CENTER FOR VIRTUAL IMAGING TRIALS

A virtual platform for evaluating medical imaging technologies from design to use

CVIT is a National Center for Biomedical Imaging and Bioengineering supported by the National Institute of Biomedical Imaging and Bioengineering (NIBIB) and operated at Duke University.

Rethinking Risk and Benefit in Medical Imaging: A More Patient-Centric Approach

Francesco Ria and Ehsan Samei recently presented the novel imaging risk framework developed in our laboratory during a featured conversation hosted by Applied Radiology Group. The discussion highlighted a new…

A New Methodology

We oﬀer an eﬃcient methodology for evaluating imaging technologies and applications in silico using computational models simulating the patient, the imaging system, and the image reader.

These models comprise our three technology, research, and development (TR&D) endeavors, where we continue to provide realistic computational models of human anatomy (virtual patients), state-of-the-art CT scanners (virtual scanners), and clinically relevant image analysis (virtual readers).

The TR&Ds work synergistically to oﬀer comprehensive resources and essential training to partners through our various collaborative and service projects as well as training and dissemination eﬀorts.

Recently Shared Software

View All Software

iTrialSpace: Programmable Virtual Lung Lesion Trials
Programmable virtual lesion trial framework that enables controlled, auditable evaluation of chest CT AI models by systematically varying lesion, anatomy, and imaging factors.
AI in Lung Health: Benchmarking Detection and Diagnostic Models
A public, reproducible benchmark for evaluating AI models for lung nodule detection and malignancy classification across multiple datasets using standardized training, validation, and external testing protocols.

Recently Shared Data

View All Data

Duke Lung Cancer Screening (DLCS) Dataset
Large-scale low-dose chest CT screening dataset with nodule annotations based on modern lung cancer screening practices.
XCAT 3.0 Personalized Phantom Library
A large-scale library of over 2,500 anatomically detailed computational phantoms generated from automated CT segmentation for virtual imaging trials and AI evaluation.