Back to search

National and Kapodistrian University of Athens

Ανάπτυξη μεθοδολογιών και εργαλείων για την αξιολόγηση, βελτιστοποίηση και διόρθωση των γεωμετρικών παραμορφώσεων και του σήματος εικόνων απεικόνισης μαγνητικού συντονισμού που χρησιμοποιούνται σε εφαρμογές στερεοτακτικής ακτινοχειρουργικής - ακτινοθεραπείας

Abstract

dc:description

The use of Magnetic Resonance Imaging (MRI) is prevalent in contemporary Radiotherapy applications, particularly in Stereotactic Radiosurgery (SRS) treatment planning for intracranial applications. The ability of MR to image soft tissue with superior contrast compared to Computed Tomography (CT), as well as to display multi-contrast images, results in better delineation and tumor characterization capabilities, overall. Especially in cases of brain metastases and other intracranial lesions (e.g., meningiomas or acoustic neuromas), MRI has established itself as the de facto imaging modality for delineating both target(s) and normal tissue.However, MR images suffer from geometrical and signal distortions, which degrade the overall treatment planning result, and may lead to underdosage of the target tumor. Furthermore, (distortions) are a major obstacle to the adoption of 3D dosimetry applications in clinical practice, such as 3D polymer gel dosimetry. Distortion correction can lead to more precise targeting of possible lesions and optimization of 3D dosimetry to obtain more complete dose information with the highest possible accuracy.After presenting the background knowledge on MRI (chapter 1), the second chapter focuses on the available techniques for the analysis and correction of geometric distortions and MR signal inhomogeneities, along with an assessment of the application of the considered methods in optimizing 3D gel polymerization dosimetry.In the first section of the experimental part, an evaluation of geometrical distortions that are independent of the applied MR pulse sequence (i.e., don’t change sign upon a read gradient polarity change) was performed. To evaluate this type of distortions, available phantoms, designed specifically for this purpose, were utilized. The phantoms were properly adjusted for the MR sequences and imaging parameters used clinically for SRS treatment planning, in MRI-based or MRI-only protocols. The first phantom incorporated 1978 holes throughout its volume, in 11 axial planes, which served as Control Points (CPs), while it was specially manufactured to occupy the maximum possible volume within a head coil, to assess distortions more accurately across the entire Field of View (FoV). The phantom was employed for the assessment of MRI manufacturer’s correction scheme performance on gradient non-linearity-induced geometric distortions. The second phantom incorporated 947 holes throughout its volume, on all three anatomical planes (axial, coronal, sagittal), and was used to evaluate residual Sequence Independent (SI) distortions, after having removed Sequence Dependent (SD) distortions. Regarding the latter, considering that the SI geometric distortions are unique to the imaging unit and the applied clinical parameters, and without having enabled the manufacturer’s distortion correction option, SI distortions ranged between 0.54 to 0.71mm on average, while maximum values ranged from 1.2 to 2.1mm. With the activation of the manufacturer’s correction algorithms, there was a significant decrease in both average as well as maximum measured values, where the former ranged from 0.47-0.59mm and the latter from 1.2-1.5mm.For the second part, residual mean SI distortion values ranged from 0.5mm to 0.79mm, while maximum values were up to 2.5mm. The uncertainty of the localization algorithm varied at satisfactory levels, with the MR CP localization uncertainty reaching an average value of (0.07 ± 0.04) mm, and a maximum of 0.17mm, while, correspondingly, the CT equivalent having an average value of (0.04 ± 0.02) mm with a maximum of 0.10mm. The peak distortion values detected were mainly located outside the typical treatment volume, at the edges of the FoV, and, consequently, their effect can be considered clinically negligible. However, distortion outside of the treatable area could still negatively affect the accuracy of the image registration process and potentially compromise treatment efficiency.The next section is dedicated to the evaluation of the geometric distortions that change sign upon read gradient polarity reversal, i.e., SD distortions. To assess this type of distortions, the phantom with the 5 planes was suitably modified, with the construction of two additional cylindrical inserts, which simulated small target volumes inside a brain, to study the spatial deformations induced by the presence of gadolinium pentaacetic acid diethylenetriamine (Gd-DTPA) paramagnetic contrast agent. The inserts were filled with Gd-DTPA solutions at various concentrations (0–20 mM), while, at the same time, reverse gradient polarity and magnetic field mapping techniques were applied, to properly evaluate the various sources of distortions. The position of the phantom inserts was markedly affected in the presence of the contrast agent, with the centroid shift reaching up to 0.068mm mM-1 (0.208ppm mM-1). Corresponding methodology was applied to patient contrast-enhanced images, following gadolinium uptake, including a total of 10 brain metastases. The overall spatial uncertainty in patient images ranged from 0.35mm to 0.87mm, depending on the location of the target tumor, with a mean of 0.54mm (2.24ppm), with Gd-DTPA induced distortion being of the order of 0.5mm for the MRI protocol used, in agreement with the phantom study.In the third section of the experimental part, geometric distortion correction schemes were studied, emphasizing on the SD geometric distortion correction, as the SI distortions can be easily evaluated and corrected using distortion maps, for a specific imaging modality. To this purpose, a 3D-printed head phantom was used, filled with polymer gel, which was irradiated in specific areas, covering the entire phantom volume. Polymerized areas served as metastases equivalents. The evaluated methods were the Field Mapping technique (FM), the Mean Image technique (MI) and the Signal Integration technique (SI). The average central positions of each target identified on the forward and reversed read gradient polarity MR scans served as the reference locations for each lesion. The same techniques and workflows were applied to correct contrast enhanced T1-weighted (T1w) MR images of 10 patients with a total of 27 brain metastases.All methods employed in the phantom study diminished spatial distortion. Median and maximum distortion magnitude decreased from 0.7mm (2.10ppm) and 0.8mm (2.36ppm), respectively, to In the last part of this thesis, correction schemes were applied to optimize a 3D polymer gel dosimetry protocol. Two identical, 3D printed, anthropomorphic head phantoms were employed. The phantoms were filled with 3D polymer gel dosimeter and subjected to CT scanning, following the clinical SRS workflow, while for the irradiation of the first phantom, a treatment plan of 11 small targets (distributed throughout the brain parenchyma) was constructed. MRIs of the second (unirradiated) phantom were used to determine the background R2 distribution in 3D, induced by B1-inhomogeneity. Acquired R2 variation maps were deducted from corresponding R2 maps of the irradiated phantom, yielding a 3D R2-matrix corrected for B1 inhomogeneities. Measured and calculated dose distributions were globally normalized to the center of a target located close to the scanner isocenter and compared in 3D. The table of doses calculated through the Treatment Planning System (TPS) was adopted as reference.By applying B1 inhomogeneity correction, an improvement in the level of agreement (up to 15%) between the measured and calculated dose profiles was observed in most targets, with an obvious severance in the corrected image quality, especially in the parenchymal region. 3D global gamma index passing rates (5%/2mm) were 77% and 92% for the uncorrected and corrected distributions, respectively. The proposed methodology successfully improved dosimetric accuracy in normalized multi-target 3D dosimetry, however, although promising, further refinement is needed for proper integration into clinical routine as multiple image processing steps and expertise in polymer gel dosimetry are required.This work significantly contributes to the understanding of geometrical distortions and signal inhomogeneities in MRI images, while it is also a step towards the potential application of distortion correction methods in clinical practice as well as their integration into 3D dosimetry protocols.

Degree

thesis:*
Grantor dc:publisher
National and Kapodistrian University of Athens
Year dc:date
2023

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Dellios, Dimitrios
  • Δέλλιος, Δημήτριος

Subjects

dc:subject × 16

Rights

Language dc:language
gre

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/54377

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Dellios, Dimitrios; Δέλλιος, Δημήτριος. Ανάπτυξη μεθοδολογιών και εργαλείων για την αξιολόγηση, βελτιστοποίηση και διόρθωση των γεωμετρικών παραμορφώσεων και του σήματος εικόνων απεικόνισης μαγνητικού συντονισμού που χρησιμοποιούνται σε εφαρμογές στερεοτακτικής ακτινοχειρουργικής - ακτινοθεραπείας. National and Kapodistrian University of Athens, 2023. http://hdl.handle.net/10442/hedi/54377