2019/01/18 by Oleksandr Bailo, Bailo, Oleksandr, DongShik Ham +3 · 2 citations
Biochemistry, Genetics and Molecular Biology · Computer Science · #Cell Image Analysis Techniques #Computer Vision and Pattern Recognition (cs.CV) #Digital Imaging for Blood Diseases #FOS: Computer and information sciences #Generative Adversarial Networks and Image Synthesis
paper · pdf · doi:10.48550/arxiv.1901.06219
openalex publication_date 2019/01/18 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
In this paper, we describe how to apply image-to-image translation techniques\nto medical blood smear data to generate new data samples and meaningfully\nincrease small datasets. Specifically, given the segmentation mask of the\nmicroscopy image, we are able to generate photorealistic images of blood cells\nwhich are further used alongside real data during the network training for\nsegmentation and object detection tasks. This image data generation approach is\nbased on conditional generative adversarial networks which have proven\ncapabilities to high-quality image synthesis. In addition to synthesizing blood\nimages, we synthesize segmentation mask as well which leads to a diverse\nvariety of generated samples. The effectiveness of the technique is thoroughly\nanalyzed and quantified through a number of experiments on a manually collected\nand annotated dataset of blood smear taken under a microscope.\n