[4C1-GS-7] Vision, speech media processing

[4C1-GS-7-05] Retinal Vessel Segmentation Using Data Augmentation Based on U-Net

〇ZHANG QIUYANG¹, YASUMURA YOSHIAKI¹ (1. Shibaura Institute of Technology )

[[Online]]

Keywords:retinal vessel segmentation, Data Augmentation, U-Net

Retinal vessels are capillaries that can be directly observed without specialty medical equipment. Since the lesions here reflect various diseases such as diabetes and stroke, the extraction of blood vessels from retinal fundus images is of great significance for the diagnosis and treatment of various diseases. Despite this high demand, machine learning-based methods are difficult to apply. This is because data in the medical field is difficult to obtain and few true data are available. To solve the problem of insufficient training data for machine learning, this report proposes a vessel segmentation method that can be trained with a small amount of training data by using a data augmentation.