National and Kapodistrian University of Athens
Development of quality control protocol and software tool for the correction of geometric distortions and signal inhomogeneities in nagnetic resonance images
Abstract
dc:descriptionMagnetic Resonance Imaging (MRI) is widely used in advanced radiotherapy applications and, especially, in stereotactic radiosurgery (SRS) treatment planning for intracranial applications. This is justified by the superior soft tissue contrast it exhibits as compared to Computed Tomography (CT) and its multi-contrast capability, which result in better tumor delineation and characterization. Especially for brain lesions, MRI has been established as the imaging modality of choice for both target and normal tissue delineation. This choice, however, comes at the expense of geometric accuracy since it is well known that MR images are geometrically distorted. Following an analysis of the underlying theoretical background, in the second part of this thesis an evaluation of the MR-related geometric distortions is performed. To this end, a prototype phantom was designed and constructed to facilitate distortion characterization for the MR pulse sequences and imaging parameters clinically employed in SRS treatment planning, in MRI-based or MRI-only protocols. The phantom incorporates 947 Control Points (CPs) and was designed to accurately fit in a typical head coil, as well as the Leksell stereotactic frame, used for patient immobilization in SRS applications. System-related distortions were characterized both with and without the presence of the frame. In the absence of the frame and following compensation for field inhomogeneities, measured average CP displacement owing to gradient nonlinearities was 0.53 mm. In presence of the frame, contrarily, detected distortion was greatly increased (up to about 5 mm) in the vicinity of the frame base due to eddy currents induced in the closed loop of its aluminum material. Although the region with the maximum observed distortion may not lie within the SRS treatable volume, frame-related distortion was obliterated at approximately 90 mm from the frame base. Severe distortions observed outside the treatable volume could possibly impinge on the delivery accuracy mainly by adversely affecting the registration process (e.g., the position of the lower part of the N-shaped fiducials used to define the stereotactic space may be miss-registered). System-related distortion was also identified in patient MR images. Using corresponding CT angiography images as a reference, an offset of 1.1 mm was detected for two vessels lying in close proximity to the frame base, while excellent spatial agreement was observed for a vessel far apart from the frame base. The same phantom was scanned at 1.5 and 3.0T and using three clinical MR imaging protocols for SRS treatment planning. B0 inhomogeneity and gradient nonlinearity related geometric distortions were assessed in this study. Areas of increased distortion were identified at the edges of the imaged volume which was comparable to a brain scan. Although mean absolute distortion did not exceed 0.5 mm on any spatial axis, maximum detected CP displacement reached 2 mm. Furthermore, the phantom was modified to incorporate two cylindrical inserts, simulating small brain metastases. The inserts were filled with various concentrations (0-20 mM) of Gd-DTPA (commonly administered in cranial SRS) in order to characterize contrast agent induced distortion. The reversed read gradient polarity was combined with the field mapping technique to distinguish between sources of distortion. Contrast agent was found to significantly affect insert position, with the centroid offset reaching on average 0.067 mm/mM (0.204 ppm/mM). Following Gd-DTPA administration, patient MR images involving a total of 10 brain metastases/targets were also studied using a similar methodology. Total target localization uncertainty was on average 0.54 mm (2.24 ppm) with the Gd-DTPA induced distortion being of the order of 0.5 mm for the MRI protocol used, in agreement with the phantom study. In an effort to establish what could be considered as acceptable geometric uncertainty, highly conformal plans were utilized to simulate irradiation of targets of different diameters (5 to 50 mm). The targets were deliberately mispositioned by 0.5 up to 3 mm. Dose Volume Histograms (DVHs) and plan quality indices clinically used for plan evaluation and acceptance were derived and used to investigate the effect of geometrical uncertainty (distortion) on dose delivery accuracy and plan quality. The latter was found to be strongly dependent on target size. For targets less than 20 mm in diameter, a spatial displacement of the order of 1 mm could significantly affect (>5%) plan acceptance/quality indices. For targets with diameter greater than 2 cm the corresponding displacement was found greater than 1.5 mm. In the last part of this thesis, distortion correction schemes were developed and/or evaluated. In specific, the efficacy of vendor-supplied distortion correction algorithms (accounting for gradient nonlinearity only) was initially assessed for a variety of scanners, following development of an advanced version of the prototype phantom for high-resolution distortion detection, incorporating nearly 2000 CPs. Moreover, the novel average-image distortion correction methodology was developed and evaluated in both phantom and patient studies. The proposed technique is based on read gradient polarity reversal and, therefore, requires two MR scans. In specific, a new image is created after averaging the signal intensities of corresponding forward and reversed polarity images, on a pixel-by-pixel basis. The method was found efficient for sequence dependent distortion minimization. Furthermore, a comparison study was also conducted involving the more well-established signal integration method. All necessary custom routines were developed in-house. Both distortion correction techniques perform equally well, minimizing the mean and median residual distortions. However, the signal integration method requires a few hours of post-imaging computational time while the average-image method is simple and efficient.
Degree
thesis:*- Grantor dc:publisher
- National and Kapodistrian University of Athens
- Year dc:date
- 2018
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Pappas, Eleftherios
- Παππάς, Ελευθέριος
Subjects
dc:subject × 16- Μαγνητική τομογραφία
- ΑΜΣ
- Παραμόρφωση εικόνας
- Ανομοιογένεια μαγνητικού πεδίου
- Γαδολίνιο
- Στερεοτακτική ακτινοχειρουργική
- Magnetic resonance
- MRI
- Image distortion
- Magnetic field inhomogeneity
- Gadolinium
- Stereotactic radiosurgery
- Ιατρική και Επιστήμες Υγείας
- Ιατρική Βιοτεχνολογία
- Medical and Health Sciences
- Medical Biotechnology
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/43142