Marcin Wierzbicki

Validation of dynamic heart models obtained using non-linear registration for virtual reality training, planning, and guidance of minimally invasive cardiac surgeries

Current minimally invasive techniques for beating heart surgery are associated with three major limitations: the shortage of realistic and safe training methods, the process of selecting port locations for optimal target coverage from X-rays and angiograms, and the sole use of the endoscope for instrument navigation in a dynamic and confined 3D environment. To supplement the current surgery training, planning and guidance methods, we continue to develop our Virtual Cardiac Surgery Planning environment (VCSP) -- a virtual reality, patient-specific, thoracic cavity model derived from 3D pre-procedural images. In this work, we create and validate dynamic models of the heart and its components. A static model is first generated by segmenting one of the image frames in a given 4D data set. The dynamics of this model are then extracted from the remaining image frames using a non-linear, intensity-based registration algorithm with a choice of six different similarity metrics. The algorithm is validated on an artificial CT image set created using an excised porcine heart, on CT images of canine subjects, and on MR images of human volunteers. We found that with the appropriate choice of similarity metric, our algorithm extracts the motion of the epicardial surface in CT images, or of the myocardium, right atrium, right ventricle, aorta, left atrium, pulmonary arteries, vena cava and epicardial surface in MR images, with a root mean square error in the 1 mm range. These results indicate that our method of modeling the motion of the heart is easily adaptable and sufficiently accurate to meet the requirements for reliable cardiac surgery training, planning, and guidance.

A High Resolution Dynamic Heart Model Based on Averaged MRI Data

We are in the process of constructing a high resolution, high signal to noise ratio (SNR) dynamic MRI dataset for the human heart using methodology similar to that employed to construct a low-noise standard brain at the Montreal Neurological Institute. Several high resolution, low SNR magnetic resonance images of 20 phases over the cardiac cycle were acquired from a single subject. Images from identical phases and temporally adjacent phases were registered, and the image intensities were averaged together to generate a high resolution, high SNR dynamic magnetic resonance image volume of the human heart. Although this work is still preliminary, and the results still demonstrate residual artifacts due to motion an sub-optimal alignment of interleaved image slices, our model has a SNR that is improved by a factor of 2.7 over a single volume, spatial resolution of 1.5 mm3, and a temporal resolution of 60 ms.

Real-time fusion of endoscopic views with dynamic 3-D cardiac images: a phantom study

Minimally invasive robotically assisted cardiac surgical systems currently do not routinely employ 3-D image guidance. However, preoperative magnetic resonance and computed tomography (CT) images have the potential to be used in this role, if appropriately registered with the patient anatomy and animated synchronously with the motion of the actual heart. This paper discusses the fusion of optical images of a beating heart phantom obtained from an optically tracked endoscope, with volumetric images of the phantom created from a dynamic CT dataset. High quality preoperative dynamic CT images are created by first extracting the motion parameters of the heart from the series of temporal frames, and then applying this information to animate a high-quality heart image acquired at end systole. Temporal synchronization of the endoscopic and CT model is achieved by selecting the appropriate CT image from the dynamic set, based on an electrocardiographic trigger signal. The spatial error between the optical and virtual images is 1.4/spl plusmn/1.1 mm, while the time discrepancy is typically 50-100 ms.

Towards subject-specific models of the dynamic heart for image-guided mitral valve surgery

Surgeons need a robust interventional system capable of providing reliable, real-time information regarding the position and orientation of the surgical targets and tools to compensate for the lack of direct vision and to enhance manipulation of intracardiac targets during minimally-invasive, off-pump cardiac interventions. In this paper, we describe a novel method for creating dynamic, pre-operative, subject-specific cardiac models containing the surgical targets and surrounding anatomy, and how they are used to augment the intra-operative virtual environment for guidance of valvular interventions. The accuracy of these pre-operative models was established by comparing the target registration error between the mitral valve annulus characterized in the pre-operative images and their equivalent structures manually extracted from 3D US data. On average, the mitral valve annulus was extracted with a 3.1 mm error across all cardiac phases. In addition, we also propose a method for registering the pre-operative models into the intra-operative virtual environment.

Determining Epicardial Surface Motion Using Elastic Registration: Towards Virtual Reality Guidance of Minimally Invasive Cardiac Interventions

Minimally invasive cardiac surgery is performed on the beating heart, through inter-costal ports. The two major limitations of these procedures are: selecting port locations for optimal target coverage (based on chest x-rays and angiograms), and navigating surgical tools through a dynamic and confined environment using a 2D endoscope. To supplement the current surgery planning and guidance strategies, we continue developing VCSP – a virtual reality, patient-specific, thoracic cavity model derived from 3D pre-procedural images. In this work, we apply elastic image registration to 4D images of the heart to model the epicardial surface over the cardiac cycle. We validated our registration algorithm on CT images of a dynamic cardiac phantom and of normal canine hearts, and found the error to be 1.14 ± 0.31 mm and 0.61 ± 0.12 mm, respectively. We believe this method of modeling the epicardial surface is sufficiently accurate to be applied in cardiac surgery planning and guidance.

Dose reduction for cardiac CT using a registration-based approach.

Two reasons for the recent rise in radiation exposure from CT are increases in its clinical applicability and the desire to maintain high SNR while acquiring smaller voxels. To address this emerging dose problem, several strategies for reducing patient exposure have already been proposed. One method employed in cardiac imaging is ECG-driven modulation of the tube current between 100% at one time point in the cardiac cycle and a reduced fraction at the remaining phases. In this paper, we describe how images obtained during such acquisition can be used to reconstruct 4D data of consistent high quality throughout the cardiac cycle. In our approach, we assume that the middiastole (MD) phase is imaged with full dose. The MD image is then independently registered to lower dose images (lower SNR) at other frames, resulting in a set of transformations. Finally, the transformations are used to warp the MD frame through the cardiac cycle to generate the full 4D image. In addition, the transformations may be interpolated to increase the temporal sampling or to generate images at arbitrary time points. Our approach was validated using various data obtained with simulated and scanner-implemented dose modulation. We determined that as little as 10% of the total dose was required to reproduce full quality images with a 1 mm spatial error and an error in intensity values on the order of the image noise. Thus, our technique offers considerable dose reductions compared to standard imaging protocols, with minimal effects on the quality of the final data.

On Enhancing Planning and Navigation of Beating-Heart Mitral Valve Surgery Using Pre-operative Cardiac Models

In an effort to reduce morbidity during minimally- invasive cardiac procedures, we have recently developed an interventional technique targeted towards off-pump cardiac interventions. To compensate for the absence of direct visualization, our system employs a virtual reality environment for image guidance, that integrates pre-operative information with real-time intra-operative imaging and surgical tool tracking. This work focuses on enhancing intracardiac visualization and navigation by overlaying pre-operative cardiac models onto the intra-operative virtual space, to display surgical targets within their specific anatomical context. Our method for integrating pre-operative data into the intra-operative environment is accurate within ~5.0 mm. Thus, we feel that our virtually-augmented surgical space is accurate enough to improve spatial orientation and intracardiac navigation.

Subject-specific models for image-guided cardiac surgery

Three-dimensional visualization for planning and guidance is still not routinely available for minimally invasive cardiac surgery (MICS). This can be addressed by providing the surgeon with subject-specific geometric models derived from 3D preoperative images for planning of port locations or to rehearse the procedure. For guidance purposes, these models can also be registered to the subject using intraoperative images. In this paper, we present a method for extracting subject-specific heart geometry from preoperative MR images. The main obstacle we face is the low quality of clinical data in terms of resolution, signal-to-noise ratio, and presence of artefacts. Instead of using these images directly, we approach the problem in three steps: (1) generate a high quality template model, (2) register the template with the preoperative data, and (3) animate the result over the cardiac cycle. Validation of this approach showed that dynamic subject-specific models can be generated with a mean error of 3.6+/-1.1 mm from low resolution target images (6 mm slices). Thus, the models are sufficiently accurate for MICS training and procedure planning. In terms of guidance, we also demonstrate how the resulting models may be adapted to the operating room using intraoperative ultrasound imaging.

Mapping Template Heart Models to Patient Data Using Image Registration

Currently, minimally invasive cardiac surgery (MICS) faces several limitations, including inadequate training methods using non-realistic models, insufficient surgery planning using 2D images, and the lack of global, 3D guidance during the procedure. To address these issues we are developing the Virtual Cardiac Surgery Platform (VCSP) – a virtual reality model of the patient specific thorax, derived from pre-procedural images. Here we present an image registration-based method for customizing a geometrical template model of the heart to any given patient, and validate it using manual segmentation as the gold standard. On average, the process is accurate to within 3.3 ± 0.3 mm in MR images, and 2.4 ± 0.3 mm in CT images. These results include inaccuracies in the gold standard, which are on average 1.6 ± 0.2 and 0.9 ± 0.2 mm for MR and CT images respectively. We believe this method adequately prepares templates for use within VCSP, prior to and during MICS.

Towards a biomechanics-based technique for assessing myocardial contractility: an inverse problem approach

This work presents the initial development and implementation of a novel 3D biomechanics-based approach to measure the mechanical activity of myocardial tissue, as a potential non-invasive tool to assess myocardial function. This technique quantifies the myocardial contraction forces developed within the ventricular myofibers in response to electro-physiological stimuli. We provide a 3D finite element formulation of a contraction force reconstruction algorithm, along with its implementation using magnetic resonance (MR) data. Our algorithm is based on an inverse problem solution governed by the fundamental continuum mechanics principle of conservation of linear momentum, under a first-order approximation of elastic and isotropic material conditions. We implemented our technique using a subject-specific ventricle model obtained by extracting the left ventricular anatomical features from a set of high-resolution cardiac MR images acquired throughout the cardiac cycle using prospective electrocardiographic gating. Cardiac motion information was extracted by non-rigid registration of the mid-diastole reference image to the remaining images of a 4D dataset. Using our technique, we reconstructed dynamic maps that show the contraction force distribution superimposed onto the deformed ventricle model at each acquired frame in the cardiac cycle. Our next objective will consist of validating this technique by showing the correlation between the presence of low contraction force patterns and poor myocardial functionality.