WBCaps: A Capsule Architecture-based Classification Model Designed for White Blood Cells Identification
Yan Liu, Ying Fu, Pu Chen · 2019
This pilot study presented a capsule architecture-based classification model named WBCaps to automatically recognize the five types of white blood cells (WBC) from normal peripheral blood smears. The entire WBCs identification workflow including two sections were also reported here. The first section automatically located the WBCs from the whole slice of blood smears through a three-step image segmentation method including color normalization, color deconvolution and cell extraction. In the second section, the extracted cells were fed into WBCaps, which was a network cascaded by traditional convolutional layer, primary capsule layer and representation capsule layer, to generate the type prediction. We employed 3-fold cross validation to validate the proposed WBCs identification method on a small clinical dataset and yielded precision 0.99, recall 0.99 and F1-score 0.99. The result outperformed that of ResNet (precision 0.97, recall 0.97, F1-score 0.97) and VGG (precision 0.97, recall 0.97, F1-score 0.98). The proposed model could be a promising clinical tool for hematology analyzer to facilitate the cytological and morphological examination.