Honors and Awards
2019-2020 UF College of Medicine University Scholar Award Recipient
3D Deformable Image Registration on Planning Lung CT
Background and Significance
The purpose of this project is to quantify the cumulative radiation damage incurred by the healthy tissue around a tumor close to or inside the lung of patients receiving proton therapy. This study is of great importance, as some patients receiving radiation therapy suffer from severe radiation toxicity comparatively due to their high radiosensitivity. While it is important to understand how much radiation a patient can tolerate, it is even more important to know how much radiation the patient is receiving. In practice, the radiation oncologist treating tumors which move significantly over the course of the breathing cycle take a time elapse CT scan of the patient in which ten volumes are produced at each tenth percentile of the breathing cycle. These ten volumes are then averaged, and the treatment plan is produced according to the positions the tumor assumes in the averaged CT volume. Surely, this will lead to a treatment plan in which the tumor is radiated for the entirety of the treatment. However, the death of the tumor cells might be accompanied by an unanticipated accrual of dead or over radiated normal tissue. The tissue response to dose intensity is a sigmoidal curve, so small errors in treatment dose near the steep incline on the tissue response curve do a great deal of unintended damage that might be common place in proton therapy treatments in which the patient is allowed to breathe. The contours of the body vary so greatly over time so that some aspect of this variation is likely to be overlooked resulting in a medical error.
My immediate goal is to compute how much radiation each portion of tissue receives, taking in to consideration the motion of the lung during proton treatment.
I will gather time-elapsed chest CT images of a representative patient (under an existing IRB-approved protocol). My first task will be to calculate how each part of the tissue displaces from one phase of the breathing cycle to the next. To do so, I will use b-spline based 3D deformable image registration. A b-spline is a smooth, continuous curve which connects a set of points. These curves are used to connect the grid points in a 3D grid which overlays the volume at hand. As the volume deforms during the breathing cycle, so do the grid points and splines. 3D deformable image registration uses non-linear transformations, similarity metrics, and an optimization algorithm to match a deformed (moving) image to a fixed image. In doing so, the grid points and splines used to deform the moving image to match with the fixed image give the information necessary to produce a vector field of tissue displacements. These displacements can be used to understand what part of a volume at one phase of the breathing cycle corresponds to a part in a volume at another phase.
Immediate Research Objectives
I will begin by researching currently available open-source solutions in MatLab, such as the xxxx method (). Pairing this information with treatment plans will allow me to compute the total radiation received through the breathing cycle at each part of the volume. With further analysis, I will calculate the total volume which receives an unhealthy dose of radiation and compare this to the expected outcome of the treatment plan.