Research
Research Overview
The iCARE Lab develops artificial intelligence and data-driven methods for medical imaging and healthcare. Our work focuses on extracting reliable and clinically meaningful information from complex medical data, particularly three-dimensional CT and CT angiography (CTA) images.
We combine medical image analysis, computer vision, machine learning, and reproducible data science to develop accurate, transparent, and practical methods that support healthcare research and clinical decision-making.
3D Medical Image Analysis
We develop computational methods for analyzing volumetric CT and CTA data, with particular interest in cardiovascular and aortic imaging. This work includes:
- Image enhancement and structure preservation
- Organ and vascular segmentation
- Centerline extraction and anatomical modeling
- Quantitative analysis of three-dimensional anatomy
Our goal is to transform complex imaging volumes into reproducible measurements that support disease characterization, longitudinal assessment, and clinical research.
Agentic and Multimodal AI
We investigate intelligent systems that combine medical images, clinical information, and specialized computational tools within coordinated workflows. Our work includes:
- Multimodal reasoning across imaging and clinical data
- Agent-based orchestration of segmentation, measurement, and reporting tasks
- Standardized analysis pipelines for complex aortic disease
- Human-in-the-loop methods for transparent and reproducible analysis
Our goal is to develop AI systems that organize complex analytical tasks while keeping their outputs traceable, verifiable, and clinically meaningful.
Reliable and Reproducible AI
We develop methods and computational practices that support trustworthy healthcare AI. Our work emphasizes:
- Transparent model evaluation and reporting
- Reproducible preprocessing, training, and validation workflows
- Robustness and generalization across datasets and institutions
- Open and reusable research software where data-use agreements permit
Our goal is to ensure that research findings can be examined, reproduced, and extended by other researchers.