We are proud to share that our Virtual Lung Screening Trial (VLST) has been published in Medical Image Analysis, a major milestone for the CVIT and the field of medical imaging research.
Why this matters
Clinical imaging trials are critical to advancing medical innovation but are often expensive, time-consuming, and ethically constrained. As the national center for virtual imaging trials, CVIT is pioneering ways to overcome these challenges using Virtual Imaging Trials (VITs), computer-based simulations that replicate key elements of clinical studies in a controlled, risk-free environment.
The VLST is a landmark achievement: it is the first-ever demonstration of a virtual platform successfully replicating a major national clinical trial, the National Lung Screening Trial (NLST), entirely in silico.
About the study
The VLST was designed to assess whether a virtual trial could emulate key outcomes of the NLST, which compared Computed Tomography (CT) and Chest Radiography (CXR) for lung cancer screening. Using simulated cancerous lung nodules, we created a cohort of 294 virtual patients modeled with anatomically realistic XCAT human phantoms. Each patient underwent both CT and CXR exams, which were interpreted by AI-based virtual readers: the AI CT-Reader and the AI CXR-Reader.
Our primary goal was to compare the diagnostic accuracy of CT and CXR, measured by the Area Under the Curve (AUC). The AI CT-Reader significantly outperformed the AI CXR-Reader, achieving an AUC of 0.92 (95% CI: 0.90–0.95) versus 0.72 (95% CI: 0.67–0.77) for CXR. Notably, at the same 94% CT sensitivity reported in the NLST, the VLST specificity of 73% closely matched the 73.4% specificity found in the NLST.
These findings reinforce that virtual trials can effectively replicate critical trends observed in real-world studies, offering a powerful new approach to evaluating imaging technologies.
Why it is a breakthrough for CVIT and the field
The VLST is a prime example of how virtual trials can integrate:
- Complex human models
- Flexible scanner simulations
- Robust AI readers
…into a fully functional, end-to-end research pipeline. This unlocks the potential to investigate imaging performance across a wide range of clinical scenarios, without the constraints or risks of live patient trials.
At CVIT, this success marks a major step forward in our goal to make virtual trials a core pillar of evidence generation in medical imaging. It demonstrates that our platform is capable not only of mirroring existing clinical outcomes, but also of enabling faster, safer, and more scalable innovation in diagnostic imaging.
“Virtual trials are no longer a concept of the future—they are an actionable platform of today.”
A new era in imaging science
In conclusion, the VLST underscores the transformative value of virtual imaging trials in advancing evidence-based medicine. As interest in CT and CXR continues to grow, both due to their diagnostic utility and high clinical volume, tools like the VLST offer a cost-effective, ethical, and scientifically rigorous alternative to traditional studies.
We are excited to continue pushing the boundaries of what is possible in imaging science, and to share this milestone with our partners, collaborators, and the broader medical community.
Read the full article in Medical Image Analysis.
Explore more resources.
Watch the VLST Workflow video.
