Sungwook Yu

Background: Deep learning has significantly advanced medical image analysis, particularly in semantic segmentation, which is essential for clinical decisions. However, existing 3D segmentation models, like the traditional 3D UNet, face challenges in balancing computational efficiency and accuracy when processing volumetric medical data. This study aims to develop an improved architecture for 3D medical image segmentation with enhanced learning strategies to improve accuracy and address challenges related to limited training data. Methods: We propose ES-UNet, a 3D segmentation architecture that achieves superior segmentation performance while offering competitive efficiency across multiple computational metrics, including memory usage, inference time, and parameter count. The model builds upon the full-scale skip connection design of UNet3+ by integrating channel attention modules into each encoder-to-decoder path and incorporating full-scale deep supervision to enhance multi-resolution feature learning. We further introduce Region Specific Scaling (RSS), a data augmentation method that adaptively applies geometric transformations to annotated regions, and a Dynamically Weighted Dice (DWD) loss to improve the balance between precision and recall. The model was evaluated on the MICCAI HECKTOR dataset, and additional validation was conducted on selected tasks from the Medical Segmentation Decathlon (MSD). Results: On the HECKTOR dataset, ES-UNet achieved a Dice Similarity Coefficient (DSC) of 76.87%, outperforming baseline models including 3D UNet, 3D UNet 3+, nnUNet, and Swin UNETR. Ablation studies showed that RSS and DWD contributed up to 1.22% and 1.06% improvement in DSC, respectively. A sensitivity analysis demonstrated that the chosen scaling range in RSS offered a favorable trade-off between deformation and anatomical plausibility. Cross-dataset evaluation on MSD Heart and Spleen tasks also indicated strong generalization. Computational analysis revealed that ES-UNet achieves superior segmentation performance with moderate computational demands. Specifically, the enhanced skip connection design with lightweight channel attention modules integrated throughout the network architecture enables this favorable balance between high segmentation accuracy and computational efficiency. Conclusion: ES-UNet integrates architectural and algorithmic improvements to achieve robust 3D medical image segmentation. While the framework incorporates established components, its core contributions lie in the optimized skip connection strategy and supporting techniques like RSS and DWD. Future work will explore adaptive scaling strategies and broader validation across diverse imaging modalities.

As the automotive industry continues to evolve rapidly, there is a growing demand for high-throughput reliable communication systems within vehicles. This paper presents the implementation and verification of a fault-tolerant Ethernet-based communication protocol tailored for automotive applications operating at 1 Gbps and above. The proposed system introduces a multi-port Network Interface Controller (NIC) architecture that supports real-time communication and robust fault handling. To ensure adaptability across various in-vehicle network (IVN) scenarios, the system allows for configurable packet sizes and transmission rates and supports diverse data formats. The architecture integrates cyclic redundancy check (CRC)-based error detection, real-time recovery mechanisms, and file-driven data injection techniques. Functional validation is performed using Verilog HDL simulations, demonstrating deterministic timing behavior, modular scalability, and resilience under fault injection. This paper presents a fault-tolerant Network Interface Controller (NIC), architecture incorporating CRC-based error detection, real-time recovery logic, and file-driven data injection. The system is verified through Verilog HDL simulation, demonstrating correct timing behavior, modular scalability, and robustness against injected transmission faults. Compared to conventional dual-port NICs, the proposed quad-port architecture demonstrates superior scalability and error tolerance under injected fault conditions. Experimental results confirm that the proposed NIC architecture achieves stable multi-port communication under embedded automotive environments. This study further introduces a novel quad-port NIC with an integrated fault injection algorithm and evaluates its performance in terms of error tolerance.

We present a mathematical model and algebraic equations for the dynamic structural analysis of the human musculoskeletal system, including the arms, muscles, and tendons. This analysis incorporates neural impulse signal transmission and other functions of the human musculoskeletal system. The intricate machinery of the human body, spanning from neurons to organs, provides a fundamental basis for the study of larger, complex systems such as human-like robotic systems. To examine the detailed movements of individual muscle groups, mathematical modeling of artificial arms is essential to understand the signal transmission system and the overall mechanism governing the arm bones in the musculoskeletal system. In this paper, we simulate the neural signal transmission system using algebraic equations and explore the model and significance of artificial arms that can be integrated into human-like robots. Our simulation results demonstrate that neural signals can be transmitted as computational values, specifically transitions between 0 and 1, in the context of human-like robot modeling. Furthermore, this paper provides a mathematical framework to demonstrate that nerve transmission signals act as physical forces, manifesting as electric signals or energy forms akin to human nerve signals. Ultimately, this research represents foundational mathematical modeling that could contribute to the development of robots capable of mimicking human movement, driven by the completion of force or energy transmission via neural impulse pathways within the human musculoskeletal system.

In recent times, many studies concerning surgical video analysis are being conducted due to its growing importance in many medical applications. In particular, it is very important to be able to recognize the current surgical phase because the phase information can be utilized in various ways both during and after surgery. This paper proposes an efficient phase recognition network, called MomentNet, for cholecystectomy endoscopic videos. Unlike LSTM-based network, MomentNet is based on a multi-stage temporal convolutional network. Besides, to improve the phase prediction accuracy, the proposed method adopts a new loss function to supplement the general cross entropy loss function. The new loss function significantly improves the performance of the phase recognition network by constraining un-desirable phase transition and preventing over-segmentation. In addition, MomnetNet effectively applies positional encoding techniques, which are commonly applied in transformer architectures, to the multi-stage temporal convolution network. By using the positional encoding techniques, MomentNet can provide important temporal context, resulting in higher phase prediction accuracy. Furthermore, the MomentNet applies label smoothing technique to suppress overfitting and replaces the backbone network for feature extraction to further improve the network performance. As a result, the MomentNet achieves 92.31% accuracy in the phase recognition task with the Cholec80 dataset, which is 4.55% higher than that of the baseline architecture.

The generative adversarial network (GAN) has demonstrated superb performance in generating synthetic images in recent studies. However, in the conventional framework of GAN, the maximum resolution of generated images is limited to the resolution of real images that are used as the training set. In this paper, in order to address this limitation, we propose a novel GAN framework using a pre-trained network called evaluator. The proposed model, higher resolution GAN (HRGAN), employs additional up-sampling convolutional layers to generate higher resolution. Then, using the evaluator, an additional target for the training of the generator is introduced to calibrate the generated images to have realistic features. In experiments with the CIFAR-10 and CIFAR-100 datasets, HRGAN successfully generates images of 64 × 64 and 128 × 128 resolutions, while the training sets consist of images of 32 × 32 resolution. In addition, HRGAN outperforms other existing models in terms of the Inception score, one of the conventional methods to evaluate GANs. For instance, in the experiment with CIFAR-10, a HRGAN generating 128 × 128 resolution demonstrates an Inception score of 12.32, outperforming an existing model by 28.6%. Thus, the proposed HRGAN demonstrates the possibility of generating higher resolution than training images.

The development of deep learning has achieved great success in object detection, but small object detection is still a difficult and challenging task in computer vision. To address the problem, we propose an improved single-shot multibox detector (SSD) using enhanced feature map blocks (SSD-EMB). The enhanced feature map block (EMB) consists of attention stream and feature map concatenation stream. The attention stream allows the proposed model to focus on the object regions rather than background owing to channel averaging and the effectiveness of the normalization. The feature map concatenation stream provides additional semantic information to the model without degrading the detection speed. By combining the output of these two streams, the enhanced feature map, which improves the detection of a small object, is generated. Experimental results show that the proposed model has high accuracy in small object detection. The proposed model not only achieves good detection accuracy, but also has a good detection speed. The SSD-EMB achieved a mean average precision (mAP) of 80.4% on the PASCAL VOC 2007 dataset at 30 frames per second on an RTX 2080Ti graphics processing unit, an mAP of 79.9% on the VOC 2012 dataset, and an mAP of 26.6% on the MS COCO dataset.

This article presents a dual-polarized, high gain multi-beam and high T/Rx channel-to-channel isolation antenna module for 24 GHz sensor applications. The proposed antenna is configured to support 2-Tx and 2-Rx channels with a pair of vertically polarized (VP) radiation pattern and a pair of horizontally polarized (HP) radiation pattern. Further, each linearly polarized T/Rx antenna is configured by 2 × 4 array with a multi-layer integrated feed network, resulting in four sets of 2 × 4 array antennas fabricated within a single printed circuit board (PCB). Since multiple RF channels must be ensured with minimal interference, high antenna-to-antenna, including Tx-to-Tx, Rx-to-Rx, and Tx-to-Rx port isolations in the proposed antenna are achieved by multi-layered feed network and four sets of T-shaped magnetic walls. To verify the performance of the proposed structure, a 2-Tx and 2-Rx antenna module was fabricated at 24 GHz. The fabricated antenna showed a measured maximum 10-dB impedance bandwidth of 3.9% with a maximum measured gain of 11.7 dBi, considering both Tx and Rx. Further, the measured channel-to-channel isolations were always better than 35.6 dB at 24 GHz.

When imaging brain vasculature with optical coherence tomography angiography (OCTA), volumetric analysis of cortical vascular networks in OCTA datasets is frequently challenging due to the presence of artifacts, which appear as multiple-scattering tails beneath superficial large vessels in OCTA images. These tails shadow underlying small vessels, making the assessment of vascular morphology in the deep cortex difficult. In this work, we introduce an image processing technique based on mean subtraction of the depth profile that can effectively reduce these tails to better reveal small hidden vessels compared to the current tail removal approach. With the improved vascular image quality, we demonstrate that this simple method can provide better visualization of three-dimensional vascular network topology for quantitative cerebrovascular studies.