Analysis of WBC structure from microscopic images and classification of cells into types is challenging. Although white blood cells can be differentiated based on their shape, color and size, one challenging aspect is that they are surrounded by other blood components such as red blood cells and platelets. In this study, transfer learning method using four network architectures that have been trained in advance is applied to classify the white blood cell images. The network architectures used are AlexNet, GoogleNet, ResNet-50 and VGG-16. A comparative analysis of the performance of these architectures was carried out in classifying the images. The evaluation method was undertaken using Confusion Matrix. The performance metrics measured in the evaluation are Accuracy, Precision, Recall and Fmeasure. The results showed that all architectures succeeded in classifying white blood cells using the transfer learning method. ResNet-50 is the network architecture that shows the highest performance in classifying white blood cell images.
Copyrights © 2021