Diagnosis of multiple sclerosis using brain stem auditory evoked potentials

Abstract:

The main obstacle in the application of neural networks to the classification of biological signals used for diagnosing diseases is the small number of data available in the average clinical study. Indeed, the number of parameters (weights) that can be used in a model should not exceed 15% of the available data, in order for the model to generalize. This puts a limit on the dimension of the input nodes. Different authors have used different approaches to solve this problem, like eliminating redundancy in the data, preprocessing the data to find centers for the radial basis functions, or extracting a small number of features that were used as inputs. It is clear that the more features we could feed into the net, the better would be the classification. In this paper the approach of incrementing the number of training elements, using randomly expanded training sets, is utilized. This way the number of original signals does not constraint the dimension of the input set in the radial basis network. A comparison with results obtained using other methods will show that with this approach the rate of success was in general greater for healthy people.

Año de publicación:

2009

Keywords:

• Neural networks
• Signal processing
• Health Sciences
• Radial Basis Functions
• Multiple Sclerosis
• wavelets

Fuente:

scopus