Brain Machine Interfaces are an emerging technology whose purpose is to allow amputees and spinal cord injury patients to control a prosthetic limb using signals derived from the brain. The proposed work will create the means for investigating how the natural plasticity of the human brain can be exploited to innovate more efficient (and thus more easily made portable) Brain Machine Interface instrumentation. This will be achieved through the development of a new simulator that simultaneously models neural adaptation in reaching tasks, a prosthetic limb, and Brain Machine Interface hardware that connects the two. A key element of this simulator will be the ability to use real-time visual and proprioceptive feedback from the modeled arm to train the virtual brain cells and thus, over time, improve the accuracy with which the brain can control the prosthesis.
The proposed work will be accomplished using three Research Aims (1) Design and implement a simulation platform that models adaptive motor control of a human limb in three-space. (2) Design and implement an instrumentation testbed capable of realizing entire families of Brain Machine Interface data acquisition subsystems and systematically manipulating their parameters. The system will handle up to 50 x 50 channels and will collect performance statistics that quantify how and where information is lost or altered in the data pathway. (3) Quantify how BMI performance can be expected to degrade in response to errors in spike detection, spike sorting, and wireless neural data transmission.
The project will guide the development of next generation Brain Machine Interface systems, especially the implantable and wireless systems that remain an obstacle to Brain Machine Interfaces becoming realistic therapeutic devices. The project will provide training to students to conduct neural engineering research and also for developing new hands-on instructional materials for teaching neural engineering at the graduate level. The research will advance techniques for modeling functional ensembles of cortical neurons while also creating pedagogical tools for conducting neural engineering education and research at Temple University and beyond.
The outcome of this project has been advances in the way in which information from neurons in the brain are acquired and analyzed. Neurons form the basic building block of brains and neural tissue, and understanding how they store, process, and transmit information is a fundamental challenge for the scientific community. One outcome of this project has been the development of electronic systems that can simultaneously acquire information in real time from an order of magnitude more neurons than was previously possible. These systems are based on field programmable gate arrays (FPGAs) which are commercially available computing systems that can be custom configured to optimally implement a specific computational need. Part of the intellectual merit of this project is that we have developed a series of system architectures that allow for enough throughput and latency to record from up to 3000 neurons using contemporary hardware. The broader impact is that the architectures we have developed will scale well with increasingly powerful FPGAs, allowing for even more channels in the future. A second main outcome of this work is the development of numerical and experimental techniques for studying neuronal plasticity in functional cell clusters. We have cultured neurons in special multi-electrode array dishes that allow us to record electrical activity from the network of cells as well as provide them with electrical stimulation to affect their connectivity. We have introduced new sensitive statistical techniques for measuring changes in network behavior over time that are capable of revealing organizational network trends that were previously not observable. This means that scientists who use cell cultures as a means of studying how neural tissue stores and processes information will have a valuable new tool for assessing function. We have also used these techniques to examine whether various commonly used electrical stimuli are capable of inducing fine-tuned localized network changes (as has been widely assumed in the research community) but have discovered that they are only capable of inducing relatively blunt network-wide activity. The broader impact of this outcome is that we have refined some commonly used techniques for studying neural function which will hopefully lead others to more conclusive findings. Finally, this project has succeeded in training a large cadre of graduate, undergraduate and high school students in electrical, biomedical, and neural engineering. Specifically, this work has supported the training of three doctoral students, two masters students, approximately 15 undergraduate students, and ten high school students.
