Ghost Artifact Detection and Correction in Diffusion Weighted Images
Gespeichert in:
| Veröffentlicht in: | ProQuest Dissertations and Theses (2025) |
|---|---|
| 1. Verfasser: | |
| Veröffentlicht: |
ProQuest Dissertations & Theses
|
| Schlagworte: | |
| Online-Zugang: | Citation/Abstract Full Text - PDF |
| Tags: |
Keine Tags, Fügen Sie das erste Tag hinzu!
|
MARC
| LEADER | 00000nab a2200000uu 4500 | ||
|---|---|---|---|
| 001 | 3213174684 | ||
| 003 | UK-CbPIL | ||
| 020 | |a 9798315752981 | ||
| 035 | |a 3213174684 | ||
| 045 | 2 | |b d20250101 |b d20251231 | |
| 084 | |a 66569 |2 nlm | ||
| 100 | 1 | |a Thai, Anh Sinh Tram | |
| 245 | 1 | |a Ghost Artifact Detection and Correction in Diffusion Weighted Images | |
| 260 | |b ProQuest Dissertations & Theses |c 2025 | ||
| 513 | |a Dissertation/Thesis | ||
| 520 | 3 | |a Magnetic resonance imaging (MRI) has become an indispensable tool in both clinical diagnoses and biomedical research. MRI systems can generate images with contrasts sensitive to specific biological processes in both function and structure. For instance, functional MRI (fMRI) identifies brain regions that are activated during specific tasks, while diffusion MRI (dMRI) reveals tissue microstructure and composition features. However, the clinical adoption of advanced MRI modalities, particularly dMRI, has been slower than the conventional MRI acquisitions. One major reason is the quantitative nature of dMRI, which is highly sensitive to image quality and prone to reproducibility issues.Like any measurement device, MRI systems have inherent imperfections that can degrade image quality. These imperfections can originate from the subject being imaged, such as excessive subject movement, metallic implants, or system-induced factors, including low signal-to-noise ratio (SNR) and distortions. Both fMRI and dMRI are more vulnerable to these imperfections due to their reliance on Echo Planar Imaging (EPI) protocol for acquisition, which enables rapid acquisition but is prone to distortions and artifacts. Additionally, recent advancements in parallel imaging techniques have further improved acquisition efficiency, enabling faster scans while maintaining high spatial resolution. However, both EPI and parallel imaging acquisition require precise calibration and preprocessing, which are not always fully addressed in practice.This dissertation focuses on detecting and correcting artifacts that persist in EPI images despite calibration, reconstruction, and preprocessing. In many clinical settings, reference scansthat can be used for scanner self-calibration are often unavailable, making the identification of artifacts difficult. To overcome this, my proposed methods rely solely on the magnitude images, excluding raw complex MRI data and reference scans. The approach leverages characteristics of images acquired with different scan parameters, such as phase-encoding directions (PEDs), allowing more effective extraction of correct anatomical information.First, I analyze how image acquisition and correction methods influence artifact characteristics. Factors such as phase-encoding direction, gradient strength, and calibration errors can alter the appearance of artifacts, making them more difficult to distinguish from actual anatomy.Next, I investigate how preprocessing steps can unintentionally modify these artifacts, posing a risk of misinterpretation in clinical and research applications. I then propose methods that can be integrated into existing dMRI preprocessing pipelines to detect and correct artifacts. As a case study, I focus on EPI Nyquist ghost artifacts in dMRI, using simulated data that utilizes Fourier Transform principles and undersampling theory to track the location and behavior of the ghost artifacts under certain constraints throughout the preprocessing pipeline. By mapping the artifacts before any preprocessing, my approach enables better tracking and isolation, improving both artifact understanding and removal.Finally, I validate this framework using both clinical and simulation datasets. Results show a 30% reduction in artifact impact in clinical data and 90% detection accuracy in simulations. This framework is the first to systematically detect and correct residual ghost artifacts in dMRI and has potential applications beyond dMRI. Specifically, the methods can be extended to other EPI-based MRI modalities, such as fMRI, and adapted for additional types of artifacts, including chemical shift artifacts. By enhancing the reliability of MRI data in cases where artifacts are unavoidable, this work ultimately improves the utility of MRI in both neuroscience research and clinical practice. | |
| 653 | |a Medical imaging | ||
| 653 | |a Electrical engineering | ||
| 773 | 0 | |t ProQuest Dissertations and Theses |g (2025) | |
| 786 | 0 | |d ProQuest |t ProQuest Dissertations & Theses Global | |
| 856 | 4 | 1 | |3 Citation/Abstract |u https://www.proquest.com/docview/3213174684/abstract/embedded/L8HZQI7Z43R0LA5T?source=fedsrch |
| 856 | 4 | 0 | |3 Full Text - PDF |u https://www.proquest.com/docview/3213174684/fulltextPDF/embedded/L8HZQI7Z43R0LA5T?source=fedsrch |