Webmaster:
bioweb (at) langate.gsu.edu

Neurons & Networks, led by Ying Zhu (Computer Science) and Paul Katz (Biology), is involved with neuroinformatics and network behavior of both biological and artificial ensembles of neurons.

The Neurons & Network Research Group has a series of informal, brown bag, research talks.  Two or three researchers will give 10 minute talks about their research with a goal of inspiring collaborations.  All interested Faculty, Students, and Post-docs are encouraged to attend.  Bring your lunch and come hear about the exciting work being done in Neurons & Networks.
Upcoming group meetings | Past Talks

Principle Investigators: Gennady Cymbalyuk / Andrey Shilnikov
Title: The Bifurcation Analysis of Neuronal Rhythmogenesis

Abstract:
Neurons exhibit various activity regimes and regime transitions that reflect their complement of ionic channels and modulatory state. Importance of the bursting activity has been established for control of rhythmic movements, and it has been widely observed in sleep and pathological brain states.  Analysis of the dynamics supporting bursting activity and classification of various transitions into bursting activity are incomplete nowadays and remain a fundamental problem for both neuroscience and the theory of dynamical systems.

We will employ methodology uniting neuroscience and the advanced theory of dynamical systems and non-local bifurcations. The dynamical systems theory identifies universal patterns. Bifurcation theory elucidates how these patterns qualitatively change as a control parameter in the model is varied. Similar methodology will be implemented in experiments. The basic idea is to introduce an artificial current through the dynamic clamp into a living neuron. Systematic variations of a control parameter determining this current will allow us to demonstrate characteristic features of the dynamical phenomena in the experiment. In this work we will implement the technique of Pontryagin's averaging method of singularly perturbed systems into the electrophysiological experiment. We will use it as a powerful tool for effective detection and bifurcation analysis of periodic orbits (tonic spiking and sub-threshold oscillations) in living neurons.

Principle Investigators: Yanquing Zhang / David Washburn
Title: Granular cognitive neural networks based on artifcial intelligence, computation intelligence and cognitive science

Abstract:
Artificial neural networks have been used in many applications. However, there are still many challenging problems such as the black box problem of traditional supervised learning algorithms and the over-fitting problem. This proposal focuses on the traditional black box problem in terms of meaningful discovered knowledge and decision-making process. NSF has the Artificial Intelligence and Cognitive Science (AICS) program focusing on advancing the state of the art in Artificial Intelligence and Cognitive Science. The AICS program supports research and related education activities fundamental to the development of computer systems capable of performing a broad variety of intelligent tasks, and to the development of computational models of intelligent behavior across the spectrum of human intelligence. So there will be future chances to apply for relevant NSF grants. Here, we plan to use traditional artificial intelligence, new computational intelligence including fuzzy logic, neural networks, evolutionary computation, granular computing and cognitive science to investigate the new theoretical neural network model that can discover hidden decision-making knowledge from data (i.e. know how the human brain learns new knowledge and makes a decision. The traditional neural learning algorithms can generate a lot of numerical weights that are not meaningful (i.e., the black box problem). Now we plan to design basic neuron groups that can contain meaningful knowledge like rules, and make new granular learning methods based on cognitive science and computational intelligence. A graduate student in computer science and a graduate student in  psychology will work on this joint project together.

Principle Investigators: Dr. Vincent Rehder / Nikolaus Dietz / Unil Perera
Title: Assessment of the effect of terahertz radiation on brain development and brain function

Abstract:
Technological developments and their introduction into the marketplace often precede studies on the adverse effects that these technical developments may have on human health. In this proposal, we attempt to reverse this trend by studying the effects of terahertz radiation on the development and function of the nervous system. Studies on teraherz emitters and detectors are presently cutting edge in physics research and this technology holds great promise to revolutionize various commercial applications in the future. Homeland security (chemical, biological, and explosive detection, seeing through walls and luggage etc.) will be a key area where THz technology will be used. Other commercial areas include, satellite imaging, skin cancer detection and secure communications. In addition, molecular spectroscopy studies of earth and planets will drastically increase as the devices are available.  As such, this technology and the radiation the devices using this technology emit or receive will have an increasing impact on the human body. To date, we are not aware of systematic studies investigating the potential effects of this radiation on the human brain. This proposal brings together three researchers from the Biology and Physics departments whose collective expertise is exquisitely suited to address this question. Dr. Rehder is an expert in Developmental Neuroscience and he has extensive expertise in culturing nerve cells, in optical imaging technology, and in the analysis of nerve growth and synapse formation. Dr. Dietz is an expert in characterization of radiation – matter interactions and the growth of thin film heterostructures. His characterization lab has several radiation sources, spectrometers, and detectors for various frequency regimes. Dr. Dietz growth facilities will be used to grow advanced terahertz emitter and detector structures that will be explored in this study.  Dr. Perera is developing semiconductor  terahertz devices  and has applied for a patent on  a THz detector. At present, Dr. Perera's group has developed detectors capable of detecting THz radiation up to 2 THz and is working on extending the range to 1 THz and beyond.  His Optoelectronics lab employs several characterization  techniques including FTIR spectrometers, monochromators covering the range from near IR to THz frequencies.  He is also the director of interaction of radiation with matter laboratory.

Principle Investigators: Dr. Paul Katz / Rajshekhar Sunderraman / Ying Zhu
Title: Identified Neuron Database Project

Abstract:
Invertebrates have uniquely identifiable neurons.  This allows neuronal circuitry to be understood at the level of cellular elements.  Thus invertebrate nervous systems serve as important models for the study of sensory motor integration, central pattern generation, and learning and memory. This database proposal will provide a means of keeping track of identified neurons and connections. The project will begin by examining identified neurons in one molluscan species, Tritonia diomedea , but the same format can be applied to any invertebrate where individual neurons can be identified. The goal is to use the database to catalog known identified neurons and their connections and then create a repository for observation of unidentified neurons that could be accumulated until those neurons can be individually identified.

Specific Aims:

1. Create a database structure that will allow us to store and retrieve information about individual neurons in Tritonia .  The structure of the database is being designed collaboratively through weekly meeting with the team. 
2. Create a graphical representation of the central nervous system of Tritonia that will allow the computer to assign unique coordinates to each neuron in the brain.
3. Create a user interface to enter data into the database.  The interface will be web-based.  It needs to be easy to use and fast so that it will be adopted by researchers in many labs.
4. Enter data into the database.  Initially this will be data on neurons that have already been identified. We call these “canonical descriptions” because they don't represent individual observations, but rather an idealized or average version of the neuron. Part of this aim is to test the database and make corrections.
5. Enter individual observations of neurons that have not yet been individually identified. Create an algorithm for recognizing and clustering observations that may represent the same neuron.  These observations can then be coalesced into a canonical description of an identified neuron.

Principle Investigators: Saeid O. Belkasim / Yi Pan / Dr. Donald Edwards
Title: Processing and analysis of confocal microscopic images of neurons

Abstract:
Confocal microscopy is a widely used imaging technique in many biological and biomedical fields. Images of Confocal microscopy require very laborious procedure to view, store and analyze these images. Automating the procedure using image processing tools may result in reduction in the amount of data to be stored, and can provide fast manipulation of the reconstructed 3-D-image. One of the basic operations in the 3-D image reconstruction process is segmentation which refers to the process of extracting the desired object (or objects) of interest from the background in an image or data volume. The segmentation process comprises a variety of techniques such as thresholding, edge detection, border following and clustering.

Most of the segmentation algorithms implemented in current digital image processing software were developed more than a decade ago. Since that time a tremendous improvement in machine processing power and new fundamental theories such as the multi resolution image analysis have been emerged. The main focus of this research proposal is to establish a selective digital image processing environment that effectively uses the multi resolution analysis at the partial segment as well as the whole image levels. This environment is capable of handling very detailed high resolution level or mixture of several resolution levels. The proposal simplifies the segmentation problem using hierarchical multi resolution image structuring procedure to select mixture of optimal resolutions to represent images by smaller segments. The processing speed and efficiency of these algorithms can be increased to achieve real time processing by harnessing the parallel image processing capabilities of the newly introduced powerful graphical processing units.