is necessary not only to reduce the costs but also to reflect the convenience of residents. In this study, we use a hybrid solution method mixing meta-heuristics and mathematical programming to find better locations within realistic computational time. We decompose the problem into two problems: garbage station allocation and collecting routes, and optimize it using meta-heuristics and mathematical programming. The placement of garbage stations reflects the residents' questionnaire as a constraint in the optimization model.

The Sixth Science, Technology and Innovation Basic Plan of the Cabinet Office points out the importance of integrating cyberspace and physical space and fostering the ability to solve problems spontaneously in order to realize a new society, "Society 5.0". In recent years, model-based development that can be developed efficiently and quickly using mathematical models has been attracting attention in the industrial world. On the other hand, in the field of education, problem-solving learning, including "problem finding" and "problem setting," is emphasized in the Courses of Study for junior high schools announced in 2009. In this study, we develop a curriculum including teaching materials for fostering problem-solving skills in the field of measurement and control programming in the junior high school technology course that incorporates the perspective of model-based developmental thinking, conduct classroom practice, and verify its usefulness based on students’ impressions and mind maps.

Disturbance suppression is an important issue in control system design. Model Predictive Controllers (MPCs) excel in suppressing known disturbances, offering uniform handling of reference tracking and disturbance response. This study aims to suppress the influence of unknown disturbances using the cost-effective Predictive Functional Control (PFC), a simplified MPC. However, addressing disturbances with PFC necessitates disturbance knowledge or estimation. This paper employs a Disturbance Observer (DOB) for estimation, but a basic DOB struggles with noisy or time-delay systems. To overcome this, our proposed method incorporates a line enhancer from speech processing. Numerical examples in this paper show that the proposed scheme works effectively and achieves the desired disturbance suppression.

In this paper, an omnidirectional source tracking method based on sensitivity difference of microphone pairs is proposed. In the conventional method, we developed the direction estimation method based on a histogram integrating the estimation results of multiple two-microphone pairs. However, the estimation accuracy degradation often occurs despite the number of microphones. When a sound source exists at a wide angle to the microphone pairs, the estimation direction indicates highly sensitive to the measurement error of TDOA(Time-Difference-Of-Arrival) between microphones, and the estimation accuracy is greatly degradated. Therefore, it is considered that the same reliability should not be given to all estimation results to the same sound source. Because the estimation accuracy differs for each microphone pairs even for the same sound source, In the proposed method, such a difficulty is resolved by selecting the microphone pair used for the estimation, then the another pair is used just to judge whether the source exists in front or in back. As a result, the estimation accuracy can be improved with half the number of microphones of the conventional method. The effectiveness of method is shown from several experimental results.

Cloud bursting is used to offload processing to the cloud when on-premise data centers run out of computing resources. Data in the on-premise data center needs to be copied to the cloud at the start of cloud bursting. Since data capacity and usage of connection between on-premise data center and cloud vary, the time required to build a cloud bursting environment will vary. This made it difficult to determine a configuration that could resume processing within the recovery time objective at a low cost in the event of a failure in the cloud. In this study, we proposed a redundancy determination method based on reliability requirements that estimates the time required to build a cloud bursting environment considering data capacity and the usage of connection between on-premise data centers and the cloud, and determines a configuration that satisfies recovery time objective at low cost. Assuming the case of cloud bursting of data analysis processing, we conducted an experiment in which a system simulating an analysis application was cloud bursted from an AZ simulating an on-premise data center to the main AZ, and the range of the system built in advance in the sub AZ was changed to simulate a sub AZ failure and failover to the sub AZ. As a result, it was confirmed that the time required for failover can be reduced by increasing the extent to which the system is preconstructed in the sub AZ. The proposed method is expected to enable us to determine a configuration that can meet the recovery time objective at a low cost.

This study conducted basic experiments to develop a detection technique for the dart-out of pedestrians from non-line-of-sight areas based on diffracted waves measured by 2.4 GHz-band radar. We demonstrated accurate classification of walking motions with high or low risks of dart-out based on deep learning of range profile images.