의료인공지능연구실

● We address the challenge of limited annotated data in medical image segmentation by applying unsupervised domain adaptation through cross-modality image translation. ● We generate post-operative MRI from pre-operative scans to help neurosurgeons estimate potential outcomes before surgery, enabling more effective surgical and treatment planning. ● Our research addresses the challenges of noise and artifacts in rs-fMRI data by proposing a comprehensive deep learning framework that integrates spatial, spectral, and temporal features to enhance the detection of noise-related components. ● We focus on survival prediction in medical imaging by combining multimodal MRI with clinical or genetic data. This allows us to capture both spatial tumor characteristics and biological context, leading to more accurate and personalized predictions. Advance multimodal learning in Medical AI by creating VLMs that seamlessly fuse medical imaging, physiological signals, clinical text, and structured knowledge. Tackle real-world clinical tasks such as diagnosis, prognosis, explanation, question answering, and more. Deliver robust, clinically deployable AI that supports healthcare professionals and ultimately improves patient care. ● Mitigating Hallucinations in Medical VLMs for Medical VQA. ● We improve trustworthiness in medical VLMs by mitigating diverse hallucination types in complex medical reasoning tasks through clinical-aware image-text alignment. ● We develop optimization methods based on Direct Preference Optimization (DPO), RLH/AIF to align model outputs with clinical image-text evidence. ● We seek to enhance the factual consistency and clinical validity of medical VLM outputs by grounding them in case-based and knowledge-based evidence. ● We develop a retrieval-augmented generatio