Computational Image Analysis Laboratory

The Computational Image Analysis Laboratory is dedicated to advancing cancer research and clinical practice by analyzing cancer imaging phenotypes and integrating genotypes using state-of-the-art image processing, radiomics, and deep learning techniques, with the goal of enhancing clinical decision-making in personalized precision oncology. We focus on developing and automating novel quantitative imaging biomarkers derived from CT, MRI, and PET modalities for a wide range of clinical applications, including (but not limited to) cancer surveillance, diagnosis, prognosis, and early response assessment across all solid tumors.

Key projects

• Response Assessment Platform Integrated with Digital Volumetric Tumor Segmentation (RAPID-VTS)
• Radiomics pipeline developed for analyzing diverse imaging phenotypes
• Clinical validation of volumetry and kinetic modeling in tumor response assessment
• Automated detection and segmentation of solid tumors on CT images
• Automation of RECIST (AutoRECIST)
• Baseline response prediction and early treatment evaluation for oncology clinical trials
• Prostate vision-language model (VLM) to automatically generate radiology reports
• Automated surveillance, diagnosis and response assessment for kidney cancer
• AI-driven prediction of post-surgical recurrence in lung cancer
• Exploring and improving the reproducibility of radiomic features and signatures
• QA programs to ensure the accurate and reproducible quantitative imaging biomarkers
• Data sharing efforts