Quantitative imaging biomarker for cardiovascular disease: cross-center radiomics for coronary stenosis and flow analysis in atrial fibrillation

Alinezhad, Lida (2026) Quantitative imaging biomarker for cardiovascular disease: cross-center radiomics for coronary stenosis and flow analysis in atrial fibrillation, [Dissertation thesis], Alma Mater Studiorum Università di Bologna. Dottorato di ricerca in Ingegneria biomedica, elettrica e dei sistemi, 38 Ciclo.
Documenti full-text disponibili:
[thumbnail of PhD Thesis Report Lida Alinezhad_Final Submission.pdf] Documento PDF (English) - Richiede un lettore di PDF come Xpdf o Adobe Acrobat Reader
Disponibile con Licenza: Salvo eventuali più ampie autorizzazioni dell'autore, la tesi può essere liberamente consultata e può essere effettuato il salvataggio e la stampa di una copia per fini strettamente personali di studio, di ricerca e di insegnamento, con espresso divieto di qualunque utilizzo direttamente o indirettamente commerciale. Ogni altro diritto sul materiale è riservato.
Download (3MB)

Abstract

Cardiovascular disease remains the leading cause of mortality worldwide. Modern cardiovascular imaging provides anatomical and functional information for diagnosis, risk assessment, and treatment planning. Quantitative imaging methods are increasingly needed to complement visual interpretation, which is reader-dependent and may vary across centers. Radiomics addresses this need by converting acquired medical images into standardized, mineable features that can reveal patterns beyond visual assessment and support diagnosis, risk stratification, and outcome prediction. However, applying radiomics in multi-center cardiovascular studies presents challenges. Scanner- and protocol-dependent variability may introduce non-biological differences that obscure biological signals. Automated and reproducible segmentation of cardiovascular structures, particularly vessels, remains technically demanding, while the small cohorts common in cardiac imaging limit model robustness and increase the risk of overfitting. This thesis presents a quantitative imaging pipeline applied to two cardiovascular scenarios. Its main contribution is a cross-center radiomics framework for predicting coronary stenosis from computed tomography angiography (CTA). The framework integrates deep learning-based coronary artery segmentation; extraction of Image Biomarker Standardisation Initiative (IBSI)-compliant features with ComBat harmonization; nested leave-one-center-out classification with predefined primary-model selection; and parallel evaluation of four classifier families. Model performance is assessed through discrimination, calibration, and threshold-based operating points. A proof-of-concept extension applies radiomics to computational fluid dynamics (CFD)-derived hemodynamic maps of the left atrium in atrial fibrillation, demonstrating its potential for quantitative analysis of complex flow patterns. The work follows methodological recommendations for multi-center radiomics studies, including the METhodological RadiomICs Score (METRICS) and Radiomics Quality Score (RQS), through transparent comparison of supplementary models and explicit reporting of negative and positive findings. Rather than presenting a clinically deployable tool, it provides a worked example of developing cross-center quantitative imaging biomarkers while defining their conditions of use. The main limitations are the two-center design and small atrial fibrillation cohort, requiring external validation before clinical translation.

Abstract
Tipologia del documento
Tesi di dottorato
Autore
Alinezhad, Lida
Supervisore
Co-supervisore
Dottorato di ricerca
Ciclo
38
Coordinatore
Settore disciplinare
Settore concorsuale
Parole chiave
cardiovascular imaging, coronary CT angiography, radiomics, quantitative imaging biomarkers, multi-center harmonization, ComBat, deep learning segmentation, cross-center validation, calibration, atrial fibrillation, computational fluid dynamics, left atrial hemodynamics.
Data di discussione
28 Settembre 2026
URI

Altri metadati

Statistica sui download

Gestione del documento: Visualizza la tesi

^