The Effectiveness of Data Augmentation for Mature White Blood Cell Image Classification in Deep Learning — Selection of an Optimal Technique for Hematological Morphology Recognition —

The data augmentation method is known as a helpful technique to generate dataset with large number of images from one small for supervised training in deep learning. However, low validity image recognition was reported recent study on artificial intelligence (AI). This aimed clarify the optimal learning model generation white blood cells (WBCs). Study Design: We conducted three different methods (rotation, scaling, and distortion) original WBC images, each AI generated by training. subjects clinical assessment were 51 healthy persons. Thin-layer smears prepared peripheral subjected May-Grünwald-Giemsa staining. Results: only significantly effective among models rotation. By contrast, effectiveness both distortion scaling poor, improved accuracy limited specific subcategory. Conclusion: Although are often used achieving high training, we consider that it necessary select medical based characteristics images.