Accounting for respiration motion during imaging helps improve targeting precision in radiation therapy. Respiratory motion can be a major source of error in determining the position of thoracic and upper abdominal tumor targets during radiotherapy. Thus, extracting respiratory motion is a key task in radiation therapy planning. Respiration-correlated or four-dimensional CT (4DCT) imaging techniques have been recently integrated into imaging systems for verifying tumor position during treatment and managing respiration-induced tissue motion. The quality of the 4D reconstructed volumes is highly affected by the respiratory signal extracted and the phase sorting method used. This thesis is divided into two parts. In the first part, two image-based respiratory signal extraction methods are proposed and evaluated. Those methods are able to extract the respiratory signals from CBCT images without using external sources, implanted markers or even dependence on any structure in the images such as the diaphragm. The first method, called Local Intensity Feature Tracking (LIFT), extracts the respiratory signal depending on feature points extracted and tracked through the sequence of projections. The second method, called Intensity Flow Dimensionality Reduction (IFDR), detects the respiration signal by computing the optical flow motion of every pixel in each pair of adjacent projections. Then, the motion variance in the optical flow dataset is extracted using linear and non-linear dimensionality reduction techniques to represent a respiratory signal. Experiments conducted on clinical datasets showed that the respiratory signal was successfully extracted using both proposed methods and it correlates well with standard respiratory signals such as diaphragm position and the internal markers’ signal. In the second part of this thesis, 4D-CBCT reconstruction based on different phase sorting techniques is studied. The quality of the 4D reconstructed images is evaluated and compared for different phase sorting methods such as internal markers, external markers and image-based methods (LIFT and IFDR). Also, a method for generating additional projections to be used in 4D-CBCT reconstruction is proposed to reduce the artifacts that result when reconstructing from an insufficient number of projections. Experimental results showed that the feasibility of the proposed method in recovering the edges and reducing the streak artifacts.
Identifer | oai:union.ndltd.org:vcu.edu/oai:scholarscompass.vcu.edu:etd-1495 |
Date | 02 May 2013 |
Creators | Dhou, Salam |
Publisher | VCU Scholars Compass |
Source Sets | Virginia Commonwealth University |
Detected Language | English |
Type | text |
Format | application/pdf |
Source | Theses and Dissertations |
Rights | © The Author |
Page generated in 0.0024 seconds