Developed deep learning-based detection and segmentation models for interventional instruments such as guidewires, catheters, and vessels used in coronary interventions.
Implemented models including U-Net, DeepLabV3+, YOLOv8 using PyTorch and OpenCV.
Created labeled datasets: over 5,000 bounding box annotations and 1,000 segmentation masks using real clinical data.
Developed an automated coronary vessel recognition algorithm for contrast-enhanced fluoroscopic images.
Built 2D Coronary Roadmap algorithms to align guidewire traces with vessel maps and compensate for cardiac motion.
Developed a direction prediction algorithm for guidewire navigation and successfully validated autonomous control through in-vivo experiments using porcine coronary models.
Integrated real-time guidance with robotic systems for commercialization.
AI Research Intern
LNRobotics
01.2022 - 08.2022
Annotated more than 5,000 images with bounding boxes for guidewire detection.
Participated in early-stage AI model development and evaluation.
Refined labeling protocols with clinical expert input.
Preprocessed medical image datasets for training pipelines.
Education
University of Ulsan - Biomedical Engineering
University of Ulsan
04.2001 - 02.2023
Skills
Python
PyTorch
OpenCV
NumPy
SciPy
Skimage
Matplotlib
Git
Ubuntu
Windows
Jupyter
QT-based UI prototypes
Data Experience
Labeled over 5,000 bounding boxes and 1,000 vessel segmentation masks from clinical data.
Managed training/validation datasets and real-time pipelines.
Applied data augmentation methods to overcome the limited availability of medical data.
Cross Functional Collaboration
Collaborated with interventional cardiologists for clinical relevance.
Worked with hardware, electrical, and regulatory teams to support product development.
Phone
+82, 10-9757-1572
Patent
A DEVICE FOR CONTROLLING A SURGICAL ROBOT THAT CONTROLS THE MOVEMENT OF A GUIDE WIRE, Registration in progress
APPARATUS AND METHOD FOR TRAINING A MACHINE LEARNING MODEL FOR PREDICTING GUIDEWIRE GEOMETRY AND DIRECTION OF TRAVE, Registration in progress