PCAderived respiratory motion surrogates from Xr

Authors: Hua Ma Gerardo Dibildox Carl Schultz Evelyn Regar Theo van Walsum

Publish Date: 2015/04/07

Volume: 10, Issue: 6, Pages: 695-705

PDF Link

Abstract

Intraoperative coronary motion modeling with motion surrogates enables prospective motion prediction in Xray angiograms XA for percutaneous coronary interventions The motion of coronary arteries is mainly affected by patients breathing and heartbeat Purpose of our work is therefore to extract coronary motion surrogates that are related to respiratory and cardiac motion In particular we focus on respiratory motion surrogates extraction in this paperWe propose a fast automatic method for extracting patientspecific respiratory motion surrogate from cardiac XA The method starts with an image preprocessing step to remove all tubular and curvilinear structures from XA images such as vessels and guiding catheters followed by principal component analysis on pixel intensities The respiratory motion surrogate of an XA image is then obtained by projecting its vesselremoved image onto the first principal componentThis breathing motion surrogate was demonstrated to get high correlation with ground truth diaphragm motion correlation coefficient over 09 on average In comparison with other related methods the method we developed did not show significant difference p005 but did improve robustness and run faster on monoplane and biplane data in retrospective and prospective scenariosPercutaneous coronary intervention PCI is a minimally invasive procedure for treating patients with advanced coronary artery disease With this technique a catheter system is introduced into patients’ circulation system through their femoral or radial artery under local anesthesia A preshaped guiding catheter is positioned into the ostium of the coronary artery Through this catheter a guide wire serving to deploy devices such as balloon catheters and stents is advanced into the branches of the artery Once a stenosis is targeted the balloon is deployed at the lesion site to fix the vessel blockage 1PCI is normally performed in a catheterization laboratory under the guidance of Xray angiography XA that coronary arteries are opacified with contrast agent However due to contrast agent toxicity its injection times are limited such that guide wire and device advancement into the target lesion is performed under “vesselfree” images In this situation interventional cardiologists have to mentally reconstruct the position of coronary arteries and stenosis based on previous images which increases the risk of failure for the procedures especially in difficult casesTo address this problem Shechter et al proposed to model the coronary motion with surrogate signals from contrastenhanced images and hence the guidance in “vesselfree” images becomes feasible by prospective motion correction with such a model 3 As the motion of coronary arteries is mainly affected by patient’s breathing motion and cardiac heart beat it is reasonable to model the coronary motion with surrogates which are correlated with patients’ respiratory and cardiac motion With such an aim we have been focusing on respiratoryinduced coronary motion modeling Therefore the purpose of our study was to develop and evaluate a method for fast and robust extraction of respiratory motion surrogates from Xray angiograms for PCIRelated works on respiratory motion surrogates have been reported Signals from external apparatus such as navigators or bellows have been used in many studies for respiratory motion modeling 2 Usage of imagebased surrogates have also been investigated One commonly used surrogate is diaphragm superioinferior SI motion 3 4 5 This is extracted by drawing a rectangular ROI on diaphragm border followed by manual tracking the diaphragm or automatic calculating the 1D translation These methods involve human interaction to draw an ROI and hence not entirely automatic Automatic diaphragm detection and tracking were reported in 6 7 These methods use morphological operation to preprocess XA images followed by a secondorder curve fitting to the diaphragm border Studies on other respiratoryrelated objects eg coronary sinus catheter and tracheal bifurcation can be seen in 4 8 These methods require specific objects being present in images which is not always the case in XA images for PCI Dimension reduction techniques have been used for studying respiratory motion as well In 9 an automatic method based on principal component analysis PCA was designed for retrospective motion gating This method first creates a mask using Hessianbased vesselness filtering and analyzes pixels inside the mask with PCA technique In another study 10 hierarchical manifolding learning was used to find correlation between image regions and respiratory motion

Keywords: