229-04: Automatic quantification of esophagus position from intra-cardiac echocardiography during atrial fibrillation ablation

Abstract

Purpose: Intracardiac echography (ICE) is used during radiofrequency ablation (RFA) of atrial fibrillation to reconstruct 3D electroanatomic maps of left atrium (LA) and to monitor potential complications, including esophageal injuries. In clinical practice, LA posterior wall (LAPW) contour is manually traced on ICE images and its distance from the esophagus only visually assessed. This tracing procedure is time-consuming and esophagus visual assessment does not allow quantitative evaluation of LAPW-esophagus relationship. The present study aimed to automatically detect esophagus position and its spatial relationship from the LAPW by ICE during RFA. Methods: An algorithm for LA segmentation was preliminary developed and validated to detect LAPW in ICE sequences. Once LAPW was detected, a fast algorithm based on the evaluation gray level intensity distribution in the image was developed to detect candidate pixels belonging to the esophagus wall. Simple second and forth order models were used for fitting resulting in distal and proximal esophagus contours, respectively. The algorithm was tested on 13 ICE acquisitions. In 13 images, one for each ICE sequence, detected esophagus boundaries were compared with manually traced (MT) ones by an experienced cardiologist. Mean esophagus thickness and the distance between esophagus proximal boundary and LAPW was also computed. Conclusion: The developed technique allows automated and accurate detection of LAPW and esophagus position in ICE images. It represents a preliminary step for real-time tracking of posterior LA anatomy and esophagus for monitoring its proximity to ablation catheter to prevent esophagus injuries. Example of LPAW and esophagous detection in one ICE image