Kalantzis, Georgios

This research demonstrates that a 3D printed bolus can be customized for electron
radiation therapy. Both extruder and powder based printers were used, along with, paraffin
wax, super stuff, and H20. The plan dose coverage and conformity for the planning target
volume (PTV), was such that the distal side of the PTV was covered by the 90% isodose
line. The structure is read, and converted into an STL file. The file is sent to a slicer to
print. The object was filled with parafin wax, superstuff or water and sealed. Materials
Hounsfield units were analyzed, along with the structure stability. This method is evaluated
by scanning the 3D printed bolus. The dose conformity is improved compared to that with
no bolus. By generating a patient specific 3D printed bolus there is an in improvement in
conformity of the prescription isodose surface while sparing immediately adjacent normal
tissues.

Simulated Annealing algorithm is utilized for Intensity Modulated Radiation Therapy IMRT optimization.
The goal in IMRT is to give the prescribed radiation dose to the tumor while minimizing the dose given
to normal organs.

Intensity modulated radiation therapy (IMRT) is a cancer treatment method in which the intensities of the radiation beams are modulated; therefore these beams have non-uniform radiation intensities. The overall result is the delivery of the prescribed dose in the target volume. The dose distribution is conformal to the shape of the target and minimizes the dose to the nearby critical organs. An inverse planning algorithm is used to obtain those non-uniform beam intensities. In inverse treatment planning, the treatment plan is achieved by using an optimization process. The optimized plan results to a high-quality dose distribution in the planning target volume (PTV), which receives the prescribed dose while the dose that is received by the organs at risk (OARs) is reduced. Accordingly, an objective function has to be defined for the PTV, while some constraints have to be considered to handle the dose limitations for the OARs.

The purpose of this research is to validate various forms of mathematical modeling
of glioblastoma multiforme (GBM) expressed as differential equations, numerically.
The first work was involved in the numerical solution of the reaction-convection
model, efficacy of which is expressed in terms of survival time. It was calculated using
simple numerical scheme for the standard-of-care treatment in clinics which includes
surgery followed by the radiation and chemotherapy. Survival time using all treatment
options increased significantly to 57 weeks compared to that of surgery close
to 14 weeks. It was also observed that survival time increased significantly to 90
weeks if tumor is totally resected. In reaction-diffusion model using simple numerical
scheme, tumor cell density patterns due to variation in patient specific tumor
parameters such as net proliferation rate and diffusion coefficient were computed.
Significant differences were observed in the patterns while using dominant diffusion
and proliferation rate separately. Numerical solution of the tumor growth model
under the anti-angiogenic therapy revealed some impacts in optimum tumor growth
control however it was not significant.

Every scheduled treatment at a radiation therapy clinic involves a series of safety
protocol to ensure the utmost patient care. Despite safety protocol, on a rare occasion
an entirely preventable medical event, an accident, may occur. Delivering a treatment
plan to the wrong patient is preventable, yet still is a clinically documented error.
This research describes a computational method to identify patients with a novel
machine learning technique to combat misadministration.The patient identification
program stores face and fingerprint data for each patient. New, unlabeled data from
those patients are categorized according to the library. The categorization of data by
this face-fingerprint detector is accomplished with new machine learning algorithms
based on Sparse Modeling that have already begun transforming the foundation of
Computer Vision. Previous patient recognition software required special subroutines
for faces and di↵erent tailored subroutines for fingerprints. In this research, the same
exact model is used for both fingerprints and faces, without any additional subroutines
and even without adjusting the two hyperparameters. Sparse modeling is a powerful tool, already shown utility in the areas of super-resolution, denoising, inpainting,
demosaicing, and sub-nyquist sampling, i.e. compressed sensing. Sparse Modeling
is possible because natural images are inherrently sparse in some bases, due to their
inherrant structure. This research chooses datasets of face and fingerprint images to
test the patient identification model. The model stores the images of each dataset as
a basis (library). One image at a time is removed from the library, and is classified by
a sparse code in terms of the remaining library. The Locally Competetive Algorithm,
a truly neural inspired Artificial Neural Network, solves the computationally difficult
task of finding the sparse code for the test image. The components of the sparse
representation vector are summed by `1 pooling, and correct patient identification is
consistently achieved 100% over 1000 trials, when either the face data or fingerprint
data are implemented as a classification basis. The algorithm gets 100% classification
when faces and fingerprints are concatenated into multimodal datasets. This suggests
that 100% patient identification will be achievable in the clinal setting.

Patients receiving Intensity Modulated Radiation Therapy (IMRT) for late stage head and neck (HN) cancer often experience anatomical changes due to weight loss, tumor regression, and positional changes of normal anatomy (1). As a result, the actual dose delivered may vary from the original treatment plan. The purpose of this study was (a) to evaluate the dosimetric consequences of the parotid glands during the course of treatment, and (b) to determine if there would be an optimal timeframe for replanning. Nineteen locally advanced HN cancer patients underwent definitive IMRT. Each patient received an initial computerized tomography simulation (CT-SIM) scan and weekly cone beam computerized tomography (CBCT) scans. A Deformable Image Registration (DIR) was performed between the CT-SIM and CBCT of the parotid glands and Planning Target Volumes (PTVs) using the Eclipse treatment planning system (TPS) and the Velocity deformation software. A recalculation of the dose was performed on the weekly CBCTs using the original monitor units. The parameters for evaluation of our method were: the changes in volume of the PTVs and parotid glands, the dose coverage of the PTVs, the lateral displacement in the Center of Mass (COM), the mean dose, and Normal Tissue Complication Probability (NTCP) of the parotid glands. The studies showed a reduction of the volume in the PTVs and parotids, a medial displacement in COM, and alterations of the mean dose to the parotid glands as compared to the initial plans. Differences were observed for the dose volume coverage of the PTVs and NTCP of the parotid gland values between the initial plan and our proposed method utilizing deformable registration-based dose calculations.