A novel technique is proposed to characterize lung tissue incompressibility variation during respiration. Estimating lung tissue incompressibility parameter variations resulting from air content variation throughout respiration is critical for computer assisted tumor motion tracking. Continuous tumor motion is a major challenge in lung cancer radiotherapy, especially with external beam radiotherapy. If not accounted for, this motion may lead to areas of radiation overdosage for normal tissue. Given the unavailability of imaging modality that can be used effectively for real-time lung tumor tracking, computer assisted approach based on tissue deformation estimation can be a good alternative. This approach involves lung biomechanical model where its fidelity depends on input tissue properties. This investigation shows that considering variable tissue incompressibility parameter is very important for predicting tumor motion accurately, hence improving the lung radiotherapy outcome.
First, anin silico lung phantom study was conducted to demonstrate the importance of employing variable Poisson's ratio for tumor motion predication. After it was established that modeling this variability is critical for accurate tumor motion prediction, an optimization based technique was developed to estimate lung tissue Poisson's ratio as a function of respiration cycle time. In this technique, the Poisson's ratio and lung pressure value were varied systematically until optimal values were obtained, leading to maximum similarity between acquired and simulated 4D CT lung images. This technique was applied in an ex vivo porcine lung study where simulated images were constructed using the end exhale CT image and deformation fields obtained from the lung's FE modeling of each respiration time increment. To model the tissue, linear elastic and Marlow hyperelastic material models in conjunction with variable Poisson's ratio were used.
The phantom study showed that the tumor motion trajectory and its final locations obtained from simulations with and without considering tissue incompressibility variation were very different. For example, tumor displacements in the z direction were −11.23 and −38.10 mm obtained with the Marlow hyperelastic material model in conjunction with constant and variable Poisson's ratio, respectively. By comparing the acquired 4D-CT image sequence of the porcine lung with their image sequence counterparts obtained from the hyperelastic model with constant and variable Poisson's ratio, it was shown that using variable tissue incompressibility reduced errors significantly in tumor motion prediction.
This investigation demonstrates the importance of incompressibility variation estimation and utilization for accurate tumor tracking in computer assisted lung external beam radiation therapy. An optimization framework was developed to estimate a Poisson's ratio function in terms of respiration cycle time using experimental image data of the lung. Utilizing this function along with respiratory system FE modeling may lead to more effective tumor targeting, hence potentially improving the outcome of lung external beam radiation therapy techniques. This is particularly true for stereotactic body radiation therapy where only one or a few fraction treatments are applied, precluding the possibility of averaging out dosimetric deviations introduced by the respiratory motion.
This research is supported by the Natural Sciences and Engineering Research Council (NSERC) of Canada and Western University. The authors would also like to acknowledge the contributions of Ting-Yim Lee, Gregory Carnes, and Jennifer Hadway who supported this research and helped in 4D-CT image acquisition and reconstruction.
I. INTRODUCTION II. METHOD AND MATERIALS II.A. Numerical phantom study II.B. Optimization algorithm for the Poisson's ratio calculation II.C. Ex vivo porcine lungmodel II.C.1. 4D-CT imaging and 3D model construction II.C.2. Lung FE model II.C.3. Validation III. RESULTS III.A. Numerical lung phantom study III.B. Poisson's ratio function of cyclic time for an ex vivo porcine model III.C. Ex vivo porcine model results validation IV. DISCUSSION AND CONCLUSIONS