2514: Biological Lung Phantom for Image-Guided Proton Therapy Validation

Abstract

To evaluate the feasibility of a biological lung phantom to validate image-guided proton beam therapy. Human lung tissue is a complex, dynamic system consisting of lung parenchyma, blood vessels, and airways, each having different electron densities. Furthermore, this system deforms and moves in a variable manner during normal respiration. In contrast, conventional radiological phantoms (e.g., the Rando® phantom) assume a static, uniform-density lung field. A more realistic phantom system that takes regional density variations and ventilation-induced motion and deformation into account is needed to verify the ability to deliver proton treatment plans to well-defined penetration depths. To this end, swine lung preserved to retain vital mechanical properties was obtained and compared to human lung for suitability as a phantom. The phantom was ventilated and imaged on a GE Lightspeed CT scanner at various lung filling states. Volume rendering of the CT image data using BioImage (SCIRun BioPSE software) was performed to visualize and determine coordinates of airway bifurcations. To demonstrate the residual range variation in proton track length in a dynamic, heterogeneous lung tissue system, a virtual lung tissue model based on lung histology and matched to measured swine lung phantom CT density at different points in the ventilatory cycle was generated. A 100 MeV proton pencil beam traversing the 10 cm thick tissue model was then simulated with the proton beam terminating in a water beamstop to calculate residual range. Preserved swine lung was found to be equivalent to human lung in terms of radiological and physical properties, lobar structure, airway architecture, volume and overall mass, and has been determined to be histologically comparable to human lung by a board-certified pathologist. Analysis of volume rendered CT images yielded a conservative minimum of 31 reproducible anatomic landmarks (airway bifurcations) evenly distributed throughout the lung. Residual ranges varying from 3 and 76 mm in the post-model water beamstop were calculated (mean of 39 mm, standard deviation of 30 mm), dependent on the breathing phase of the lung phantom. The preserved swine lung lasted approximately 8 weeks before degrading, which may be extended through improved preservation techniques. Simulation results suggest that respiratory motion not only compromises the therapeutic dose delivery to a moving target, but can result in significant excess dose delivery to critical structures downstream of the target volume (e.g., spinal cord, esophagus) through variations in residual range. Advantages of the dynamically ventilated preserved swine lung phantom for proton depth-dose studies with a proton range telescope include realistic geometry, motion, deformation, and the ability to implant radio-opaque clips and tumor surrogates. Disadvantages include compromises in true lung deformation due to external ventilation instead of physiological respiration, as well as the inevitable degradation of the system. We believe that this dynamic biological lung phantom will be useful as a verification tool for image-guided proton therapy, in that it will allow for the planning, scanning, and delivery of proton dose to a reproducible, complex, spatially and temporally variable system.