Efficient Annotation Deep Learning Techniques for Medical Image Analysis

Time: 2026-01-29 08:10:00

Topic: Efficient Annotation Deep Learning Techniques for Medical Image Analysis
Speaker: Yanwei Chen, Professor (Invited by: Ye Qi)    Ritsumeikan University (Japan)
Time: January 29  10:30--11:30

Location: Room 111, West Building, School of Mathematics

Abstract:

    In recent years, deep learning (DL) has played a pivotal role across numerous academic and industrial domains, particularly in computer vision and image recognition. In medical image analysis, DL has also emerged as a powerful tool for automating critical tasks, such as lesion detection, segmentation, and classification. However, prevailing methods predominantly rely on supervised learning paradigms that necessitate large-scale, expert-annotated datasets—an approach that is both cost-prohibitive and impractical in real-world clinical settings. This presentation will focus on label-efficient deep learning algorithms designed to overcome these limitations. Specifically, it will explore techniques such as semi-supervised learning, knowledge (language)-guided learning, and foundation model fine-tuning, aiming to alleviate the annotation burden while maintaining high accuracy and robustness in medical image analysis. In this talk, I will present our recent advances in label-efficient deep learning for enhancing medical image analysis, with a primary focus on three research directions: (1) semi-supervised learning; (2) knowledge (language)-guided learning; and (3) foundation model fine-tuning.