Artificial intelligence is opening up new possibilities for supporting medical diagnosis and healthcare, particularly with advances in computer vision, machine learning, and the growing availability of medical data. However, translating research results into real-world practice remains challenging. These challenges involve not only methods and technologies, but also data, domain knowledge, interaction with users, and collaboration among experts from different disciplines.
This presentation introduces several approaches and research efforts in applying artificial intelligence and intelligent interaction to healthcare, covering problems such as medical image classification, detection, and segmentation; semi-supervised learning for three-dimensional medical imaging data; and computer-assisted analysis of endoscopic, laryngoscopic, and X-ray images.
Through selected examples, the presentation also discusses several issues encountered in the research process, including the role of domain knowledge, the availability and quality of annotated data, interaction and collaboration with domain experts, and the gap between experimental results and practical deployment.
The presented work is not intended to offer complete solutions, but rather to share some possibilities, experiences, and open questions that deserve further investigation. In this spirit, the presentation aims to encourage exchange and mutual learning, while highlighting the importance of interdisciplinary collaboration and contributions from different areas of expertise toward the development of useful, appropriate, and responsible AI technologies for healthcare.