An Efficient Deep Learning Hybrid Model for Improving the Classification Accuracy of Blood Cell Images

Abstract

Blood is the lifeline of human body and is also a key indicator to many underlying conditions or may predict the onset of serious conditions or disorders. Careful analysis with data inferred from the blood cell analysis is crucial for early detection and treatment of a number of life-threatening disorders. Amongst the various types of blood cells present in a blood cell image, WBCs or white blood cells give precise information about a number of such medical disorders. Analysis of WBC has been taken as the primary objective of this research work by invoking state-of-the-art image processing methods. A deep learning-based hybrid model has been proposed and implemented in this research work effective classification of WBCs from blood cell images obtained from benchmark dataset namely BCCD. The classification thus obtained has been compared with existing benchmark methods and in each case, the computed performance metrics indicate that proposed hybrid model outperforms existing methods.