Imaging biomarkers for cardiovascular risk prediction with CT techniques
<p><b>Title:</b> Imaging biomarkers for cardiovascular risk prediction with CT techniques</p> <p><b>Background:</b> Coronary plaque formation, progression and rupture are still the major cause for myocardial infarction. Although many ways for the identifica...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | English |
| Published: |
2022
|
| Subjects: |
| _version_ | 1797108609787625472 |
|---|---|
| author | Kluner, LV |
| author2 | Antoniades, C |
| author_facet | Antoniades, C Kluner, LV |
| author_sort | Kluner, LV |
| collection | OXFORD |
| description | <p><b>Title:</b> Imaging biomarkers for cardiovascular risk prediction with CT techniques</p>
<p><b>Background:</b> Coronary plaque formation, progression and rupture are still the major cause for myocardial infarction. Although many ways for the identification of high-risk plaques have been developed and evaluated over the last decade, they are still not able to forecast the development and rupture of an individual plaque precisely. With the increase of new techniques such as radiomics and the fat attenuation index (FAI), new risk stratification techniques for coronary plaques could be established. </p>
<p><b>Methods:</b> The first part developed an automated coronary plaque segmentation tool for calcified and non-calcified plaque components. The second part compared statistical versus radiomics-based automated plaque classification approaches. These two parts aimed to facilitate an automated coronary plaque analysis on future big cohorts. In the last part, FAI around plaques (pFAI) was analysed and differences between individual coronary plaque categories were compared.</p>
<p><b>Results:</b> After showing excellent results on intra- and interobserver analysis for the manual segmented dataset, a model for calcified plaque components achieved very good results, even outperforming metrics from the interobserver analysis. The radiomics-based approach for non-calcified plaque detection proved the feasibility of this method by showing good initial results. While an automated plaque classification based on two first-order statistical values has already worked well, radiomics were able to even outperform this in terms of accuracy and diversification of plaque types. Measured pFAI revealed significant differences between individual plaque classes based on their degree of calcification.</p>
<p><b>Discussion:</b> Although the size of the manual labeled and segmented datasets was limited, meaningful models for an automated coronary plaque segmentation and classification were developed. Further, it was shown that FAI around different plaque types provides additional information on their inflammatory status and, therefore, eventually on their individual risk, which has to be investigated on plaque-specific outcome data in the future.</p> |
| first_indexed | 2024-03-07T07:31:15Z |
| format | Thesis |
| id | oxford-uuid:ff99e0e3-708b-41c5-bfe1-fbe9e9e1da3e |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-03-07T07:31:15Z |
| publishDate | 2022 |
| record_format | dspace |
| spelling | oxford-uuid:ff99e0e3-708b-41c5-bfe1-fbe9e9e1da3e2023-01-11T13:11:39ZImaging biomarkers for cardiovascular risk prediction with CT techniquesThesishttp://purl.org/coar/resource_type/c_db06uuid:ff99e0e3-708b-41c5-bfe1-fbe9e9e1da3eHeart--ImagingEnglishHyrax Deposit2022Kluner, LVAntoniades, CNeubauer, STomlins, P<p><b>Title:</b> Imaging biomarkers for cardiovascular risk prediction with CT techniques</p> <p><b>Background:</b> Coronary plaque formation, progression and rupture are still the major cause for myocardial infarction. Although many ways for the identification of high-risk plaques have been developed and evaluated over the last decade, they are still not able to forecast the development and rupture of an individual plaque precisely. With the increase of new techniques such as radiomics and the fat attenuation index (FAI), new risk stratification techniques for coronary plaques could be established. </p> <p><b>Methods:</b> The first part developed an automated coronary plaque segmentation tool for calcified and non-calcified plaque components. The second part compared statistical versus radiomics-based automated plaque classification approaches. These two parts aimed to facilitate an automated coronary plaque analysis on future big cohorts. In the last part, FAI around plaques (pFAI) was analysed and differences between individual coronary plaque categories were compared.</p> <p><b>Results:</b> After showing excellent results on intra- and interobserver analysis for the manual segmented dataset, a model for calcified plaque components achieved very good results, even outperforming metrics from the interobserver analysis. The radiomics-based approach for non-calcified plaque detection proved the feasibility of this method by showing good initial results. While an automated plaque classification based on two first-order statistical values has already worked well, radiomics were able to even outperform this in terms of accuracy and diversification of plaque types. Measured pFAI revealed significant differences between individual plaque classes based on their degree of calcification.</p> <p><b>Discussion:</b> Although the size of the manual labeled and segmented datasets was limited, meaningful models for an automated coronary plaque segmentation and classification were developed. Further, it was shown that FAI around different plaque types provides additional information on their inflammatory status and, therefore, eventually on their individual risk, which has to be investigated on plaque-specific outcome data in the future.</p> |
| spellingShingle | Heart--Imaging Kluner, LV Imaging biomarkers for cardiovascular risk prediction with CT techniques |
| title | Imaging biomarkers for cardiovascular risk prediction with CT techniques |
| title_full | Imaging biomarkers for cardiovascular risk prediction with CT techniques |
| title_fullStr | Imaging biomarkers for cardiovascular risk prediction with CT techniques |
| title_full_unstemmed | Imaging biomarkers for cardiovascular risk prediction with CT techniques |
| title_short | Imaging biomarkers for cardiovascular risk prediction with CT techniques |
| title_sort | imaging biomarkers for cardiovascular risk prediction with ct techniques |
| topic | Heart--Imaging |
| work_keys_str_mv | AT klunerlv imagingbiomarkersforcardiovascularriskpredictionwithcttechniques |