Scheduled System Maintenance:
Some services will be unavailable Sunday, March 29th through Monday, March 30th. We apologize for the inconvenience.
By Topic

Wavelet neural networks for clinical diagnosis

Sign In

Cookies must be enabled to login.After enabling cookies , please use refresh or reload or ctrl+f5 on the browser for the login options.

Formats Non-Member Member
$31 $13
Learn how you can qualify for the best price for this item!
Become an IEEE Member or Subscribe to
IEEE Xplore for exclusive pricing!
close button

puzzle piece

IEEE membership options for an individual and IEEE Xplore subscriptions for an organization offer the most affordable access to essential journal articles, conference papers, standards, eBooks, and eLearning courses.

Learn more about:

IEEE membership

IEEE Xplore subscriptions

1 Author(s)
Dickhaus, H. ; Dept. of Med. Informatics, Heidelberg Univ., Germany

Summary form only given. The wavelet neural network (WNN) approach can be regarded as a hybrid method for pattern-recognition tasks of non stationary biosignals, e.g. electrocardiograms, evoked potentials, or electromyograms occurring together with clinical data of the patient in medical diagnosis. The concept of WNN for classification aims to combine aspects of the wavelet transformation for purposes of feature extraction and selection with the characteristic decision capabilities of neural network approaches. A small number of features are calculated from high-dimensional input patterns - in most cases sample points of a signal - by time frequency transformation. A. certain number of easily interpretable parameters of the transformation controls this feature extraction process. The calculated features are regarded as inputs to an artificial neural network. During the training phase, the WNN not only learns adequate decision functions and arbitrarily complex decision regions defined by the weight coefficients, but also looks for those parts of the parameter space that are suited for a reliable categorization of the input signals. This procedure allows an optimal automatic adjustment of the input parameters in dependence on the correct classification. The approach can be expanded to multidimensional wavelet nodes. This can be useful for applications where more than one input signal should be. considered. Furthermore, features that cannot be extracted from recorded signals, e.g. data of the subject's age, sex, or medication etc., can be used as additional inputs to the net's classification part, that expands this approach to a hybrid system. Excellent results are achieved by this method for many clinical applications, for example, identification of high-risk patients in cardiology or discrimination of children suffering from attention deficit hyperactivity disorder in neurology.

Published in:

Engineering in Medicine and Biology Society, 2001. Proceedings of the 23rd Annual International Conference of the IEEE  (Volume:4 )

Date of Conference: