Automatic Interpretation of MRI Brain Images Using Probabilistic Neural Network

Automatic interpretation of various brain abnormalities is possible if classification of magnetic resonance (MR) human brain images can be carried out in an efficient manner. This paper proposes a new approach for automatic diagnosis using wavelet transform in combination with probabilistic neural network. The proposed method consists of two stages namely (1) feature extraction and (2) classification. In the first stage, features are extracted using discrete wavelet transformation because wavelet transform based methods are well known tool for extracting frequency space information from non-stationary signals. Subsequently in the classification stage a classifier based on probabilistic neural network has been developed. Investigations have shown that probabilistic neural network provides a general solution to the pattern classification problems and its classification accuracy is more than the commonly used backpropagation network. The pathological images considered were those of brains suffering from stroke, degenerative disease and infectious disease. The proposed approach classifies the pathological images for an accuracy ranging from 93% to 100%.