Center for Integrative Biomedical Computing

Collaborating Investigator(s): Natalia Trayanova, PhD
Institution: Johns Hopkins

Dr. Trayanova is one of the current leaders in the application of multiscale simulation to cardiac rhythm disturbances (arrhythmias) and their treatment through techniques such as defibrillation and catheter ablation. She and her group have pioneered advances in computational modeling of activation in the heart and in the geometric modeling of cardiac tissues and whole hearts, e.g., in the use of rule-based algorithms to assign details of myocardial fiber structure and the assignment of altered electrophysiology to regions of the atria based on MRI imaging.

This project is an extension of the existing collaborator role that Dr. Trayanova and her group have had in the present Center and recognizes both the success to date and the exciting opportunities for future progress. Dr. Trayanova seeks to leverage a career of experience in creating realistic simulations of cardiac electrical activity to the clinical setting and this is also the focus of this DBP relationship. The demands of rapidly creating accurate models of the thorax and hearts of each patient under the clinical time constraints are enormous and require a careful balance between precision and speed. From this need comes a natural drive to the Image and Geometric Analysis TR&D, which is well placed to improve both segmentation and mesh generation for such models. Within the aspects of image analysis, the analysis of the shape of the heart is another driving aspect of this relationship. Dr. Trayanova has already carried out preliminary studies using very simple geometric elements of heart shape and has one of the largest collections of segmented heart geometries from patients to which we will apply sophisticated statistical shape analysis.

This DBP is also an ideal partner for the Simulation & Estimation TR&D in the general setting of simulation and estimation and also in evaluating the role of uncertainty, especially because the results of their simulations are set to become part of clinical practice. Physicians (and their patients) require information on the confidence and accuracy of the predictions that will guide diagnosis and therapy and uncertainty quantification seeks to provide such estimates. This DBP will also drive the Simulation & Estimation TR&D as it presents an ideal opportunity to develop support for integration of third-party simulation tools within the SCIRun framework. Dr. Trayanova is one of the original users and drivers for the CARP simulation software, and within the Center, we have already developed applications that leverage CARP and SCIRun for improved integration and management of simulations. Another area of new potential within this DBP is the application of cardiac inverse solution approaches to the problems that Dr. Trayanova traditionally approaches from a forward sense.

Finally, this DBP is closely related to the Marrouche/CARMA DBP, with which there have already been interactions with the aim of extracting patient-specific models of patients with atrial fibrosis from the CARMA studies and then using simulations to study the underlying mechanisms of atrial fibrillation. The Center will facilitate these interactions with the specific goal of uncovering the putative mechanistic relationship between structural features of the remodeled, fibrotic, and scarred heart and reentry and arrhythmias.

This project is highly innovative for its potential to touch so many aspects of computational medicine as it can be applied to diseases of the heart. Dr. Trayanova is poised to provide physicians with a means to predict arrhythmias from image data and then to use the power of simulation to guide therapies. Such disruptive technology represents the absolute cutting edge of computation cardiology.
Collaborating Investigators: Alvaro Pascual-Leone, MD, PhD, Michael Fox, MD, PhD, Mark Halko, PhD, Moushin Shafi, MD, Brad Manor PhD
Institution: Beth Israel Deaconess Medical Center and Harvard Medical School

Dr. Pascual-Leone, is the director of the Berenson-Allen Center for Noninvasive Brain Stimulation at Beth Israel Deaconess Medical Center. He and his collaborators carry out extensive research and clinical studies in the application of all types of transcranial neuromodulation to the treatment of neuropsychiatric disorders such as depression and schizophrenia, epilepsy, and chronic pain as well as studies of mechanisms and effects of stimulation on both normal and disease populations. The other investigators listed are either already actively involved with CIBC (Drs. Fox and Halko) or are in the initial planning stages of projects (Drs. Shafi and Manor). The primary interaction for this DBP is with the Simulation & Estimation TR&D, in the areas of effective algorithms for broad and flexible simulation and optimization of transcranial modulation, fast model building for simulation in population studies, and quantification of uncertainty in simulations and optimizations. In addition the BrainStimulator package, in an initial form, is already being used by Dr. Halko and this DBP is a primary driver of its development. However, we also anticipate significant interaction with the Visualization TR&D in the area of uncertainty visualization, an essential advance in order for uncertainty quantification to be of use to our DBP collaborators. Finally, in our effort to streamline model building while keeping error and uncertainty within acceptable levels, we have employed shape modeling and shape statistical methods developed under the Image and Geometric Analysis TR&D.

There are many practitioners adopting transcranial modulation methods around the world, but there are very few studies establishing mechanisms, limits, and best practices for control of current deposition in realistic geometries, and a striking paucity of clear mechanisms of action or effective dose. At the same time, there are only modest numbers of published studies using computational methods to predict, and even fewer to optimize, current deposition with transcranial technologies. Validation remains a huge problem, as does the availability of effective, efficient, usable, open source, and well supported software tools to study these problems from a computational point of view. Useful progress in all these realms can only come from a close partnership between leading clinicians and scientists in the field of applied neuromodulation and a strong algorithm, modeling, and software center such as CIBC. Our key innovation, then, is truly this research collaboration itself. The innovative products of that collaboration will include modeling and optimization tools that are tested and validated in a diverse set of stimulation scenarios, the ability to carry out such computational studies on more significant population scales, the provision of modeling tools with usable controls to clinical and research practitioners, quantification and visualization of the uncertainty associated with computational predictions, and increased understanding of the contribution of stimulation parameters and bioelectric anatomy and function to the broader questions of mechanism and dose. In our discussions with Dr. Pascual-Leone, he has underlined to us his belief that only through innovation in effective modeling methods can progress be made at a sufficiently rapid rate towards these longer-term goals. Moreover, because we freely distribute and actively disseminate and support our software, and in light of our established relationships with the other DBPs and collaborators in the Neuromodulation and Brain Source Localization Research Cluster, our work has a broad impact across the country and the world among the rapidly growing community in transcranial neuromodulation research and treatment.
Collaborating Investigator(s): Alison Marsden, PhD
Institution: University of California, San Diego

Professor Alison Marsden directs the Cardiovascular Biomechanics Computation (CBC) Lab at the University of California, San Diego. Her laboratory pursues the development and application of cardiovascular blood flow simulation tools to tackle clinically relevant questions in adult and pediatric cardiovascular disease. These developments include multiscale modeling and fluid structure interaction methods, with validations of simulation results against clinical data. Clinical applications include models of reverse graft stenosis to identify precursors of vein graft failure in post-CABG patients and thrombotic risk assessment in children with coronary artery aneurysms caused by Kawasaki disease. The CBC Laboratory also develops and releases open source tools, SimVascular to facilitate the simulation of cardiovascular function.

The CBC Laboratory uses computational models to study the interactions between vascular morphology and its effects on blood flow. These computational models typically include patient-specific geometries that are derived from 3D images (and in some cases, hypothetical modifications to vasculature). The results of these simulations help scientists and clinicians understand the relationships among structural, cardiovascular pathologies (e.g., congenital heart defects), and cardiovascular function. The work of Professor Marsden's laboratory is providing a means to predict the outcome of surgeries, systematically test and optimize new surgical approaches and devices, and personalize treatments for individual patients—and is thereby leading to paradigm shifts in clinical practice.

Several aspects of this work benefit directly with the technology in the Center and the goals in the Image and Geometric Analysis TRD. First, segmentation of vascular blood pools remains an important aspect of this work, and the CBC group is typically looking for more efficient tools/methods for different types of data. Second, open source tools for meshing continue to be a challenge for this group (and many others), and there is a concerted effort to examine how the Center's meshing technology might facilitate this research. Finally, there is a strong interest from the CBC Laboratory in statistical shape analysis and uncertainty quantification. In particular, important aspects of cardiovascular function are affected by pathological morphology, and the investigators hope to use a combination of simulation and shape analysis to form general quantifiable properties of morphology that help determine pathological function (e.g., parameters derived through projections onto low-dimensional shape space). There are also important needs surrounding the uncertainty in both the patient-specific geometries and the resulting simulations. Thus, the uncertainty analysis work in the Image and Geometric Analysis TRD and the uncertainty visualization work in the Visualization TRD are directly motivated and influenced by this DBP.
Collaborating Investigator(s): Michael Okun, MD, Kelly Foote, MD
Institution: University of Florida

Deep brain stimulation (DBS) is a therapy in which a surgically implanted system can provide relief from disabling symptoms for a variety of disorders. It is an FDA-approved therapy for the treatment of Parkinson's disease (PD) and essential tremor (ET). DBS can modulate several different circuits in the vicinity of the electrode, which in turn can induce therapeutic effects and/or side effects for a wide range of neurological disorders. The DBS community is challenged to understand both the motor and non-motor effects of DBS, and to create systems that improve symptoms and avoid side effects. The broad goals of this research group are:

  1. to develop technology with which to measure and improve quality of life for patients treated using DBS;
  2. to form a data repository of patient outcomes that enables analysis of much larger patient populations than would be possible in a single center or clinical trial; and
  3. to use this data to accurately predict who will respond to DBS and how treatment should be applied to achieve the best possible outcomes for future patients.

Drs. Okun and Foote are both full professors and are co-directors of the Center for Movement Disorders & Neurorestoration at the University of Florida; both are world experts in their fields. They currently lead a group that is at the forefront of using DBS to treat patients with a range of neurological conditions including movement disorders (e.g., PD, ET, dystonia), psychiatric disorders (e.g., depression, Tourette syndrome), and neurodegenerative disorders (e.g., Alzheimer's). Combined, they have over 300 peer reviewed publications and their center has several ongoing scientific studies and clinical trials.

The fundamental purpose of DBS is to modulate neural activity with applied electric fields, but few investigators have the tools to enable a quantitative understanding of how changes in lead location or stimulation settings will modify activity in different brain circuits. Recently a number of studies have documented the electric fields generated during DBS. The computational power and computer science skills necessary to effectively implement such models are not available at most DBS centers. Nonetheless, many practitioners now recognize the value of computational models and interactive visualization for guiding the clinical application of DBS, as well as providing a basis for novel scientific studies. In addition, the stakes are high for patients, many of whom have received life- changing improvements in treatment, but some of whom have accepted the risks of invasive brain surgery and have failed to benefit from it. We believe that these two factors form the foundation of the relationship between the TR&Ds and this DBP. In order to make the most of this relationship, new technologies are needed to provide:

  1. tools that are easy to use. In the best case, the TR&D will develop interfaces that greatly facilitate medical image management and registration.
  2. Tools with an intuitive interface. Mobile apps are particularly attractive in this regard, especially those that can be easily integrated into a clinical workflow.
  3. Tools that facilitate the dissemination of knowledge among centers.
  4. Quantification of uncertainty.

These technical developments will allow investigators to make much better use of data collected during standard care, and will further allow them to predict how to best treat each DBS patient based on information gathered from large populations of prior patients. The major technical innovation is in the integration of patient imaging, DBS settings, clinical outcomes, and novel statistical models that will enable quantification of the effects of DBS in such a way that will enable the prediction of outcomes, effect size, and uncertainty for future patients. These capabilities are relevant for many areas of neuromodulation beyond DBS and we will argue that these technologies are critical because, in contrast to pharmaceuticals, there currently is no well-defined concept of a dose in DBS or any other neuromodulation therapy.

This DBP relies on several enabling technologies provided by the CIBC. The Simulation & Estimation TR&D provides tools and expertise in bioelectric field modeling and neuron population modeling and expertise in integrating clinical outcome data into patient-specific computational models. The Visualization TR&D provides tools and expertise in interactive models on a variety of platforms including desktop applications and mobile platforms, and visualization of effect size and uncertainty. The ultimate goal is to translate the anticipated results into improving patient outcomes and identifying new therapeutic technologies.
Collaborating Investigator(s): Gabrielle Kardon, PhD
Institution: University of Utah

Gabrielle Kardon is an Associate Professor of Human Genetics at the University of Utah. The Kardon Lab is focused on the development and regeneration of muscle and the role of the surrounding muscle connective tissue and adjacent tendons in regulating these processes. How muscle, connective tissue, and tendon are assembled into a functional limb musculoskeletal system is largely unknown and defective in multiple human genetic syndromes. Using sophisticated mouse genetic reagents to manipulate these three tissues, the Kardon Lab is dissecting the molecular mechanisms and cell-cell interactions required for normal limb development and that is aberrant in human genetic syndromes.

Understanding normal limb and diaphragm development and determining the genetic and cellular mechanisms underlying limb and diaphragm birth defects has great impact. This increased understanding will pave the way for future therapeutic treatments of these defects. Diaphragmatic birth defects (Congenital Diaphragmatic Hernias) are both common (1:3000 births) and 50% fatal. The estimated cost of treatment of CDH patients in the US is $250 million/year, and thus there is a strong incentive to devise early interventions. In adults, successful regeneration of muscle damaged during injury and disease is critical for proper function of the musculoskeletal system. The 4D (3D + time) data sets generated by this Driving Biomedical Project of diaphragm development limb muscle development and regeneration will be broadly useful to the research community.

This Driving Biomedical Project interacts extensively with Visualization and Image and Geometric Analysis TR&Ds to address several major technical challenges in the analysis of 4D confocal and multiphoton data sets. Of primary concern is the size of the data sets. The Kardon Lab continues to develop new genetic tools to fluorescently label tissues and as their ability to image more tissues simultaneously and to image them while developing or regenerating ex vivo grows, the size of the data sets increases exponentially. Current confocal data sets are generally in the range of 5 GB, but 4D datasets based on the new multiphoton data sets are 1–5 TB. The Image and Geometric Analysis and Visualization TR&Ds are addressing the required scalability of software tools through their development of level-of-detail and streaming methods for such large datasets. Another serious need addressed by the TR&Ds is the lack of quantitative tools to analyze such large image datasets. For instance, quantification of differences in the number and proliferative status of cells or size and shape of tissues between mutant and control mice is critical for understanding how muscle develops and regenerates. The Image and Geometric Analysis and Visualization TR&Ds develop interactive methods for localization, segmentation, and quantification of collective cell migration.