A5025485089- Cell Image Analysis Techniques
- AI in cancer detection
- Computer Graphics and Visualization Techniques
- Image Processing Techniques and Applications
- 3D Shape Modeling and Analysis
- Advanced Electron Microscopy Techniques and Applications
- Radiomics and Machine Learning in Medical Imaging
- Advanced Numerical Analysis Techniques
- Advanced Neural Network Applications
- Image and Signal Denoising Methods
- Advanced MRI Techniques and Applications
- Sparse and Compressive Sensing Techniques
- Advanced Fluorescence Microscopy Techniques
- Medical Image Segmentation Techniques
- Advanced Vision and Imaging
- Medical Imaging Techniques and Applications
- Data Visualization and Analytics
- Digital Imaging for Blood Diseases
- Cloud Computing and Resource Management
- Functional Brain Connectivity Studies
- Photoacoustic and Ultrasonic Imaging
- Parallel Computing and Optimization Techniques
- Scientific Computing and Data Management
- Advanced Data Compression Techniques
- Tensor decomposition and applications
Korea University
2020-2025
Boramae Medical Center
2020-2023
Seoul National University
2020-2023
Korea Institute of Science and Technology
2023
Ulsan National Institute of Science and Technology
2011-2021
Seoul Metropolitan Government
2020
Institute of Science and Technology
2017
University of Utah
2004-2011
Harvard University
2009-2011
Harvard University Press
2010-2011
Compressed Sensing MRI (CS-MRI) has provided theoretical foundations upon which the time-consuming acquisition process can be accelerated. However, it primarily relies on iterative numerical solvers still hinders their adaptation in time-critical applications. In addition, recent advances deep neural networks have shown potential computer vision and image processing, but to reconstruction is an early stage. this paper, we propose a novel learning-based generative adversarial model,...
Cellular-resolution connectomics is an ambitious research direction with the goal of generating comprehensive brain connectivity maps using high-throughput, nano-scale electron microscopy. One main challenges in developing scalable image analysis algorithms that require minimal user intervention. Deep learning has provided exceptional performance classification tasks computer vision, leading to a recent explosion popularity. Similarly, its application connectomic analyses holds great...
This paper presents DXR, a toolkit for building immersive data visualizations based on the Unity development platform. Over past years, in augmented and virtual reality (AR, VR) have been emerging as promising medium sense-making beyond desktop. However, creating remains challenging, often require complex low-level programming tedious manual encoding of attributes to geometric visual properties. These can hinder iterative idea-to-prototype process, especially developers without experience 3D...
In this paper we propose a novel computational technique to solve the Eikonal equation efficiently on parallel architectures. The proposed method manages list of active nodes and iteratively updates solutions those until they converge. Nodes are added or removed from based convergence measure, but management does not entail an extra burden expensive ordered data structures special updating sequences. has suboptimal worst-case performance but, in practice, real synthetic datasets, runs faster...
Pathology Artificial Intelligence Platform (PAIP) is a free research platform in support of pathological artificial intelligence (AI). The main goal the to construct high-quality pathology learning data set that will allow greater accessibility. PAIP Liver Cancer Segmentation Challenge, organized conjunction with Medical Image Computing and Computer Assisted Intervention Society (MICCAI 2019), first image analysis challenge apply datasets. was evaluate new existing algorithms for automated...
This paper presents the first volume visualization system that scales to petascale volumes imaged as a continuous stream of high-resolution electron microscopy images. Our architecture dense, anisotropic because it: (1) decouples construction 3D multi-resolution representation required for from data acquisition, and (2) sample access time during ray-casting size hierarchy. is designed around scalable virtual memory handles missing naturally, does not pre-compute any such an octree, can...
This paper presents a programmable digital finite-impulse response (FIR) filter for high-performance and low-power applications. The architecture is based on computation sharing multiplier (CSHM) which specifically targets re-use in vector-scalar products can be effectively used the low-complexity FIR design. Efficient circuit-level techniques, namely new carry-select adder conditional capture flip-flop (CCFF), are also to further improve power performance. A 10-tap was implemented...
Data sets imaged with modern electron microscopes can range from tens of terabytes to about one petabyte. Two new tools, Ssecrett and NeuroTrace, support interactive exploration analysis large-scale optical-and electron-microscopy images help scientists reconstruct complex neural circuits the mammalian nervous system.
With the advent of advances in self-supervised learning, paired clean-noisy data are no longer required deep learning-based image denoising. However, existing blind denoising methods still require assumption with regard to noise characteristics, such as zero-mean distribution and pixel-wise noise-signal independence; this hinders wide adaptation method medical domain. On other hand, unpaired learning can overcome limitations related on which makes it more feasible for collecting training...
Abstract Mapping neural connections within the brain has been a fundamental goal in neuroscience to understand better its functions and changes that follow aging diseases. Developments imaging technology, such as microscopy labeling tools, have allowed researchers visualize this connectivity through high-resolution brain-wide imaging. With this, image processing analysis become more crucial. However, despite wealth of images generated, access an integrated pipeline process these data is...
We demonstrate a self-homodyne detection method to stabilize continuous-wave 1550-nm laser 1-km optical fiber delay line, achieving frequency instability of 6.3 × 10 −15 at 16-ms averaging time. This result, limited by thermal noise, is achieved without the need for vacuum system, highlighting potential our approach ultra-stable systems in non-laboratory environments. The system utilizes only few passive optic components and single balanced photodetector, significantly simplifying...
Very long baseline interferometry (VLBI) enables high-angular-resolution observations in astronomy and geodesy by synthesizing a virtual telescope with baselines spanning hundreds to thousands of kilometres. Achieving high instrumental phase stability VLBI relies on the generation high-quality, atomic-referenced RF local oscillator (LO) RF-comb signals for effective downconversion celestial precise calibration, respectively. As observing frequencies move into higher ranges wider bandwidth,...
Immersive analytics is gaining attention across multiple domains due to its capability facilitate intuitive data analysis in expansive environments through user interaction with data. However, creating immersive systems for specific tasks challenging the need programming expertise and significant development effort. Despite introduction of various visualization authoring toolkits, domain experts still face hurdles adopting into their workflow, particularly when faced dynamically changing...
PurposeBiliary tract cancer, also known as intrahepatic cholangiocarcinoma (IHCC), is a rare disease that shows no clear symptoms during its early stage, but prognosis depends highly on the cancer subtype. Hence, an accurate subtype classification model necessary to provide better treatment plans patients and reduce mortality. However, annotating histopathology images at pixel or patch level time-consuming labor-intensive for giga-pixel whole slide images. To address this problem, we propose...
In this paper we present a method to compute and visualize volumetric white matter connectivity in diffusion tensor magnetic resonance imaging (DT-MRI) using Hamilton-Jacobi (H-J) solver on the GPU (Graphics Processing Unit). Paths through volume are assigned costs that lower if they consistent with preferred directions. The proposed finds set of voxels DTI contain paths between two regions whose within threshold optimal path. result is path analysis, which driven by clinical scientific...
Recent advances in scanning technology provide high resolution EM (Electron Microscopy) datasets that allow neuro-scientists to reconstruct complex neural connections a nervous system. However, due the enormous size and complexity of resulting data, segmentation visualization processes data is usually difficult very time-consuming task. In this paper, we present NeuroTrace, novel volume system consists two parts: semi-automatic multiphase level set with 3D tracking for reconstruction...
This paper presents an efficient, fine-grained parallel algorithm for solving the Eikonal equation on triangular meshes. The equation, and broader class of Hamilton–Jacobi equations to which it belongs, have a wide range applications from geometric optics seismology biological modeling analysis geometry images. ability solve such accurately efficiently provides new capabilities exploring visualizing parameter spaces inverse problems that rely in forward model. Efficient solvers...
Sand mining, among the many activities that have significant effects on bed changes of rivers, has increased in parts world recent decades. Numerical modeling plays a vital role simulation long term; however, computational time remains challenge. In this paper, we propose sand mining component integrated into bedload continuity equation and combine it with high-performance computing using graphics processing units to boost speed simulation. The developed numerical model is applied Mekong...
We study the use of neural network algorithms in surface reconstruction from an unorganized point cloud, and meshing implicit surface. found that for such applications, most suitable type networks is a modified version growing cell structure we propose here. The algorithm works by sampling randomly target space, usually cloud or surface, adjusting accordingly network. adjustment includes connectivity Doing several experiments gives satisfactory results some challenging situations involving...
Recent advances in high-resolution microscopy let neuroscientists acquire neural-tissue volume data of extremely large sizes. However, the tremendous resolution and high complexity neural structures present big challenges to storage, processing, visualization at interactive rates. A proposed system provides exploration petascale (petavoxel) volumes resulting from high-throughput electron streams. The can concurrently handle multiple support simultaneous voxel segmentation data. Its...