A Comparison of Human Crafted Features and Machine Crafted Features on White Blood Cells Classification

Leukocytes can be divided into five types and each type can be classified according to the characteristics of their morphological structure, cytoplasm, and nucleus. It is essential to classify leukocytes precisely because each type of leukocytes has a distinctive characteristic to provide defense against anomaly in our body. Unfortunately, doctors usually use manual verification to classify leukocytes one by one. This method is time-consuming and susceptible to error. To solve this problem, we used machine-crafted features method and human-crafted features method. In this study, we compare both methods and verify the results by comparing their quantitative measurement and computational time. The machine-crafted features method gave better result compared human-crafted features method, but the drawback machine crafted feature method is computational challenging compared to human-crafted feature method.