Volumetric ultrasound image reconstruction from a single-element forward-looking intracardiac steerable catheter using 3D adaptive normalized convolution

Abstract

In interventional cardiology catheters are routinely used to access and treat defects and diseases in the heart. Image guidance using forward-looking (FL)ultrasound transducers at the tip of the catheter could give the physician visual feedback during complex procedures such as valve replacement or transseptal puncture. In this work, we investigate FL 3D imaging by integrating a 7 MHz single-element ultrasound transducer at the tip of a novel multi-steerable intracardiac catheter together with an optical shape sensing fiber (OSS). We tested the imaging capability of the integrated device on an ex-vivo pig heart. By acquiring ultrasound A-lines at different locations while steering the catheter tip, a sparse 3D image is obtained. To reconstruct a volumetric image from the sparse data we implemented an adaptive Normalized Convolution (NC)algorithm were the dimension, orientation and angle of the 3D anisotropic kernel changes dynamically according to the scanning path. We acquired ultrasound A-lines of the tricuspid valve and we computed the 3D image using NC with both an isotropic kernel and an anisotropic kernel. We successfully interpolated the sparse data obtaining 3D volumes of the heart. By using an anisotropic kernel better 3D reconstruction is achieved with higher detail information compared to the reconstruction obtained using an isotropic kernel. This pilot experiment demonstrates the potential of FL image guidance during intracardiac procedures using a single-element transducer integrated in a steerable catheter with an OSS fiber.