2026년 4월 5일
After years of collaborative research between MAILab and Seoul Asan Medical Center, our latest study has been published in Scientific Reports:
“Real-time deep learning interpretation of echocardiographic video for automated detection of anatomical features associated with Tetralogy of Fallot in pediatric patients: a feasibility study”
🔹 Authors: Mi Jin Kim, Jeong Jin Yu, Seulgi Cha, Jae Suk Baek, Dongha Yang & Yeon Jin Jang
🔹 Read the full article here: https://www.nature.com/articles/s41598-026-45943-x
Tetralogy of Fallot (TOF) is a complex congenital heart defect that requires precise and timely diagnosis. With a global shortage of pediatric cardiologists, early detection can be challenging.
Our study demonstrates how AI and deep learning can assist in automating TOF detection from echocardiographic videos:
✅ Trained on 174 pediatric patients (2018–2023)
✅ Used Detectron2 and Mask R-CNN for feature-level detection
✅ Achieved AUC ≈ 1.0 and F1 score 96.8%, with over 97% accuracy across videos
This approach has the potential to:
Improve diagnostic accessibility in regions lacking specialists
Reduce pediatric cardiologists’ workload
Serve as a model for other congenital heart diseases
A big thank you to the entire team at MAILab and Seoul Asan Medical Center for their dedication and collaboration. This is a step forward in making life-saving diagnostics faster, more precise, and more widely accessible.