BioPSE visualizing the electrical field generated by an ICD device.

One of the exciting projects in which the SCI Institute is collaborating is the development of new tools and techniques to assist doctors optimize the placement of Implantable Cardiac Defibrillators (ICDs) prior to surgery.

The use of ICDs has greatly increased over the last few years due to their efficacy in preventing sudden cardiac death (SCD) in patients with congenital heart defects or heart disease. These devices work by continually monitoring the rhythm of the patient's heart and immediately delivering a corrective electric shock if a life-threatening tachycardia is detected. Through this innovation, thousands of lives are saved each year. Surprisingly, these devices are sometimes implanted in newborns and older children with congenital heart defects. Pediatric patients present a particular challenge to the surgeons planning an implantation due to the wide variety of shapes and sizes of torsos. It often has proven difficult for physicians to determine the ideal placement and orientation of the electrodes prior to surgery. Accurate placement of the electrodes is crucial to ensure successful defibrillation with a minimum amount of electric current and to minimize potential damage to the heart and the surrounding tissues.

The SCI Institute's NIH Center for Integrative Biomedical Computing (CIBC) is working in close collaboration with Dr. John Triedman and Dr. Matthew Jolley at Children's Hospital in Boston in order to develop the methods and software infrastructure necessary to allow physicians to simulate - prior to surgery - the placement and activation of ICDs implanted within their patients.

To achieve this goal, the scientists and developers involved have outlined a process for generating patient specific computer models and conducting the simulations. By developing a set of tools and Power Applications for the SCIRun/BioPSE Problem Solving Environment specifically designed to facilitate this process, they hope to significantly improve the physician's ability to provide accurate and effective treatment to patients with minimal invasiveness.

(A) Segmentation of CT-scan (B) Placement of Electrodes (C) Computational Mesh

The process starts by performing a CT or MRI scan of the patient from the neck to the pelvis. The scanner generates a series of grayscale 2D slice images of the patient's internal anatomy which are assembled into a 3D volume. The data is then loaded into a segmentation tool such as Seg3D which is used to separate and identify the various tissues and organs displayed in the images. The software generates a Finite Element Mesh (FEM) of the patient's anatomy, breaking the volume into thousands of tiny cubic boxes which are each assigned a series of values to describe the physical properties of the tissues they represent. Each box is assigned a tissue type (e.g., heart, blood, bone, lung, kidney, liver, muscle, connective tissue) as well as a level of electrical conductivity based on a set of known typical values. Segmentation is the most challenging part of the process because the computer must accurately identify all of the tissues based on the grayscale images produced by the scanner. This step still requires involvement by the physician to make judgment calls in identifying the tissue types. Accuracy in this step is critical to insure that the computer model is representative of the actual patient physiology.

Results of defibrillation current projected onto a cross-sectional plane.

Once the finite element mesh has been created, the physician loads the model into a system called BioPSE, or Biomedical Problem Solving Environment. This software tool developed at the Scientific Computing and Imaging Institute provides an intuitive three-dimensional simulation environment that allows users to construct and to simulate a wide range of biomedical and scientific problems. Once the model has been loaded into BioPSE, the physician is able to select the type of ICD device planned and then to implant it virtually into the patient's torso. The user may adjust the electrode placement interactively and immediately view the resulting defibrillation current throughout the tissues in the patient's torso. The voltages at every location are represented by color. BioPSE allows the physician to view the model from any angle and to change interactively the positioning of the electrodes and immediately to see the resulting changes in the field. This instant visual feedback allows the doctor to locate the ideal placement for the electrodes before cutting into a patient.

"We utilize SCIRun and BioPSE to interactively explore novel locations for implantable cardiac defibrillators (ICDs) in children. ICD electrode placement in kids is difficult due to their size and other anatomical limitations. As a result, many children require a unique placement performed on an ad hoc basis. SCIRun and BioPSE provide an easy to use environment to place virtual electrodes in individual child torso models and subsequently compute and visualize the expected defibrillating electric fields, allowing logical placement of electrodes rather than clinical trial and error. Our research would not be possible without the ongoing development and customization of these tools by the SCI Institute." -- Dr. Matthew Jolley, Department of Cardiology, Children's Hospital Boston.

Visualization of currents on the surface of the heart during a defibrillation simulation.

In order to achieve successful defibrillation, the device must deliver fields of 5V/cm over 95% of the cardiac tissue. At the same time, it is important to avoid delivering more than about 60V/cm to any tissues in order to avoid damage. Electricity flows through the human anatomy in a complex manner resulting in relative hot spots and dead zones. Small changes in the placement of the electrodes may cause large changes to the resulting field. Placing the electrodes directly on the surface of the heart may efficiently deliver the electricity to the heart tissue with minimal spreading to the surrounding organs, however it also results in an uneven field distribution with hot spots near the electrodes. Placing the electrodes further from the heart (e.g., just under the skin) may result in a more even field on the heart but this requires a higher current and more electricity is delivered to the other tissues. Finding the right balance is a primary challenge to the physician.

Making this process feasible for physicians without the involvement of computer scientists, and accurate enough to represent real patients, is just the type of challenge SCI scientists are well suited for. The resulting simulation process must be quick, highly automated, and reproducible. A major focus of this research is the process of automatic segmentation necessary for converting a series of MRI or CT scans into an accurate model of the patient's torso. Currently, the physician's help is still required for accurately identifying tissues. Developers at SCI are working on new algorithms to improve this process which will be integrated into Seg3D, SCI's segmentation software. We are also developing task specific tools within BioPSE to make ICD simulation as simple and accurate as possible.

Our collaborating physicians at Children's Hospital in Boston have used this system in a number of cases including both children and adults. With this experience, they have been able to demonstrate that simulations developed in this manner are able to provide reasonably accurate results. Our collaborators were also able to gain important insights into electrode placement and design that would have been impossible previously.

Through continued close collaboration with these physicians in the field, we are confident that this system has the potential to become an indispensable tool for planning ICD implantation.