The Ice/Agar Phantom:  The development of a realistic phantom designed to 
verify the accuracy of a computer algorithm that calculates the volume of 
the human cerebral ventricular system based on CT images 

J. Evans, N. Linney, M. Schmidt

Abstract
Radiologists currently diagnose shunt blockage in hydrocephalus patients by 
subjective visual scrutiny of CT images, which is complicated by the complexity 
of the ventricular system and by differences in scanning angles across sessions.  
The current research focused on developing an agar phantom that is adequately 
comparable to a real brain, to be used to verify the accuracy of a computer 
algorithm designed to calculate the volume of the ventricular system based on CT 
images.  The phantoms developed consisted of a water-filled space contained within 
an agar body.  Both a simple and a complex phantom were developed, neither of 
which had the water contained within an unrealistic plastic or rubber membrane, 
and both of which used agar and water, which approximated the densities and 
textures of real brain tissue and cerebral spinal fluid, respectively.  
The complex model came near in structural complexity to the human ventricular 
system.  Verification of the algorithm using various models with known 
ventricular volumes produced accuracies with a percent error all below 5.68%, 
with an average of 2.36%.  Future research should use the models to determine 
the smallest detectable change that the algorithm can detect in order to 
continue to assess the applicability of the algorithm in a clinical setting.