AI pathology Lab
Projects
2022 SKIN PATH AI
A learning dataset was created for a total of 30,000 sheets of diseases that need to be distinguished from Seoul National University Hospital, Catholic University, National Cancer Center, Samsung Seoul Hospital, and six normal skin tumors in the 2022 NIA project. In model validation for dataset validation, we also demonstrate high performance above AUC 0.97.
2021 STANBAI -
Non-Gynecology Cytology AI
-
2021 CodiPAI -DIP-OPTIMUS
1) Panorgan pancancer classification AI
2) AI for RCC
3) GI cancer MSI prediction AI
In 2021, it was chosen for the development of a digital pathology-based cancer-specialized AI analysis solution in the KHIDI project, and a total of 100,000 slides and metadata were collected from 12 university hospitals nationwide over a 5-year period for a total of 20 types of cancer. We created a multi-organ pathology AI model, a comprehensive kidney cancer pathology diagnosis system, and a gastrointestinal cancer MSI molecular mutation and prognosis prediction system using the collected data sets. Currently, five SCI-level papers have been published, and two more are in the works, as well as six patents for image preprocessing and pathology AI-related technologies (PCT/KR2021/017402, etc.).
2021 DEEP:URO -
AI for Urine Cytology
-
It was chosen for an NRF project in 2021 and is currently developing a cytopathology image analysis and diagnosis system.
-
5000 urine samples in total The entirety of each slide is gathered, annotated, and quality checked.
-
Using the CLAM-MIL algorithm, AUC was 0.99 at the whole slide level and accuracy was 0.97, resulting in significantly better performance than conventional pathologists' 50-70% accuracy.
2018 DEEP:CRC -
Pathology AI for CRC Bx
In 2018, the Korean SGER project created an artificial intelligence diagnostic system for colon cancer based on 20,000 images of biopsy tissue pathology, published related papers, registered intellectual property (C-2020-008022), and finished one patent application. It led all stages of the development of an artificial intelligence pathology image diagnostic system, and it will proceed with approval from the Ministry of Food and Drug Safety in the form of 3rd party software currently running on INFINITT Digital Pathology Solution in the first half of this year (DPS).
2016 ImmunoGenius -
ML-Based IHC Interpretation Mobile app
The 2016 Korean SGER project created an artificial intelligence diagnostic estimation system as a mobile application to analyze immunostaining data, demonstrated high diagnostic accuracy in 1,000 lymphoma and unknown tumor data (up to 94.7%), published a paper in CIKM, one of the five major academic societies in computer science, and published two SCIE papers. It has also registered intellectual property rights (C-2018-0007646) and is promoting it for easy use following its release on Google Android and Apple App Store. We gained extensive experience in the processes of underlying research, technology commercialization, and technology commercialization for software development available in the actual clinical field as a result of this project.
2022 Student Pathology AI Study
Pathology Image Analysis Using Artificial intelligence
-
A knowledge of pathology
- Basic familiarity with pathology
- Knowledge of tumors
-
Understanding Artificial Intelligence
- Artificial Intelligence Foundations and History
- Machine learning and deep learning types and understanding
- Knowledge of CNN
-
Experience with Group Projects
- Developing research abilities through collaborative research
-
Writing research papers and attending conferences experience