An improved deep learning based classification of human white blood cell images
Md. Abu Ismail Siddique, Abu Zahid Bin Aziz, Abdul Matin
Abstract
White blood cells (WBC) are part of the immune systems which defend both infectious diseases and foreign invaders. There are various types of white blood cells in our body and each of these blood cells has a specific function in our body. The differential test is the traditional way to classify white blood cells in that it calculates the percentage of different types of white blood cells. In this test, the efficiency is low and time-consuming. Various machine learning and deep learning methods have been developed over the years that produced good results. In this work, we applied a deep learning based convolutional neural network (CNN) called “SqueezeNet” to classify white blood cells. After fine-tuning the hyperparameters, we trained our model and tested its performance in the testing dataset. Our method achieved 93.8% accuracy in the test data which is better than the existing classifiers. This proves that our method can be a useful approach for this task.