Augmented mapping and analysis of pancreatic fat by quantitative MRI
<p>The incidence of pancreas pathology is rising rapidly. While early detection is critical, these are often "silent" conditions that only become symptomatic at a late stage, when they may already be untreatable. Incidental findings, where the target organ is near the pancreas, offer...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | English |
| Published: |
2022
|
| Subjects: |
| _version_ | 1797111865352912896 |
|---|---|
| author | Triay Bagur, A |
| author2 | Bulte, D |
| author_facet | Bulte, D Triay Bagur, A |
| author_sort | Triay Bagur, A |
| collection | OXFORD |
| description | <p>The incidence of pancreas pathology is rising rapidly. While early detection is critical, these are often "silent" conditions that only become symptomatic at a late stage, when they may already be untreatable. Incidental findings, where the target organ is near the pancreas, offer a way to detect pancreas pathology in time. Pancreatic ectopic fat is an early manifestation of disease, and evaluation using MRI is receiving increasing interest. However, it is challenged by the remarkable variability of the pancreas across individuals and its physiological complexity. Quantitative Magnetic Resonance Imaging (MRI) has shown potential for imaging biomarker development, including Proton Density Fat Fraction (PDFF) and T1, though some challenges remain.</p>
<p>The aim of the research from this thesis has been to produce an augmented representation of pancreatic disease state using quantitative MRI, focusing on practical implementation. First, regional quantification of pancreas imaging biomarkers, namely PDFF and T1, is explored through the development of a novel automated pancreas subsegmentation method that employs groupwise registration of UK Biobank data sets. The subsegmentation is validated with expert annotations using direct metrics (Dice overlap) and indirect metrics (PDFF quantification). Second, a novel PDFF reconstruction method is proposed based on chemical shift imaging and field inhomogeneity mapping. The method is robust to artefacts in one data set with challenging B0 field inhomogeneities. Third, semi-automated quantification of pancreas PDFF and T1 is demonstrated in a large population with Long COVID, enabling efficient processing. Finally, image-based quality control of PDFF is implemented through an extension of an error mapping technique previously applied to T1, and validated using simulations and in-vivo data sets. The output error maps are robust, intuitive, and can inform spatial disease heterogeneity and longitudinal change.</p> |
| first_indexed | 2024-03-07T08:16:25Z |
| format | Thesis |
| id | oxford-uuid:f526df4f-3173-4c9e-b9ea-0241a24db896 |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-03-07T08:16:25Z |
| publishDate | 2022 |
| record_format | dspace |
| spelling | oxford-uuid:f526df4f-3173-4c9e-b9ea-0241a24db8962024-01-03T11:48:28ZAugmented mapping and analysis of pancreatic fat by quantitative MRIThesishttp://purl.org/coar/resource_type/c_db06uuid:f526df4f-3173-4c9e-b9ea-0241a24db896Magnetic resonance imagingFatPancreasEnglishHyrax Deposit2022Triay Bagur, ABulte, DBrady, M<p>The incidence of pancreas pathology is rising rapidly. While early detection is critical, these are often "silent" conditions that only become symptomatic at a late stage, when they may already be untreatable. Incidental findings, where the target organ is near the pancreas, offer a way to detect pancreas pathology in time. Pancreatic ectopic fat is an early manifestation of disease, and evaluation using MRI is receiving increasing interest. However, it is challenged by the remarkable variability of the pancreas across individuals and its physiological complexity. Quantitative Magnetic Resonance Imaging (MRI) has shown potential for imaging biomarker development, including Proton Density Fat Fraction (PDFF) and T1, though some challenges remain.</p> <p>The aim of the research from this thesis has been to produce an augmented representation of pancreatic disease state using quantitative MRI, focusing on practical implementation. First, regional quantification of pancreas imaging biomarkers, namely PDFF and T1, is explored through the development of a novel automated pancreas subsegmentation method that employs groupwise registration of UK Biobank data sets. The subsegmentation is validated with expert annotations using direct metrics (Dice overlap) and indirect metrics (PDFF quantification). Second, a novel PDFF reconstruction method is proposed based on chemical shift imaging and field inhomogeneity mapping. The method is robust to artefacts in one data set with challenging B0 field inhomogeneities. Third, semi-automated quantification of pancreas PDFF and T1 is demonstrated in a large population with Long COVID, enabling efficient processing. Finally, image-based quality control of PDFF is implemented through an extension of an error mapping technique previously applied to T1, and validated using simulations and in-vivo data sets. The output error maps are robust, intuitive, and can inform spatial disease heterogeneity and longitudinal change.</p> |
| spellingShingle | Magnetic resonance imaging Fat Pancreas Triay Bagur, A Augmented mapping and analysis of pancreatic fat by quantitative MRI |
| title | Augmented mapping and analysis of pancreatic fat by quantitative MRI |
| title_full | Augmented mapping and analysis of pancreatic fat by quantitative MRI |
| title_fullStr | Augmented mapping and analysis of pancreatic fat by quantitative MRI |
| title_full_unstemmed | Augmented mapping and analysis of pancreatic fat by quantitative MRI |
| title_short | Augmented mapping and analysis of pancreatic fat by quantitative MRI |
| title_sort | augmented mapping and analysis of pancreatic fat by quantitative mri |
| topic | Magnetic resonance imaging Fat Pancreas |
| work_keys_str_mv | AT triaybagura augmentedmappingandanalysisofpancreaticfatbyquantitativemri |