An optimized multisource acquisition and registration workflow for imaging guided ventricular tachycardia ablation

Abstract Background Catheter ablation is a cornerstone treatment for scar-related ventricular arrhythmias arising in patients affected by structural heart disease. Imaging integration with electroanatomical mapping (EAM) suites for ventricular tachycardia (VT) ablation procedural guidance is an established step for advanced substrate characterization and treatment. Purpose To compare and validate different approaches for intraprocedural multisource imaging registration and integration with EAM suites for VT ablation guidance. Methods Thirty consecutive patients prospectively enrolled in the ongoing VOYAGE clinical trial for imaging guided/aided VT ablation at our center were retrospectively analyzed. Multidetector computed tomography (MDCT) and late gadolinium enhancement cardiac magnetic resonance (LGE-CMR) imaging was performed before procedure. MDCT and LGE-CMR images were both acquired during end-expiratory breath hold. Same cardiac cycle phase images were used. Wide band LGE-CMR was acquired in case of a cardiac implantable electronic device was present to reduce artifacts. Images were imported in ADAS3D software. MDCT and CMR images were manually registered using established anatomical landmarks (left ventricle outflow tract, left ventricle apex, papillary muscles, left and right ventricle junction), were then respectively processed for aortic/ventricular anatomical reconstruction and myocardial characterization and then imported in CARTO3 electroanatomical mapping suite for procedural guidance. Registration between EAM and imaging was either performed with ascending aorta fast anatomical mapping (FAM) and 3D landmark alignment with aortic MDCT or via intracardiac echocardiography (ICE) left ventricle (LV) chamber segmentation and 3D landmark alignment with ventricular MDCT. ICE segmentation was also obtained for every case regardless of aortic FAM. Surface match between ICE and MDCT LV chamber reconstructions was calculated for method comparison. Results All 30 patients had an ICE segmentation of the LV, with a mean of 15,66±4,89 contours. Aortic FAM was obtained and used for aortic MDCT registration in 10 subjects. LV ICE was instead used for imaging alignment in 20 subjects. When compared to aortic FAM, LV ICE to LV MDCT alignment achieved lower average surface distance (3,86±1,41 mm vs 8,884±6,77 mm; p<0,001), lower minimum surface distance (0,002±0,03 vs 0,07±0,06; p= 0,001) and lower maximum surface distance (14,16±4,37 vs 22,93±11,10; p<0,001). Conclusion An optimized workflow consisting of ICE LV chamber segmentation was superior in terms of 3D surface distance to aortic FAM reconstruction to obtain reliable multisource imaging registration for MDCT/CMR guided/aided VT ablation.