CLASSIFICATION OF BLOOD LEUKEMIA USING NEURAL NETWORKS AND IMAGE PRE-PROCESSING

In this paper, we have presented an efficient mechanism for the classification of blood leukemia using microscopic images employing a Probabilistic Neural Network (PNN). Probabilistic Neural Network is based on Bayes's theorem of Conditional Probability and is a famed paradigm for data classification for systems employing artificial intelligence. Pre-processing has been achieved using gray scaling and thresholding. Discrete Wavelet Transform (DWT) has been used as a tool to remove the abrupt variations in the calculated feature values. Principal component analysis (PCA) has been used to find particular trends in the computed feature data and minimize the redundancy. We have shown that the proposed technique achieves 98.33% percent classification accuracy which can be attributed to the highly rigorous pre-processing and feature extraction mechanisms which culminates to training a Probabilistic Neural Network which is used for the final classification of the data.