Transit Dosimetry for Patient Treatment Verification with an Electronic Portal Imaging Device

Sean Lawrence Berry

Transit Dosimetry for Patient Treatment Verification with an Electronic Portal Imaging Device
Berry, Sean Lawrence
Thesis Advisor(s):
Wuu, Cheng-Shie
Ph.D., Columbia University
Applied Physics and Applied Mathematics
Persistent URL:
\The complex and individualized photon fluence patterns constructed during intensity modulated radiation therapy (IMRT) treatment planning must be verified before they are delivered to the patient. There is a compelling argument for additional verification throughout the course of treatment due to the possibility of data corruption, unintentional modification of the plan parameters, changes in patient anatomy, errors in patient alignment, and even mistakes in identifying the correct patient for treatment. Amorphous silicon (aSi) Electronic Portal Imaging Devices (EPIDs) can be utilized for IMRT verification. The goal of this thesis is to implement EPID transit dosimetry, measurement of the dose at a plane behind the patient during their treatment, within the clinical process. In order to achieve this goal, a number of the EPID's dosimetric shortcomings were studied and subsequently resolved. Portal dose images (PDIs) acquired with an aSi EPID suffer from artifacts related to radiation backscattered asymmetrically from the EPID support structure. This backscatter signal varies as a function of field size (FS) and location on the EPID. Its presence can affect pixel values in the measured PDI by up to 3.6%. Two methods to correct for this artifact are offered: discrete FS specific correction matrices and a single generalized equation. The dosimetric comparison between the measured and predicted through-air dose images for 49 IMRT treatment fields was significantly improved (p << .001) after the application of these FS specific backscatter corrections. The formulation of a transit dosimetry algorithm followed the establishment of the backscatter correction and a confirmation of the EPID's positional stability with linac gantry rotation. A detailed characterization of the attenuation, scatter, and EPID response behind an object in the beam's path is necessary to predict transit PDIs. In order to validate the algorithm's performance, 49 IMRT fields were delivered to a number of homogeneous and heterogeneous slab phantoms. A total of 33 IMRT fields were delivered to an anthropomorphic phantom. On average, 98.1% of the pixels in the dosimetric comparison between the measured and predicted transit dose images passed a 3%/3mm gamma analysis. Further validation of the transit dosimetry algorithm was performed on nine human subjects under an institutional review board (IRB) approved protocol. The algorithm was shown to be feasible for patient treatment verification. Comparison between measured and predicted transit dose images resulted in an average of 89.1% of pixels passing a 5%/3mm gamma analysis. A case study illustrated the important role that EPID transit dosimetry can play in indicating when a treatment delivery is inconsistent with the original plan. The impact of transit dosimetry on the clinical workflow for these nine patients was analyzed to identify improvements that could be made to the procedure in order to ease widespread clinical implementation. EPID transit dosimetry is a worthwhile treatment verification technique that strikes a balance between effectiveness and efficiency. This work, which focused on the removal of backscattered radiation artifacts, verification of the EPID's stability with gantry rotation, and the formulation and validation of a transit dosimetry algorithm, has improved the EPID's dosimetric performance. Future research aimed at online transit verification would maximize the benefit of transit dosimetry and greatly improve patient safety.
Item views
text | xml
Suggested Citation:
Sean Lawrence Berry, , Transit Dosimetry for Patient Treatment Verification with an Electronic Portal Imaging Device, Columbia University Academic Commons, .

Columbia University Libraries | Policies | FAQ