Restoration of optimal ellipsoid left ventricular geometry: lessons learnt from in silico surgical modelling.

OBJECTIVES: Several issues that are inherent in the surgical techniques of surgical ventricular restoration (SVR) need specialized devices or techniques to overcome them, which may not always result in optimal outcomes. We used a non-invasive novel in silico modelling technique to study left ventricular (LV) morphology and function before and after SVR. The cardiac magnetic resonance imaging derived actual pre- and postoperative endocardial morphology and function was compared with the in silico analysis of the same.

METHODS: Cardiac magnetic resonance steady state free precession (SSFP) cine images were employed to segment endocardial surface contours over the cardiac cycle. Using the principle of Hausdorff distance to examine phase-to-phase regional endocardial displacement, dyskinetic/akinetic areas were identified at the instant of peak basal contraction velocity. Using a three-dimensional (3D) surface clipping tool, the maximally scarred, dyskinetic or akinetic LV antero-apical areas were virtually resected and a new apex was created. A virtual rectangular patch was created upon the clipped surface LV model by 3D Delaunay triangulation. Presurgical endocardial mechanical function quantified from cine cardiac magnetic resonance, using a technique of spherical harmonics (SPHARM) surface parameterization, was applied onto the virtually clipped and patched LV surface model. Finally, the in silico model of post-SVR LV shape was analysed for quantification of regional left ventricular volumes (RLVVs) and function. This was tested in 2 patients with post-myocardial infarction antero-apical LV aneuryms. Left ventricular mechanical dysynchrony was evaluated by RLVV analysis of pre-SVR, in silico post-SVR and actual post-SVR LV endocardial surface data.

RESULTS: Following exclusion of the scarred areas, the virtual resected LV model demonstrated significantly lesser areas of akinesia. The decreases in regional LV volumes in the in silico modelling were significant and comparable with the actual decreases following SVR. Both the regional end diastolic volume (EDV) and end systolic volume (ESV) at the apex decreased significantly corresponding to greater reductions in apical volumes by the technique of rectangular patch plasty (apical EDV 2.1607 ± 0.20577 to 0.4774 ± 0.1775 ml, P = 0.007; apical ESV 1.9708 ± 0.36451 to 0.442 ± 0.047 ml, P = 0.013).

CONCLUSIONS: This pilot study was done using novel in silico techniques for virtual surgical modelling, which helped in accurate estimation and planning of optimal LV restoration by SVR.