An integrated platform for 2‐D and 3‐D optical and electrical mapping of arrhythmias in Langendorff‐perfused rabbit hearts

Figure 1. Schematic diagram and photographs of the integrated experimental setup for simultaneous panoramic optical and electrical mapping A, schematic representation of the three‐dimensional (3‐D) system integrating the optical mapping system (light green) and the electrical mapping system (light purple). Three high‐speed cameras (CAM A, B and C) are positioned around the Langendorff‐perfused heart for panoramic optical mapping. The electrical mapping system includes three custom‐designed multi‐electrode arrays (MEAs), labelled #1 to #3; and a torso‐tank model for non‐invasive mapping labelled #4. These MEAs are placed over the RA, LA and ventricular surface, respectively. Detailed representations of the MEAs, including their real dimensions, are shown at the bottom of the schematic. B, top‐view photograph of the experimental setup mounted on the optical table, highlighting the positioning of the cameras. A dashed box indicates the custom‐built 3‐D support structure designed to acquire the sequence of images for the 3‐D reconstruction with the optical camera A. C, photograph of the system during an optical mapping experiment, showing red LEDs (650 nm) illuminating the heart.

Figure 2. 3‐D acquisition and calibration A, the optical system acquires images by rotating a single camera 360° around the heart (left). Calibration is performed using a triangular prism placed at the centre of the setup. B, segmentation, reconstruction and smoothing: a semi‐automatic segmentation interface (left) identifies the heart silhouette across all frames. The reconstruction process (centre) carves regions of interest (ROIs) to generate the 3‐D mesh. Comparison between initial and smoothed meshes (right). C, geometry: refined meshes obtained from the volume boundaries, after applying simplification and filters. D, interface (left) used to align and refine the mapping between optical and geometric data. Inverse projection examples from three cameras are shown (right). E, reprojection accuracy of the 3‐D mesh: validation of the alignment between 2‐D and 3‐D data using projection contours (bottom row). F, electrode location: python interface to manually identify electrodes location using the 2‐D optical data (left) and the already computed electrodes over the 3‐D heart surface (right). G, 3‐D surface projections showing atrial depolarisation (left), ventricular depolarisation…

Figure 3. Simultaneous optical and invasive electrical mapping during sinus rhythm with AV block (SR/AVB), ventricular tachycardia with a premature atrial contraction (VT/PAC) and atrial fibrillation with AV block (AF) A, upper: showing representative optical and EGM signals recorded from selected regions on the heart surface, illustrating cardiac activity during SR/AVB (left), VT/PAC (middle) and AF (right). Lower: yellow arrows and circles indicate the anatomical locations where these signals were acquired, corresponding to the electrode positions shown in the camera‐acquired images. B, 3‐D LAT maps for electrical (top) and optical (bottom) data during SR/AVB (left), VT/PAC (middle) and AF (right). This illustrates the synchronisation of data capture and the differentiation between atrial (red) and ventricular (blue) activity. All isochronal lines in LAT are 5 ms apart, except AF with 2 ms. C, 3‐D phase maps for electrical (top) and optical (bottom) data under SR/AVB (left), VT/PAC (middle) and AF (right).

Figure 4. Comparative analysis of VT and VF A, electrophysiological recordings and anatomical context. Left: electrode 3 (E3) voltage signals and fractional change in fluorescence over time during VT, with anatomical position marked on the camera‐acquired images. Right: electrode 3 (E3) voltage signals and fractional change in voltage over time during VF, with electrode position indicated on the camera‐acquired images. Locations where the signals were obtained are highlighted with a yellow arrow and circle, corresponding to the electrode positions visualised in multi‐angle camera‐acquired images. B, optical LAT maps acquired from the three camera views, and contact electrical LAT map from the ventricular MEA, all isochronal lines in LAT are 5 ms apart. C, 3‐D optical phase maps acquired from the three camera views, and a 3‐D contact electrical phase map from the ventricular MEA.

Figure 5. Non‐invasive mapping of a SR with AV node blockage A, torso‐tank setup with labelled faces and electrodes; the heart remains centrally positioned throughout the experiment. B, torso‐tank signals during SR, with electrode numbers and tank face labels (Fa–Ff) matching their positions. C, 250 ms segment from electrodes 33, 18 and 58 during SR (left); six red lines mark time points for torso‐tank potential maps (right), where white ± signals indicate relative potential levels. D, CC between EGMs (black) and rEGMs (red). Left and middle: representative 130 ms segments from four RA and four LA electrodes, with signals overlaid and aligned by computed lag; CC values and lags are shown. Right: boxplots of CC across all 16 electrodes per region. All signals normalised between –1 and 1. E, left: 4.5 s segment showing, from top to bottom, optical (RA, LA), electrical (RA, LA, tank) and reconstructed (rEGM) signals. Right: corresponding frequency spectra.

Figure 6. ECGi analysis during VT A, left: three‐dimensional rEGM potential map with highlighting ventricular electrode positions, alongside the corresponding two‐dimensional EGM potential map. Four representative electrodes (EL13–16) are shown with normalised EGM (black) and rEGM (red) signals overlaid, including their CC and lag values. Right: boxplot of CC values across the 16 ventricular electrodes. B, left: two‐second signal segments recorded from the ventricular region, showing optical fluorescence, measured EGMs, torso‐tank signals and the reconstructed rEGM. Right: frequency spectra of each signal.