Neuroimitator Overview
The Neuroimitator (NI) program is a new tool which allows the formation of arbitrarily structured neural nets. Parameters are assigned to the neurons and synaptic connections specified that are similar to observed in electrophysiological experiments. Arbitrary input influences can be assigned to any elements of the network or the whole network, and the network functioning as a whole simulated. The NI provides a means for "materialization" of physiological hypotheses.
Electrophysiological parameters of the neurons in the NI are phenomenological and are not connected with molecular mechanisms underlying the parameters in neuron (metabolic transport, ionic conductances, etc.).
The user interface of the Neuroimitator consists of the specialized graphical editors: neural network editor, function editor, input influences editor, project manager and two viewers.
The neural network editor allows construction of the network (quantity of neurons, their location and interconnections, parameters of neurons and connections, etc.). It has the working space 100x100 places for the neurons. The parameters of the neuron which can be defined: - resting potential - threshold - absolute refracterity - relative refracterity (function, usually exponential) - form of action potential (function, usually complex, consisting from a spike function and a trace potential which may be positive, negative or biphasic function).
The synaptic connection parameters are: - delay - weight (amplitude, positive for EPSP, negative for IPSP) - relaxation time of a function of the rear front of postsynaptic potential.
The function editor is used to define and edit the functions which are used in other editors and programs of the Neuroimitator in the cases where the parameters are not constant. The editor library contains several functions: exponential, linear, rectangular, saw-like and sinusoidal. It is possible to add to the library with user defined functions and control their scaling and calibration.
The editor of input influences is assigned to condition the functioning of the neural net during influence of external stimuli, i.e. for creating the model of external input. Two types of input influences may be assigned: an event which always invokes action potential in the neuron, and a change in the membrane potential defined by arbitrary function.
The User output interface contains two specialized visualizers: - visualizer of the membrane potential changes, threshold and other parameters of individual neurons This allows the demonstration on the monitor of practically the same picture which a physiologist observes during the intracellular experiment - the visualizer of spike spreading and synaptic activity in the net. This allows the demonstration on the monitor the whole neural net or selected regions with pseudocolors indicating the dynamics of spike generation in neurons, their spreading in the network and the generation of synaptic potentials.
The program has 100x100 grid of places for neurons. Amplitude can be defined with 0.1mV precision, time - 0.5 msec.
In the present version simulation of the network functioning occurs without plastic changes in synapses or neurons. Version 4 will have plasticity simulation (habituation, accommodation, Hebbian plasticity, etc.)
System Requirements
Neuroimitator 3.1 requires the following minimum configuration:
- An IBM Personal Computer.
- A VGA monitor or better.
- A 386 or higher processor (coprocessor is desirable).
- A hard disk with enough disk space to install Neuroimitator (about 1-2 megabytes) and to store result files (depends on the complexity of your networks and the time interval simulated).
- Microsoft Windows or Microsoft Windows for Workgroups version 3.1 or higher.
- At least 4M RAM. The more RAM you have on your computer the more complicate networks you can simulate.
Neuroimitator V3.2 for Windows 95 requires a 486 or higher processor and 3 megabytes of space.