Deep Learning-Based Algorithm for Disease Prediction from Blood Microscopic Images

Abstract

The current method of diagnosing blood disorders relies on the expertise of a haematologist, which is time-consuming, prone to errors, and has limitations. To aid clinical decision-making, an automated visual image analysis system is necessary, particularly in cases of leukemia, a form of cancer that affects white blood cells. Early and precise diagnosis is essential for effective treatment and patient survival. Although traditional methods involve microscopic examination of blood smears, machine learning techniques have also been explored but with significant misclassification errors. To enhance accuracy, a deep learning system can be employed, which comprises two modules: detection and classification. The function of the detection module is to distinguish white blood cells from other elements present in the images, whereas the classification module applies a back propagation network technique to recognize and quantify the relevant white blood cells that aid in the diagnosis of leukemia.