Elena Valderrama

Polyimide cuff electrodes for peripheral nerve stimulation.

This paper describes a new tripolar spiral cuff electrode, composed of a thin (10 microm) and flexible polyimide insulating carrier and three circumneural platinum electrodes, suitable for stimulation of peripheral nerves. The cuffs were implanted around the sciatic nerve of two groups of ten rats each, one in which the polyimide ribbon was attached to a plastic connector to characterize the in vivo stimulating properties of the electrode, and one without a connector for testing possible mechanical nerve damage by means of functional and histological methods. The polyimide cuff electrodes induced only a very mild foreign body reaction and did not change the nerve shape over a 2-6 month implantation period. There were no changes in the motor and sensory nerve conduction tests, nociceptive responses and walking track pattern over follow-up, and no morphological evidence of axonal loss or demyelination, except in one case with partial demyelination of some large fibers after 6 months. By delivering single electrical pulses through the cuff electrodes graded recruitment curves of alpha-motor nerve fibers were obtained. Recruitment of all motor units was achieved with a mean charge density lower than 4 microC/cm(2) for a pulse width of 50 micros at the time of implantation as well as 45 days thereafter. These data indicate that the polyimide cuff electrode is a stable stimulating device, with physical properties and dimensions that avoid nerve compression or activity-induced axonal damage.

Peripheral nerve regeneration through microelectrode arrays based on silicon technology.

This paper describes some developments, made to obtain a chronic neural interface to record signals from regenerated peripheral nerves. Microperforated silicon dices, fabricated by techniques compatible with CMOS processes, were coupled in silicone nerve chambers and implanted between the severed ends of peripheral nerves in rats. Three configurations of perforated dices with 25 via-holes of 100 μm diameter, 121 via-holes of 40 μm and 400 via-holes of 10 μm were assessed. The feasibility of axonal regeneration through the dices via-holes was proved by histological and physiological methods over 3 months post-implantation. The regenerated nerves were organized in fascicles corresponding to the grid pattern of the via-holes. However, nerve regeneration was difficult and distal re-innervation delayed with respect to simple tubulization repair. The size of the via-holes and the total open area are determinants of the degree and quality of regeneration. Further improvements are needed in both the microelectrode dice design and in neurobiological stimulation of regeneration.

Infraclinic Breast Carcinoma: Application of Neural Networks Techniques for the Indication of Radioguided Biopsias

The aim of this work was to determine if the utilization of an artificial neural network (ANN) for the indication of radioguided biopsias can reduce the percentage of negative biopsias. The ANN was constructed as a three-layer, feed-forward network. The input layer consist of 15 input nodes corresponding to the radiologic and clinico-epidemiological data to evaluate. The ANN was trained using a supervised learning algorithm on 190 cases (122 benign, 68 malignant cases) and tested on 47 cases (30 benign, 17 malignant cases). Performance of the network was evaluated in terms of sensitivity and specificity over a range of decision threshold and was expressed as a receiver operating characteristic curve (ROC).

Noninvasive measurement of torque development in the rat foot: measurement setup and results from stimulation of the sciatic nerve with polyimide-based cuff electrodes

In neural rehabilitation, selective activation of muscles after electrical stimulation is mandatory for control of paralyzed limbs. For an evaluation of electrode selectivity, a setup to noninvasively measure the force development after electrical stimulation in the rat foot was developed. The setup was designed in accordance to the anatomical features of the rat model to test the isometric torque development at given ankle positions in an intact leg. In this paper, the setup design and development is presented and discussed. In a first study, the selectivity of small nerve cuffs with 12 electrodes implanted around the rat sciatic nerve was investigated. Special attention was drawn to the performance of the torque measurement setup in comparison to electrophysiological data obtained from compound muscle action potential recordings. Using one cuff around the nerve, electrical stimulation on different electrode tripoles led to plantarflexion and dorsiflexion of the foot without an a priori alignment of the cuff.

Design of a low-cost and high-speed neurocomputer system

This paper presents a new parallel computer architecture for high-speed emulation of any neural network model. The system is based on a new ASIC (Application Specific Integrated Circuit) that performs all required arithmetical operations. The essential feature of this ASIC is its ability to adapt the internal parallelism dynamically to the data precision for achieving an optimal utilization of the available hardware resources. Four ASICs are installed on one board of the neurocomputer system and emulate in parallel a neural network in a synchronous operation mode (SIMD architecture). By additional boards the system performance and also the size of the neural networks that can be simulated is increased. The main advantage of the system architecture is the simplicity of the design allowing the construction of low cost neurocomputer systems with a high performance. The achieved performance depends on the data precision, and the number of installed boards. In the case of 16 bit weights and only one board a performance of 480 MCPs and 120 MCUPs (using backpropagation) can be obtained.

DISCRETE LEARNING IN FEED-FORWARD NEURAL NETWORKS

We introduce a learning algorithm for feed-forward neural networks with synapses which can only take a discrete number of values. Taking into account the inherent limitations associated to these networks, we think that the performance of the method is quite efficient as we have shown through some simple results. The main novelty with respect to other discrete learning techniques is a different strategy in the search for solutions. Generalizations to any arbitrary distribution of discrete weights are straightforward.

Adopting New Competences Experience in the Face of Engineering Learning

The so-called Bologna process has opened a stage of deep changes in the university teaching methods in most European countries. Our work, which is focused on Spanish university reality, describes the changes that an engineering student must face in his/her learning habits as well as the new education methodologies followed by teaching staff to motivate and promote these changes. Some of these methodologies are oriented to create/reinforce a learning culture in the student and favor an intensive use of e-learning. Both elements play an important role as basis of the lifelong learning

A Massively Parallel Neurocomputer with a Reconfigurable Arithmetical Unit

This paper presents a massively parallel neurocomputer system which is mainly based on a new reconfigurable arithmetical unit optimized for the simulation of neural networks. The system offers a very high performance for all typical neural network operations combined with a high flexibility to adapt the available hardware resources to the requirements of a user-selected neural network model. The main system features are the support of many different bitlengths, a high memory bandwidth, a good scalability and a dynamic reconfigurability.

Learning Algorithm for Feed-Forward Neural Networks with Discrete Synapses

We introduce a learning algorithm for feed-forward neural networks with synapses which only can take a discrete number of values. The main novelty respect to other discrete learning techniques is a different strategy in the search for solutions which turns out to be quite effective. Generalizations to any arbitrary distribution of discrete weights are straightforward.

Synthesis of a Processor

In this chapter the design of a complete digital system is presented. The system to be synthesized is a generic (general purpose) component, able to execute different algorithms. The particular algorithm is defined by the contents of a memory block that stores a (so-called) program. This type of system is generally called a processor, in this case a very simple one. It is an example of application of the synthesis methods described all along the previous chapters.