Auflistung nach Autor:in "Xie, Jingyang"

Gerade angezeigt 1 - 1 von 1

Ventricular tachycardia target registration and cardiac motion estimation for stereotactic arrhythmia radioablation
(2025) Xie, Jingyang
Ventricular tachycardia (VT) is a severe life-threatening arrhythmia originating in the ventricles, potentially causing sudden cardiac death. Stereotactic arrhythmia radioablation (STAR) is a novel and non-invasive bailout treatment option for refractory VT. The paramount goal of STAR is to precisely deliver focused high-dose radiation beams to treat the VT targets in the heart ventricles while minimizing exposure to the surrounding organs at risk. As a novel therapeutic approach, STAR presents several challenges including VT target transfer from the electroanatomical mapping (EAM) system to the radiation treatment planning system (TPS), as well as cardiac motion estimation of the cardiac clinical target volume (CTV) and American Heart Association (AHA) 17 segments of the left ventricle (LV) for motion management. On the one hand, unlike typical radiotherapy tumors that are easily identifiable in computed tomography (CT) scans, the VT substrate is primarily characterized by its electrophysiological properties which are typically determined through an electrophysiological study using a three-dimensional (3D) EAM procedure. Studies show that freehand delineation of the VT target region on the treatment planning CT, as defined on the EAM surface, has poor inter-observer consistency, even among experienced electrophysiologists and radiation oncologists. Additionally, clinically relevant errors have been reported with such freehand VT target registration method. Therefore, practical VT target registration methods are crucial for accurately transferring the VT target from the EAM system to treatment planning imaging data. On the other hand, as a moving organ, the heart contains respiratory and cardiac motion. During STAR treatments, respiratory motion can be effectively managed with gating, deep inspiration breath-hold or robotic tracking techniques. However, cardiac motion, particularly the movement of the cardiac CTV and the 17 LV segments, poses a significant challenge in the precise definition of cardiac internal target volume (ITV). This movement can lead to misalignments, which may reduce STAR treatment effectiveness and increase the risk of harm to nearby organs at risk via dose wash-out. Estimating cardiac motion is essential for defining an appropriate cardiac ITV margin, thereby enhancing the effectiveness of STAR and patient outcomes. The aim of this dissertation is to investigate four main aspects in the field of STAR: (1) practical methods for VT target registration, (2) validation of these methods using real-world VT patient data, (3) accuracy assessment of target registration methods in the absence of ground truth, and (4) a patient- and segment-specific cardiac motion estimation method. To address (1), a software was developed, which includes three practical semi-automatic VT target registration methods, namely the AHA 17-segment model registration, 3D-3D registration and 2D-3D registration. The AHA 17-segment model registration method divides the LV myocardium structure contoured from cardiac CT into 17 segments according to the AHA 17-segment model, enabling the assessment of targeted LV segment(s) and follow-up studies. The 3D-3D registration method reads vendor-specific EAM raw data and transfers the 3D VT ablation points to the 3D LV contours with respect to the treatment planning imaging data. The 2D-3D registration method is a versatile approach that supports any EAM system and enables the transfer of the VT target region marked on the 2D EAM screenshots in standard anatomical views to the 3D LV contours with respect to the treatment planning imaging data. These three registration methods are semi-automatic rather than fully automatic due to strict accuracy requirements in clinical applications. Automatic registration methods may not be sufficiently reliable, as the LV and aorta structures are derived from different modalities, which can introduce inaccuracies and data incompleteness, making them unsuitable for clinical use. In contrast, the proposed semi-automatic registration methods have demonstrated practical feasibility on real-world STAR datasets. They provide the necessary flexibility, allowing clinicians to refine the registration process based on their expertise and the specific characteristics of each STAR case, ensuring both accuracy and clinical applicability. For aspect (2), the software was successfully validated as a quality assurance tool in the STAR treatment planning procedure for 5 VT cases within the German RAVENTA trial. Particularly, the 2D-3D registration method eliminates the need for interpreting proprietary formats exported from different EAM systems. The semi-automated VT target registration methods enable quality assurance of the manually transferred cardiac CTV, reducing clinician-dependent inconsistencies and enhancing the safety and robustness of the VT target registration. Additionally, retrospective findings of incorrectly transferred VT target could potentially help explain VT recurrences. In a cross-validation study addressing (3), the proposed 2D-3D registration method and the 3D-3D registration method from the 3D Slicer extension EAMapReader outputted nearly identical cardiac CTV structures. This result indicates that both methods are suitable for quality assurance and VT target transfer to avoid mistargeting and provide standardized workflows. Finally, regarding aspect (4), this dissertation presents an electrocardiogram-gated cardiac CT-based patient- and segment-specific cardiac motion estimation method using the intensity-based non-rigid automatic image registration in STAR for VT. The method was utilized on case data from 10 STAR-treated VT patients, and the estimated cardiac motion demonstrated considerable individual variability in cardiac CTVs and 17 LV segments across different VT patients, highlighting the need for individualized cardiac ITV margins and motion management strategies to enhance accuracy and effectiveness in STAR. Additionally, this analysis provides reference data on cardiac motion for STAR treatment planning in VT patients. This method has been integrated into the proposed software as a module. In summary, three practical semi-automatic VT target registration methods were developed and validated, and a patient- and segment-specific cardiac motion estimation method was proposed. These methods bridge the gap between EAM systems and radiation TPS, enhancing STAR performance and improving VT patient outcomes, with potential for future clinical applications.