In step with the development of Industry 4.0, research on automatic operation technology and components related to automobiles is continuously being conducted. In particular, the torque angle sensor (TAS) module of the steering wheel system is considered to be a core technology owing to its precise angle, torque sensing, and high-speed signal processing. In the case of conventional TAS modules, in addition to the complicated gear structure, there is an error in angle detection due to the backlash between the main and sub-gear. In this paper, we propose a multi-track encoder-based vehicle steering system, which is incorporated with a TAS module structure that minimizes the number of components and the angle detection error of the module compared with existing TAS modules. We also fabricated and tested an angle detection signal processing board and evaluated it on a test stand. As a result, we could confirm its excellent performance of an average deviation of 0.4° and applicability to actual vehicles by evaluating its electromagnetic interference (EMI) environmental reliability. The ultimate goal of the TAS module is to detect the target steering angle with minimal computation by the steering or main electronic control unit (ECU) to meet the needs of the rapidly growing vehicle technology. The verified angle detection module can be applied to an actual steering system in accordance with the mentioned technical requirements.

Current deep learning-based speech enhancement methods focus on enhancing the time–frequency representation of the signal. However, conventional methods can lead to speech damage due to resolution mismatch problems that emphasize only specific information in the time or frequency domain. To address these challenges, this paper introduces a speech enhancement model designed with a dual-path structure that identifies key speech characteristics in both the time and time–frequency domains. Specifically, the time path aims to model semantic features hidden in the waveform, while the time–frequency path attempts to compensate for the spectral details via a spectral extension block. These two paths enhance temporal and spectral features via mask functions modeled as LSTM, respectively, offering a comprehensive approach to speech enhancement. Experimental results show that the proposed dual-path LSTM network consistently outperforms conventional single-domain speech enhancement methods in terms of speech quality and intelligibility.

Recently, the development of medical rehabilitation technology has resulted in an increased interest in speech therapy equipment. In particular, research on articulation therapy for communication disorders is being actively conducted. The existing methods for the diagnosis and treatment of speech disorders, such as traditional tactile perception tests and methods based on the empirical judgment of speech therapists, have many limitations. Moreover, the position and contact force of the tongue are key factors in speech disorders with regards to articulation. This is a very important factor in the distinction of Korean characters such as lax, tense and aspirated consonants. In this study, we proposed a Korean-electropalatography (EPG) system to easily measure and monitor the position and contact force of the tongue during articulation treatment and diagnosis. In our proposed K-EPG system, a sensor was fabricated using an AgCl electrode and biocompatible silicon. Furthermore, the measured signal was analyzed by implementing a bio-signal processing module and monitoring program. In particular, the bio-signal was measured by inserting the device into the palate of an experimental healthy test group (four subjects). Through these experiments, we confirmed that our K-EPG system could be applied to clinical treatment in speech therapy.

Background: Breast cancer diagnosis increasingly relies on ultrasound imaging, but challenges related to operator dependency and image quality limitations necessitate automated segmentation approaches. Traditional U-Net architectures, while widely used for medical image segmentation, suffer from shallow encoder structures that limit feature extraction capabilities. Methods: This study proposes an enhanced segmentation model that replaces the conventional U-Net encoder with an Inception architecture and employs transfer learning using ImageNet pre-trained weights. The model was trained and evaluated on a dataset of 900 breast ultrasound images from Kyungpook National University Hospital. Performance evaluation utilized multiple metrics including Intersection over Union (IoU), Dice coefficient, precision, and recall scores. Results: The proposed Inception U-Net achieved superior performance with an IoU score of 0.7774, Dice score of 0.8491, precision score of 0.7081, and recall score of 0.7174, demonstrating approximately 5% improvement over baseline U-Net architecture across all evaluation metrics. Conclusions: The integration of Inception modules within the U-Net architecture effectively addresses feature extraction limitations in breast ultrasound segmentation. Transfer learning from ImageNet datasets proves beneficial even across domain differences, establishing a foundation for broader medical imaging applications.

Although recent deep-learning-based speech enhancement (SE) methods significantly outperform traditional approaches, their computational demands often scale proportionally with their performance. This makes them usually impractical for deployment on resource-constrained computing platforms (edge devices) like smartphones, IoT devices, and digital hearing aids, which process data locally. In this paper, we propose a novel lightweight spectral enhancement network (LSENet) that is designed to estimate high-quality speech with minimal computational overhead. The network consists of an encoder-decoder architecture enhanced by a group-dilated convolutional module, which efficiently leverages time-frequency domain information while significantly reducing resource consumption through dilated convolutional groups and spectral-wise attention modules. In addition, to capture the long-range contextual dependencies of the extracted features, an improved dual-path recurrent neural network is introduced between the encoder-decoder structure. Experimental results show that the proposed model achieves competitive performance with state-of-the-art baseline models on the Voicebank + Demand and DNS-Challenge datasets while requiring only 39.4 thousand model parameters and 237 million multiply-accumulate operations.

Background and Objectives: Petechial cerebral hemorrhages can be caused by various factors, such as traumas, cerebral infarctions, and aging, and is related to the disruption of the blood–brain barrier or the cellular damage of blood vessels. However, there is no animal model that recapitulates cerebral petechial hemorrhages. Materials and Methods: Here, we implemented a petechial hemorrhage using a novel technology, i.e., microbubble-assisted focused ultrasound (MB + FUS). Results: This method increases the permeability of the blood–brain barrier by directly applying mechanical force to the vascular endothelial cells through cavitation of the microbubbles. Microbubble-enhanced cavitation has the advantage of controlling the degree and location of petechial hemorrhages. Conclusions: We thus generated a preclinical rat model using noninvasive focal MB + FUS. This method is histologically similar to actual petechial hemorrhages of the brain and allows the achievement of a physiologically resembling petechial hemorrhage. In the future, this method shall be considered as a useful animal model for studying the pathophysiology and treatment of petechial cerebral hemorrhages.