MERCIS LAB

Traditional coronary artery bypass graft (CABG) surgery has undesirable side effects that range from cognitive loss to increased hospital stay that are believed to be related to artificial heart pumps.

This project aims to develop intelligent robotic tools for performing off-pump CABG surgery. The project's main focus are development of intelligent control algorithms, designing of milli- and macro- scale intelligent robotic instruments for CABG surgery, and development of sensing system for tracking heart motion.

For more information please visit project page.

We are developing a millirobotic gripper with integrated actuation for minimal invasive surgery. The diameter and length of the tool is restricted in order to give maneuverability to the millirobot and to crate less damage to the tissue during the incision. The design is aimed to have a high gripping force and small dimensions than existing designs.

Video: Hybrid Gripper (16.8 MB, Video Compression: DivX®)

Rehabilitation Robotics

EEG Signal Analysis for Stroke Rehabilitation:

Electroencephelography (EEG) and related technologies have long been helpful tools for physicians in their efforts to rehabilitate victims of stroke. However, because EEG signals are the aggregate of all the brain's complex electricortical activities, there exists a great challenge in deciphering EEG signals. One such challenge is to see if the EEG signals of stroke victims can be systematically differentiated from the EEG data of healthy individuals. Since stroke disables parts of a human's brain, we would expect such damage to be reflected in EEG recordings and somehow discernable. Some of the approaches to this problem include spectral analysis, electromyograph (EMG) noise rejection, Bayesian analysis, and event-related time-series analysis.

GiPSi^TM is an open source/open architecture framework for developing organ level surgical simulations. Main goal is to facilitate shared development of reusable models, to accommodate heterogeneous models of computation, and to provide a framework for interfacing multiple heterogeneous models. The framework provides an intuitive API for interfacing dynamic models defined over spatial domains. It is specifically designed to be independent of the specifics of the modeling methods used and therefore facilitates seamless integration of heterogeneous models and processes. I/O interfaces for visualization and haptics for real-time interactive applications have also been provided.

GiPSi^TM API version 1.0 is currently available in the GiPSi^TM website. The downloadable GiPSi^TM code includes the specifications of the Core GiPSi^TM API, the GiPSi^TM Computational Tools set, and a number of very simple sample model implementations and a basic visualization engine which follow the GiPSi^TM API specifications.

For more information please visit GiPSi ^TM (General Physical Simulation Interface) website. 

By expanding and adding a network extension to GiPSi^TM provides benefit to user to use Surgical Simulation over the network. All users can access and perform the simulation any time from any appropriate network access point. Network extension of GiPSi^TM involves a middleware module (GiPSiNet) to improve the lack of network QoS and to enhance the user-perceived quality of a networked simulation.

The long-term objective of this research is to develop and validate a compelling and effective virtual environment-based simulator to provide training in the field of endoscopic neurosurgery, with the following specific aims:
- Development of the enabling technologies for construction of a virtual environment-based training simulator for endoscopic neurosurgery.
- Application of the virtual environment for skill and procedure training in endoscopic neurosurgery.
- Establishment of the construct validity of the surgical training simulator.

Our research focuses on segmentation of MRI data with the goal of producing mesh models of anatomical structures with minimal user interaction. The level set and fast marching methods are the primary algorithms being investigated. By incorporating into these methods ontological information about the regions being segmented and competitive algorithms for region expansion, we hope to achieve high quality image segmentation with little or no manual input. The segmented data will then be used to create detailed models for surgical simulation.

An important area of research in the teleoperation literature is to develop systematic methods to quantitatively compare different manipulator designs in application critical tasks. Since teleoperation systems are mostly executed in the extreme environment, there are constraints in designing the mechanism and choosing sensors. Such quantitative methods are especially important during design of the manipulators to make an informed decision among various design alternatives.

We have developed a novel approach to quantitatively compare different sensory schemes. This evaluation is done by comparing the norm of the a posteriori error covariance matrices of the Kalman filters for each configuration. The main advantage of this method is that it allows to quantitatively compare arbitrary sensory configurations.

We have also developed a quantitative comparison method for the overall teleoperation system designs. This method is based on H_∞ framework. The upper H_∞ norm bound of the system including H_∞ sub optimal controller is used as the performance index. As a case study, the method is applied to a real teleoperation system to study the effects of sensory configuration and back-drivability of the mechanism on the performance of the system in tasks which involve different environment impedances.

The value of haptic interaction in surgical simulation applications has led to a great deal of research interest into the challenges involved in providing haptic force-feedback in virtual environment simulations with deformable surfaces. The key obstacle to overcome in haptic interaction is the difference in update rates of the simulation, which typically is linked to the graphical update rate of between 10 and 60 Hz, and the update rate requirement of the haptic interface, which must be on the order of 1 KHz in order to be convincing to the operator. Techniques to bridge this gap are explored in this project through multi-rate simulation, where the virtual environment in its full complexity is simulated at the visual update rate, while a simpler simulation encompassing parts of the environment local to the virtual instrument is run in parallel at the haptic update rate and periodically re-synced to the full model. Specifically, multi-rate simulation techniques are being adapted for use with mass-spring and FEM models to simulate tissue properties for surgical simulation applications.

Emergence of "Systems Biology" provided a comprehensive and integrative perspective to examine the structure and function at the cellular and organism level instead of focusing on the isolated parts. Integration of multilevel and multiscale physiological models is an important requirement for such a system-based approach. Mathematical models for physiological processes have been developed in all levels from cell, up to organs and organ systems. However, little has been done in the name of integrating individual models to comprehensively study the whole system. This is due to the complexity to integrate multiscale and multilevel models of independent physiological processes.

This project aims to build a software framework where the multilevel and multiscale models for the complex biological system can be integrated.

For more information please visit project page.

The characteristics of excitable cell mathematical models can be used for the development of new techniques in simulating the electrical behavior of the human heart. While very simple models of such behavior can be simulated at real-time or better speeds on powerful computing equipment, the use of realistic cell models or organmagnitude cell networks make the simulations computationally infeasible. In this project, an examination of the FitzHugh-Nagumo model and its response to stimulus is utilized. In order to move toward the goal of a full cardiac simulation, a method of optimizing single-cell calculations through local interpolation techniques and a separate method of optimizing multi-cell simulations by tracking cellular activations are introduced.