Neural networks: In search of computer-aided diagnosis

Abstract

The authors introduce the Bonsett Protocol, which quantifies a physician's intuitive and experienced opinions. Neural networks and EMG (electromyographic) signals have been used to enhance the diagnosis of neuromuscular disorders. Neural networks in both supervised and unsupervised learning environments have provided valuable insight

however, there has been no method available to quantify the expert opinion of the examining physician. This study combines the quantitative EMG features that are derived from the motor unit action potential, and the expert physician's opinion, thus forming the input to the neural network. Expert physicians will be able to create their own neural network models by quantifying their examination protocols and combining it with the quantitative EMG data.