Virtual Phantoms

This folder contains 5 patient-derived virtual phantoms. Virtual phantoms were developed from patient scan data to provide flexible, programmable, and quantiative ground truths for reconstruction testing while maintaining realistic models of patient phyisology and bolus propogation.Parameters of pharmacokinetic models were fit from patient data and used with patient images to define virtual phantoms. These phantoms were implemented as a stored set of parameters and evaluation functions and are designed as inputs to a virtual scan simulator also developed for this project. Code for the virtual scanner can be found, along with a demo, at https://github.com/tyo8/ECA_Demo .The phantoms used in this experiment modeled noisy acquisition with signal from three different classes of sources; vessel, lesion, and background. Vessel and lesion signal components come from distinct models but a single procedure; all signal components are additive.Vessels and lesions were segmented from patient images and perfusion maps were created from ultrafast data, using a flow constraint method developed by Chengyue Wu at the University of Texas (reference).In addition to the virtual phantoms, each MATLAB file in this folder contains a minimal complete set of inputs to run the phantom through the virtual scanner, including virtual MRI acquisition parameters (e.g., TR and TE) and an acquisition sequence.