{"id":5754,"date":"2022-08-05T06:27:42","date_gmt":"2022-08-04T21:27:42","guid":{"rendered":"http:\/\/hvcl.korea.ac.kr\/?page_id=5754"},"modified":"2026-02-04T13:46:30","modified_gmt":"2026-02-04T04:46:30","slug":"research_connect","status":"publish","type":"page","link":"https:\/\/hvcl.korea.ac.kr\/?page_id=5754","title":{"rendered":"Research"},"content":{"rendered":"\n<script>\nfunction link_to_publication(category){\n  document.location=\"https:\/\/hvcl.korea.ac.kr\/?page_id=5770&category=\"+category; \n\/\/  location.replace(\"https:\/\/hvcl.korea.ac.kr\/?page_id=5770&category=\"+category);\n}\n<\/script>\n<!-- resize -->\n<script>\nwindow.addEventListener(\"resize\", handlePageSize);\nfunction handlePageSize() {\n    const item_list=document.getElementsByClassName(\"research-category-body-row\");\n    for(let j=0;j<item_list.length;j++){\n        if(window.innerWidth<800){\n            item_list[j].style.display=\"block\";\n        }\n        else{\n            item_list[j].style.display=\"flex\";\n        }\n    }\n    const card_list=document.getElementsByClassName(\"research-category-card\");\n    for(let j=0;j<card_list.length;j++){\n        if(window.innerWidth<800){\n            card_list[j].style.width=\"90%\";\n        }\n        else{\n            card_list[j].style.width=\"calc(33% - 40px)\";\n        }\n    }\n    const highlight_list=document.getElementsByClassName(\"research-highlight-container\");\n    for(let j=0;j<highlight_list.length;j++){\n        if(window.innerWidth<800){\n            highlight_list[j].style.display=\"block\";\n        }\n        else{\n            highlight_list[j].style.display=\"flex\";\n        }\n    }\n    const highlight_left_list=document.getElementsByClassName(\"research-highlight-left\");\n    for(let j=0;j<highlight_left_list.length;j++){\n        if(window.innerWidth<800){\n            highlight_left_list[j].style.width=\"95%\";\n        }\n        else{\n            highlight_left_list[j].style.width=\"40%\";\n        }\n    }\n    const highlight_right_list=document.getElementsByClassName(\"research-highlight-right\");\n    for(let j=0;j<highlight_right_list.length;j++){\n        if(window.innerWidth<800){\n            highlight_right_list[j].style.width=\"95%\";\n        }\n        else{\n            highlight_right_list[j].style.width=\"60%\";\n        }\n    }\n}\n<\/script>\n  <meta charset=\"utf-8\">\n  <meta name=\"viewport\" content=\"width=device-width, initial-scale=1\">\n  <link rel=\"stylesheet\" href=\"https:\/\/maxcdn.bootstrapcdn.com\/bootstrap\/3.4.1\/css\/bootstrap.min.css\">\n  <script src=\"https:\/\/ajax.googleapis.com\/ajax\/libs\/jquery\/3.6.0\/jquery.min.js\"><\/script>\n  <script src=\"https:\/\/maxcdn.bootstrapcdn.com\/bootstrap\/3.4.1\/js\/bootstrap.min.js\"><\/script>\n<div class=\"container full-contents\">\n  <div id=\"slider\" class=\"carousel slide\" data-ride=\"carousel\">\n    <!-- Indicators -->\n    <ol class=\"carousel-indicators\">\n      <li data-target=\"#slider\" data-slide-to=\"0\" class=\"active\"><\/li>\n      <li data-target=\"#slider\" data-slide-to=\"1\"><\/li>\n      <li data-target=\"#slider\" data-slide-to=\"2\"><\/li>\n<li data-target=\"#slider\" data-slide-to=\"3\"><\/li>\n<li data-target=\"#slider\" data-slide-to=\"4\"><\/li>\n    <\/ol>\n    <!-- Wrapper for slides -->\n    <div class=\"carousel-inner\">\n<div class=\"item active\">\n                <div class=\"container research-highlight-container\">\n                    <div class=\"research-highlight-left\">\n                        <img decoding=\"async\" src=\"https:\/\/hvcl.korea.ac.kr\/AAAI2026.png?_t=1763960142\">\n                    <\/div>\n                    <div class=\"research-highlight-right\">\n                                <p class=\"research-highlight-title\">Paper accepted to AAAI 2026!<\/p>\n                                <p class=\"research-highlight-subtitle\">Virtual Multiplex Staining for Histological Images using a Marker-wise Conditioned Diffusion Model<\/p>\n                                <p class=\"research-highlight-body\">\n                          In this work, we present a virtual multiplex staining framework that synthesizes up to 18 immunofluorescence marker images directly from standard H&amp;E histology. Our conditional diffusion model shares a single architecture across markers with marker-wise one-hot conditioning and pixel-level fine-tuning, achieving state-of-the-art accuracy and color fidelity on the HEMIT and Orion-CRC datasets.\n                                <\/p>\n                    <\/div>\n                <\/div>\n      <\/div>\n<div class=\"item\">\n                <div class=\"container research-highlight-container\">\n                    <div class=\"research-highlight-left\">\n                        <img decoding=\"async\" src=\"https:\/\/hvcl.korea.ac.kr\/wp-content\/papercite-data\/images\/kr_miccai_image-removebg-preview.png?_t=1721024040\">\n                    <\/div>\n                    <div class=\"research-highlight-right\">\n                                <p class=\"research-highlight-title\">Paper accepted to MICCAI 2024!<\/p>\n                                <p class=\"research-highlight-subtitle\">Reference-free Axial Super-resolution of 3D Microscopy Images using Implicit Neural Representation with a 2D Diffusion Prior<\/p>\n                                <p class=\"research-highlight-body\">\n                          In this project, we reconstruct anisotropic micoroscopy (x8, x10) images by leverageing implicit neural representation and diffusion models. Our method optimizes a continuous volumetric representation from low-resolution axial slices, using a 2D diffusion prior trained on high-resolution lateral slices without requiring isotropic volumes.\n                                <\/p>\n                    <\/div>\n                <\/div>\n      <\/div>\n<div class=\"item\">\n                <div class=\"container research-highlight-container\">\n                    <div class=\"research-highlight-left\">\n                        <img decoding=\"async\" src=\"https:\/\/hvcl.korea.ac.kr\/wp-content\/papercite-data\/images\/sk_miccai_image-removebg-preview.png?_t=1721024040\">\n                    <\/div>\n                    <div class=\"research-highlight-right\">\n                                <p class=\"research-highlight-title\">Paper accepted to MICCAI 2024!<\/p>\n                                <p class=\"research-highlight-subtitle\">Clinical-grade Multi-Organ Pathology Report Generation for Multi-scale Whole Slide Images via a Semantically Guided Medical Text Foundation Model<\/p>\n                                <p class=\"research-highlight-body\">\n                           We propose a novel Patient-level Multi-organ Pathology Report Generation (PMPRG) model, which uti- lizes the multi-scale WSI features from our proposed multi-scale regional vision transformer (MR-ViT) model and their real pathology reports to guide VLM training for accurate pathology report generation. The model then automatically generates a report based on the provided key features- attended regional features.\n                                <\/p>\n                    <\/div>\n                <\/div>\n      <\/div>\n<div class=\"item\">\n                <div class=\"container research-highlight-container\">\n                    <div class=\"research-highlight-left\">\n                        <img decoding=\"async\" src=\"https:\/\/hvcl.korea.ac.kr\/ECCV2024_-_framework-removebg-preview.png?_t=1721613208\">\n                    <\/div>\n                    <div class=\"research-highlight-right\">\n                                <p class=\"research-highlight-title\">Paper accepted to ECCV 2024!<\/p>\n                                <p class=\"research-highlight-subtitle\">Co-synthesis of Histopathology Nuclei Image-Label Pairs using a Context-Conditioned Joint Diffusion Model<\/p>\n                                <p class=\"research-highlight-body\">\nWe propose a novel framework for co-synthesizing histopathology nuclei images and paired semantic labels using a context-conditioned joint diffusion model. By incorporating spatial and structural context information, our method generates high-quality synthetic data that improves nuclei segmentation and classification performance.                             <\/p>\n                    <\/div>\n               <\/div>      \n        <\/div>\n      <div class=\"item\">\n                <div class=\"container research-highlight-container\">\n                    <div class=\"research-highlight-left\">\n                        <img decoding=\"async\" src=\"https:\/\/hvcl.korea.ac.kr\/MICCAI2024_-_framework-removebg-preview.png?_t=1721613208\">\n                    <\/div>\n                    <div class=\"research-highlight-right\">\n                                <p class=\"research-highlight-title\">Paper accepted to MICCAI 2024!<\/p>\n                                <p class=\"research-highlight-subtitle\">Controllable and Efficient Multi-Class Pathology Nuclei Data Augmentation using Text-Conditioned Diffusion Models<\/p>\n                                <p class=\"research-highlight-body\">\n                           We propose a two-stage framework for multi-class nuclei data augmentation using text-conditional diffusion models. First, we generate semantic labels and instance maps with a joint diffusion model conditioned by text prompts. Next, we use a latent diffusion model to create high-quality pathology images matching these labels. The method is validated on large pathology nuclei datasets.\n                                <\/p>\n                    <\/div>\n                <\/div>\n      <\/div>\n \n    <\/div>\n    <!-- Left and right controls -->\n    <a class=\"left carousel-control\" href=\"#slider\" data-slide=\"prev\">\n      <span class=\"glyphicon glyphicon-chevron-left\" aria-hidden=\"true\"><\/span>\n      <span class=\"sr-only\">Previous<\/span>\n    <\/a>\n    <a class=\"right carousel-control\" href=\"#slider\" data-slide=\"next\">\n      <span class=\"glyphicon glyphicon-chevron-right\"><\/span>\n      <span class=\"sr-only\">Next<\/span>\n    <\/a>\n  <\/div>\n<\/div>\n<hr>\n<section class=\"research-category\">\n    <div id=\"research-category-header\">\n        <p class=\"text-research-main-title\"> Computer Vision <\/p>\n        <p class=\"text-research-main-body\"> Computer Vision is field of artificial intelligence that extracts useful information from images, videos and other visual inputs. Especially, we focus on computer vision for medical imaging which generates meaningful information or insight through enhancing, classifying, segmenting the images in medical domain. Computer vision in medical domain have brought significant development in simplifying the complex and challenging tasks.<\/p>\n    <\/div>\n    <div class=\"research-category-body-row\">\n        <button class=\"research-category-card\" onclick=\"link_to_publication('Biomedical_Imaging')\">\n            <div id=\"research-category-card-top\">\n                        <img decoding=\"async\" src=\"https:\/\/hvcl.korea.ac.kr\/research\/assets\/img\/compressed.png\">\n            <\/div>\n            <div id=\"research-category-card-bottom\">\n                        <p class=\"text-research-card-title\"> Biomedical Imaging <\/p>\n                        <p class=\"text-research-card-body\">Biomedical Imaging is deriving quantitative information form biomedical images.<\/p>\n            <\/div>\n        <\/button>\n        <button class=\"research-category-card\" onclick=\"link_to_publication('Image_Processing')\">\n            <div id=\"research-category-card-top\">\n                        <img decoding=\"async\" src=\"https:\/\/hvcl.korea.ac.kr\/research\/assets\/img\/highlight1.png\">\n            <\/div>\n            <div id=\"research-category-card-bottom\">\n                        <p class=\"text-research-card-title\"> Image Processing <\/p>\n                        <p class=\"text-research-card-body\">Image Processing is the process of converting an image into a digital form and operating to obtain useful information from it. Recognition, sharpening and restoration, pattern recognition is mainly dealt.<\/p>\n            <\/div>\n        <\/button>\n        <button class=\"research-category-card\" onclick=\"link_to_publication('Histopathology')\">\n            <div id=\"research-category-card-top\">\n                        <img decoding=\"async\" src=\"https:\/\/hvcl.korea.ac.kr\/research\/assets\/img\/histopathology.png\">\n            <\/div>\n            <div id=\"research-category-card-bottom\">\n                        <p class=\"text-research-card-title\"> Histopathology <\/p>\n                        <p class=\"text-research-card-body\"> Histopathology is using deep learning to assist digital analysis of tissue and cell examination under microscope by classification, regression and segmenatation. <\/p>\n            <\/div>\n        <\/button>\n    <\/div>\n    <div class=\"research-category-body-row\">\n        <button class=\"research-category-card\" onclick=\"link_to_publication('Deep_Learning')\">\n            <div id=\"research-category-card-top\">\n                        <img decoding=\"async\" src=\"https:\/\/hvcl.korea.ac.kr\/research\/assets\/img\/deep_learning.png\">\n            <\/div>\n            <div id=\"research-category-card-bottom\">\n                        <p class=\"text-research-card-title\"> Deep Learning <\/p>\n                        <p class=\"text-research-card-body\"> Interactive Deep Learning is designing the algorithms and intelligent user interface that enables machine learning with the intervention with human interaction. <\/p>\n            <\/div>\n        <\/button>\n    <\/div>\n<\/section>\n<hr>\n<section class=\"research-category\">\n    <div id=\"research-category-header\">\n        <p class=\"text-research-main-title\"> Visualization <\/p>\n        <p class=\"text-research-main-body\">Visualization is any technique for creating images, diagrams, or animations to communicate a message. Visualization through visual imagery has been an effective way to communicate both abstract and concrete ideas since the dawn of humanity. <\/p>\n    <\/div>\n    <div class=\"research-category-body-row\">\n        <button class=\"research-category-card\" onclick=\"link_to_publication('Scientific_Visualization')\">\n            <div id=\"research-category-card-top\">\n                    <img decoding=\"async\" src=\"https:\/\/hvcl.korea.ac.kr\/research\/assets\/img\/bio_imag.jpg\">\n            <\/div>\n            <div id=\"research-category-card-bottom\">\n                        <p class=\"text-research-card-title\"> Scientific Visualization <\/p>\n                        <p class=\"text-research-card-body\">Scientific visualization is the process of representing raw, scientific data as images and providing external aid to improve interpretation of large data sets. <\/p>\n            <\/div>\n        <\/button>\n        <button class=\"research-category-card\" onclick=\"link_to_publication('Information_Visualization')\">\n            <div id=\"research-category-card-top\">\n                    <img decoding=\"async\" src=\"https:\/\/hvcl.korea.ac.kr\/research\/assets\/img\/visual_analytics.png\">\n            <\/div>\n            <div id=\"research-category-card-bottom\">\n                        <p class=\"text-research-card-title\"> Information Visualization <\/p>\n                        <p class=\"text-research-card-body\">Information visualization is representing the data in a visual, informative ways to convey information and enhance interpretation of the given data.<\/p>\n            <\/div>\n        <\/button>\n        <button class=\"research-category-card\" onclick=\"link_to_publication('Visual_Analytics')\">\n            <div id=\"research-category-card-top\">\n                    <img decoding=\"async\" src=\"https:\/\/hvcl.korea.ac.kr\/research\/assets\/img\/dxplorer_web_teaser.png\">\n            <\/div>\n            <div id=\"research-category-card-bottom\">\n                        <p class=\"text-research-card-title\"> Visual Analytics <\/p>\n                        <p class=\"text-research-card-body\">Visual Analytics is combination of visualization of the data, human factor, and data analysis to gain knowledge from data with human interactive exploration. <\/p>\n            <\/div>\n        <\/button>\n    <\/div>\n    <div class=\"research-category-body-row\">\n        <button class=\"research-category-card\" onclick=\"link_to_publication('VR\/AR')\">\n            <div id=\"research-category-card-top\">\n                    <img decoding=\"async\" src=\"https:\/\/hvcl.korea.ac.kr\/research\/assets\/img\/VRAR.jpg\">\n            <\/div>\n            <div id=\"research-category-card-bottom\">\n                        <p class=\"text-research-card-title\"> VR\/AR <\/p>\n                        <p class=\"text-research-card-body\"> VR\/AR researches is enabling highly immersive experience in VR or AR environment by studying the interaction or perceptions. Especially, our lab focuses on \nthe use of VR\/AR in medical domain by creating VR\/AR environment suitable for specific activities such as surgical navigation, neuron tracing, or etc. We also seek to establish next-generation mixed reality workspace by studying the utilization of the VR\/AR environment.\n<\/p>\n            <\/div>\n        <\/button>\n        <button class=\"research-category-card\" onclick=\"link_to_publication('Biomedical_Visualization')\">\n            <div id=\"research-category-card-top\">\n                    <img decoding=\"async\" src=\"https:\/\/hvcl.korea.ac.kr\/research\/assets\/img\/bio_vis.png\">\n            <\/div>\n            <div id=\"research-category-card-bottom\">\n                        <p class=\"text-research-card-title\"> Biomedical Visualization <\/p>\n                        <p class=\"text-research-card-body\">Biomedical Visualization is a multidisciplinary field that draws upon and integrates subject matter from a variety of disciplines (e.g. anatomy, biochemistry, genetics, molecular and cell biology, neuroscience, physiology, and surgery, as well as art, graphic design, animation, and computer science).<\/p>\n            <\/div>\n        <\/button>\n        <button class=\"research-category-card\" onclick=\"link_to_publication('Connectomics')\">\n            <div id=\"research-category-card-top\">\n                    <img decoding=\"async\" src=\"https:\/\/hvcl.korea.ac.kr\/research\/assets\/img\/connecto.png\">\n            <\/div>\n            <div id=\"research-category-card-bottom\">\n                        <p class=\"text-research-card-title\"> Connectomics <\/p>\n                        <p class=\"text-research-card-body\">Connectomics is the study of the brain\u2019s structural and functional connections between cells, which is visualized as a connectome. The connectome is a map of all neural connections in a brain and connectomics is the mapping of these connections. Connectomics can provide insights about the brain and many incurable diseases that are associated with it.\n<\/p>\n            <\/div>\n        <\/button>    \n    <\/div>\n<\/section>\n<hr>\n<section class=\"research-category\">\n    <div id=\"research-category-header\">\n        <p class=\"text-research-main-title\"> Parallel and Distributed Computing<\/p>\n        <p class=\"text-research-main-body\"> Distributed computing is often used in tandem with parallel computing. Parallel computing on a single computer uses multiple processors to process tasks in parallel, whereas distributed parallel computing uses multiple computing devices to process those tasks.\n<\/p>\n    <\/div>\n    <div class=\"research-category-body-row\">\n        <button class=\"research-category-card\" onclick=\"link_to_publication('Parallel_Computing\/GPGPU')\">\n            <div id=\"research-category-card-top\">\n                    <img decoding=\"async\" src=\"https:\/\/hvcl.korea.ac.kr\/research\/assets\/img\/parallel.png\">\n            <\/div>\n            <div id=\"research-category-card-bottom\">\n                        <p class=\"text-research-card-title\"> Parallel Computing <\/p>\n                        <p class=\"text-research-card-body\">We deal with increasing the performance of visualization tasks through parallel computing.<\/p>\n            <\/div>\n        <\/button>\n        <button class=\"research-category-card\" onclick=\"link_to_publication('Parallel_Computing\/GPGPU')\">\n            <div id=\"research-category-card-top\">\n                    <img decoding=\"async\" src=\"https:\/\/hvcl.korea.ac.kr\/wp-content\/papercite-data\/images\/sisc21.PNG\">\n            <\/div>\n            <div id=\"research-category-card-bottom\">\n                        <p class=\"text-research-card-title\"> GPGPU <\/p>\n                        <p class=\"text-research-card-body\">General-purpose computing on graphics processing units (GPGPU) is the use of a graphics processing unit (GPU) to perform computation in applications traditionally handled by the central processing unit (CPU).<\/p>\n            <\/div>\n        <\/button>\n    <\/div>\n<\/section>\n<script>\nhandlePageSize();\n<\/script>\n<!-- regacy -->\n<!-- embed type=\"text\/html\" src=\"https:\/\/hvcl.korea.ac.kr\/research\" style=\"width:100%; height:3500px\" -->\n\n\n\n<p><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Paper accepted to AAAI 2026! Virtual Multiplex Staining for Histological Images using a Marker-wise Conditioned Diffusion Model In this work, we present a virtual multiplex staining framework that synthesizes up to 18 immunofluorescence marker images directly from standard H&amp;E histology. Our conditional diffusion model shares a single architecture across markers with marker-wise one-hot conditioning and <a class=\"read-more\" href=\"https:\/\/hvcl.korea.ac.kr\/?page_id=5754\">Read More<\/a><\/p>\n","protected":false},"author":8,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":[],"_links":{"self":[{"href":"https:\/\/hvcl.korea.ac.kr\/index.php?rest_route=\/wp\/v2\/pages\/5754"}],"collection":[{"href":"https:\/\/hvcl.korea.ac.kr\/index.php?rest_route=\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/hvcl.korea.ac.kr\/index.php?rest_route=\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/hvcl.korea.ac.kr\/index.php?rest_route=\/wp\/v2\/users\/8"}],"replies":[{"embeddable":true,"href":"https:\/\/hvcl.korea.ac.kr\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=5754"}],"version-history":[{"count":117,"href":"https:\/\/hvcl.korea.ac.kr\/index.php?rest_route=\/wp\/v2\/pages\/5754\/revisions"}],"predecessor-version":[{"id":7835,"href":"https:\/\/hvcl.korea.ac.kr\/index.php?rest_route=\/wp\/v2\/pages\/5754\/revisions\/7835"}],"wp:attachment":[{"href":"https:\/\/hvcl.korea.ac.kr\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=5754"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}