IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 57, NO. 9, SEPTEMBER 2010 2209 Evaluation of Model-Enhanced Ultrasound-Assisted Interventional Guidance in a Cardiac Phantom Cristian A. Linte*, Student Member, IEEE, John Moore, Chris Wedlake, and Terry M. Peters, Fellow, IEEE Abstract—Minimizing invasiveness associated with cardiac pro- cedures has led to limited visual access to the target tissues. To address these limitations, we have developed a visualization en- vironment that integrates interventional ultrasound (US) imaging with preoperative anatomical models and virtual representations of the surgical instruments tracked in real time. In this paper, we present a comprehensive evaluation of our model-enhanced US-guidance environment by simulating clinically relevant inter- ventions in vitro. We have demonstrated that model-enhanced US guidance provides a clinically desired targeting accuracy better than 3-mm rms and maintains this level of accuracy even in the case of image-to-patient misalignments that are often encountered in the clinic. These studies emphasize the benefits of integrating real-time imaging with preoperative data to enhance surgical nav- igation in the absence of direct vision during minimally invasive cardiac interventions. Index Terms—In vitro therapy evaluation, minimally invasive cardiac interventions, model-enhanced visualization and naviga- tion, surgical tracking technologies, ultrasound (US) imaging. I. INTRODUCTION T HE DEVELOPMENT of minimally invasive alternatives to conventional cardiac therapy has been under active in- vestigation [1]–[3]. However, these techniques have inevitably led to reduced access to the sites that require treatment and re- duced visual access to the surgical field itself. As an example, in a typical robot-assisted procedure [4]–[6], surgical access is achieved using robotic instruments that are inserted inside the patient’s thorax through access ports, while surgical visualiza- tion is achieved via real-time laparoscopic video.Moreover, a wide range of minimally invasive cardiac ablation procedures for arrhythmia treatment [7]–[9] or valvular interventions [10], Manuscript received December 27, 2009; revised March 18, 2010; accepted May 12, 2010. Date of publication May 27, 2010; date of current version August 18, 2010. This work was supported by the Canadian Institutes of Health Research under Grant MOP 179298, by the Natural Sciences and Engineer- ing Research Council under Grant 155108-07 and through Canada Graduate Scholarships Doctoral Award, by the Heart and Stroke Foundation of Canada through Doctoral Research Award, by the Canada Foundation for Innovation under Grant 5184, and by the Ontario Research Fund under Project Imaging for Cardiovascular Therapeutics (ICT). Asterisk indicates corresponding author. *C. A. Linte is with the Imaging Research Laboratories, Robarts Research Institute, University of Western Ontario, London, ON N6A 5K8, Canada (e-mail: clinte@imaging.robarts.ca). J. Moore, C. Wedlake, and T. M. Peters are with the the Imaging Research Lab- oratories, Robarts Research Institute, University of Western Ontario, London, ON N6A 5K8, Canada (e-mail: jmoore@imaging.robarts.ca; cwedlake@ imaging.robarts.ca; tpeters@imaging.robarts.ca). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TBME.2010.2050886 [11] may be performed percutaneously; catheters are navigated through larger blood vessels into the heart, typically under real- time fluoroscopic image guidance, or as recently shown, using integrated X-ray fluoroscopy and MRI [12]–[14], often referred to as X-MR guidance. Nevertheless, most intracardiac proce- dures are still performed on the empty, arrested heart, primarily due to the challenges associated with safe access and adequate visualization required for therapy delivery inside the beating heart. Prior to performing an intervention, diagnostic images of the patient are reviewed offline and together with preopera- tive images are used to plan out the procedure [15]. It is also commonly assumed that these images represent the intraop- erative morphology with sufficient fidelity to enable adequate therapy guidance [16]. Nevertheless, cardiac therapy remains a challenging problem for image guidance due to the complex soft-tissue structure and motion of the heart, and consequently, due to the limited accuracy achieved when modeling the in- traoperative heart from preoperative data. Therefore, real-time intraoperative visualization is critical to enable minimally inva- sive beating-heart therapy. In response to these challenges, we have developed an inter- ventional guidance platform that relies on multimodality med- ical imaging for visualization and manipulation of intracardiac structures in absence of direct vision [17], [18]. Our platform in- tegrates transesophageal echocardiography (TEE) for real-time visualization augmented with preoperative models of the car- diac anatomy and electrical potential maps and virtual repre- sentations of the surgical instruments tracked in real time using magnetic-tracking technologies [19]. The end result is a model- enhanced ultrasound (US) surgical-guidance environment— one of the first attempts toward bridging diagnosis and surgical planning with interventional guidance, thus allowing the 2-D in- traoperative US data to be interpreted within the 3-D anatomical context provided by the preoperative models [20]. However, before any novel image-guidance platform is trans- lated into a clinical setting, a robust quantitative assessment of its surgical navigation capabilities is critical. To perform a true surgical-accuracy assessment, precise knowledge of both the surgical tool and target locations is required. While an in vivo beating-heart assessment is preferred, it would entail a very in- vasive process of implanting tracked targets inside the in vivo porcine myocardium, closing up the thoracic cavity, acquiring the necessary preoperative images, and ultimately, performing the procedure. As a tradeoff, several groups have resorted to the use of representative phantoms to simulate in vitro clinical procedures for image-guidance evaluation [21]–[24]. We have also used various phantoms to conduct qualitative assessments 0018-9294/$26.00 © 2010 IEEE