In the pre-procedural planning and guidance of ventricular tachycardia (VT) electroanatomic mapping (EAM) and catheter ablation procedures, the exact location and extent of myocardial scar is important to decide whether the procedure will be epicardial or endocardial, as well as to reduce intervention time. Today, delayed enhanced magnetic resonance imaging (DE-MRI) is considered the imaging gold standard for the assessment of scar tissue. However, multi-detector computed tomography (MDCT) could be an interesting alternative. The main reasons are related to the reduced artefacts caused by the implantable cardioverter-defibrillator (ICD), the higher spatial resolution when compared to DE-MRI with which we can have detailed information about the anatomy of the heart (e.g. trabeculae and coronary arteries) as well as the reliability in visualizing epicardial fat distribution. In the identification of re-entry circuits during an epicardial intervention the knowledge of the location and extent of epicardial fat is useful because it presents voltage characteristics similar to scar tissue and is often confused with the latter. However, fat is often neglected in the ablation procedures because it needs a very time consuming manual segmentation. The purpose of this work was to construct a 3D multi-parametric model of the heart by segmenting automatically ventricular cavities, left myocardium, scar, epicardial fat and coronaries from MDCT images. For the anatomical segmentation a 3D level set algorithm based on a new multi-scale directional stopping function was developed. The stopping function consists in decreasing the image scale space in two steps, letting the level set to expand at each step, preventing in this way over-segmentation. Myocardial scar was accessed by automatic segmentation of delayed enhanced MDCT scan and/or myocardial thinning from the angiographic scan. A thick Fat layer was also assumed to be scar if an epicardial intervention was demanded. This framework was applied to patients with recurrent VT undergoing angiographic and delayed MDCT before electro-anatomic mapping (EAM) and radiofrequency ablation. The accuracy of our model was assessed by comparing it with the manual segmentations of expert radiologists and by findings of EAM constructed by arrhythmologists before radiofrequency ablation. Preliminary results suggest that our method could be integrated into the clinical surgery software system as an effective tool to assist the surgeon during the ablation procedure.
Durante la mappatura elettroanatomica e la procedura di ablazione, l'esatta localizzazione della cicatrice miocardiaca è importante per decidere se la procedura sara epicardiaca o endocardiaca, cosi come per ridurre il tempo d'intervento. Oggi, la risonanza magnetica con mezzo di contrasto è considerato il gold standard per valutare il tessuto miocardiaco. Tuttavia, la TAC potrebbe essere un interessante alternativa. Le principali ragioni sono la riduzione degli artefatti causata dal defibrillatore, la maggiore risoluzione spaziale e l'affidabilità nella visualizzazione del grasso epicardiaco quando confrontata con la risonanza. Nell'identificazione dei circuiti di rientro durante un intervento epicardiaco, la conoscenza della localizzazione del grasso epicardiaco è utile, perche il grasso presenta caratteristiche di voltaggio simili al tessuto cicatriziale ed è spesso confuso con quest'ultimo. Tuttavia, il grasso è spesso trascurato nelle procedure di ablazione perche richiede una onerosa segmentazione manuale. Lo scopo di questo lavoro è stato costruire un modello 3D multi parametrico del cuore, segmentando automaticamente le cavita ventricolari, il miocardio sinistro, la cicatrice, il grasso epicardiaco e le coronarie da immagini TAC. Per la segmentazione anatomica è stato sviluppato un’algoritmo level set basato su un filtro multiscala e direzionale. La cicatrice miocardiaca è stata segmentata analizzando lo scan tardivo e/o l’assottigliamento della parete miocardiaca. Questo approccio è stato applicato su pazienti con tachicardia ventricolare ricorrente. L’accuratezza del nostro modello è stata verificata confrontandolo con le segmentazioni manuali di esperti radiologi e con i risultati della mappa elettroanatomica creata dall’aritmologo durante l’intervento di mappatura. I risultati suggeriscono che il nostro metodo potrebbe essere integrato nel software di ablazione a radiofrequenza come uno strumento efficace per assistere l’aritmologo durante l’intervento.
Patient-specific multi-parametric model of the heart from MDCT images to guide ventricular tachycardia ablation procedures
GONCALVES ANTUNES, SOFIA
Abstract
In the pre-procedural planning and guidance of ventricular tachycardia (VT) electroanatomic mapping (EAM) and catheter ablation procedures, the exact location and extent of myocardial scar is important to decide whether the procedure will be epicardial or endocardial, as well as to reduce intervention time. Today, delayed enhanced magnetic resonance imaging (DE-MRI) is considered the imaging gold standard for the assessment of scar tissue. However, multi-detector computed tomography (MDCT) could be an interesting alternative. The main reasons are related to the reduced artefacts caused by the implantable cardioverter-defibrillator (ICD), the higher spatial resolution when compared to DE-MRI with which we can have detailed information about the anatomy of the heart (e.g. trabeculae and coronary arteries) as well as the reliability in visualizing epicardial fat distribution. In the identification of re-entry circuits during an epicardial intervention the knowledge of the location and extent of epicardial fat is useful because it presents voltage characteristics similar to scar tissue and is often confused with the latter. However, fat is often neglected in the ablation procedures because it needs a very time consuming manual segmentation. The purpose of this work was to construct a 3D multi-parametric model of the heart by segmenting automatically ventricular cavities, left myocardium, scar, epicardial fat and coronaries from MDCT images. For the anatomical segmentation a 3D level set algorithm based on a new multi-scale directional stopping function was developed. The stopping function consists in decreasing the image scale space in two steps, letting the level set to expand at each step, preventing in this way over-segmentation. Myocardial scar was accessed by automatic segmentation of delayed enhanced MDCT scan and/or myocardial thinning from the angiographic scan. A thick Fat layer was also assumed to be scar if an epicardial intervention was demanded. This framework was applied to patients with recurrent VT undergoing angiographic and delayed MDCT before electro-anatomic mapping (EAM) and radiofrequency ablation. The accuracy of our model was assessed by comparing it with the manual segmentations of expert radiologists and by findings of EAM constructed by arrhythmologists before radiofrequency ablation. Preliminary results suggest that our method could be integrated into the clinical surgery software system as an effective tool to assist the surgeon during the ablation procedure.File | Dimensione | Formato | |
---|---|---|---|
2015_02_PhD_Antunes.pdf
accessibile in internet per tutti
Descrizione: Testo della tesi
Dimensione
9.88 MB
Formato
Adobe PDF
|
9.88 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/108582